diff --git "a/full_fine-tuning/Salesforce/codegen-350M-multi/trainer_state.json" "b/full_fine-tuning/Salesforce/codegen-350M-multi/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/full_fine-tuning/Salesforce/codegen-350M-multi/trainer_state.json"
@@ -0,0 +1,9492 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.995008319467554,
+  "eval_steps": 500,
+  "global_step": 1350,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0022185246810870773,
+      "grad_norm": 4.3018317222595215,
+      "learning_rate": 3.703703703703704e-07,
+      "loss": 0.8259,
+      "step": 1
+    },
+    {
+      "epoch": 0.004437049362174155,
+      "grad_norm": 6.3028764724731445,
+      "learning_rate": 7.407407407407408e-07,
+      "loss": 0.7536,
+      "step": 2
+    },
+    {
+      "epoch": 0.0066555740432612314,
+      "grad_norm": 5.622929573059082,
+      "learning_rate": 1.1111111111111112e-06,
+      "loss": 0.8631,
+      "step": 3
+    },
+    {
+      "epoch": 0.00887409872434831,
+      "grad_norm": 4.8698930740356445,
+      "learning_rate": 1.4814814814814817e-06,
+      "loss": 0.7016,
+      "step": 4
+    },
+    {
+      "epoch": 0.011092623405435386,
+      "grad_norm": 3.686124086380005,
+      "learning_rate": 1.8518518518518519e-06,
+      "loss": 0.7406,
+      "step": 5
+    },
+    {
+      "epoch": 0.013311148086522463,
+      "grad_norm": 5.313748359680176,
+      "learning_rate": 2.2222222222222225e-06,
+      "loss": 0.6944,
+      "step": 6
+    },
+    {
+      "epoch": 0.01552967276760954,
+      "grad_norm": 4.853567600250244,
+      "learning_rate": 2.5925925925925925e-06,
+      "loss": 0.716,
+      "step": 7
+    },
+    {
+      "epoch": 0.01774819744869662,
+      "grad_norm": 3.770028829574585,
+      "learning_rate": 2.9629629629629633e-06,
+      "loss": 0.8874,
+      "step": 8
+    },
+    {
+      "epoch": 0.019966722129783693,
+      "grad_norm": 8.172672271728516,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.7306,
+      "step": 9
+    },
+    {
+      "epoch": 0.022185246810870772,
+      "grad_norm": 2.7724807262420654,
+      "learning_rate": 3.7037037037037037e-06,
+      "loss": 0.8935,
+      "step": 10
+    },
+    {
+      "epoch": 0.024403771491957847,
+      "grad_norm": 8.790614128112793,
+      "learning_rate": 4.074074074074075e-06,
+      "loss": 1.0276,
+      "step": 11
+    },
+    {
+      "epoch": 0.026622296173044926,
+      "grad_norm": 3.552311897277832,
+      "learning_rate": 4.444444444444445e-06,
+      "loss": 0.6588,
+      "step": 12
+    },
+    {
+      "epoch": 0.028840820854132,
+      "grad_norm": 4.177034854888916,
+      "learning_rate": 4.814814814814815e-06,
+      "loss": 0.8304,
+      "step": 13
+    },
+    {
+      "epoch": 0.03105934553521908,
+      "grad_norm": 3.1599209308624268,
+      "learning_rate": 5.185185185185185e-06,
+      "loss": 1.0338,
+      "step": 14
+    },
+    {
+      "epoch": 0.033277870216306155,
+      "grad_norm": 5.0901641845703125,
+      "learning_rate": 5.555555555555556e-06,
+      "loss": 0.7944,
+      "step": 15
+    },
+    {
+      "epoch": 0.03549639489739324,
+      "grad_norm": 3.609119415283203,
+      "learning_rate": 5.925925925925927e-06,
+      "loss": 0.8199,
+      "step": 16
+    },
+    {
+      "epoch": 0.03771491957848031,
+      "grad_norm": 3.859593391418457,
+      "learning_rate": 6.296296296296296e-06,
+      "loss": 0.7405,
+      "step": 17
+    },
+    {
+      "epoch": 0.03993344425956739,
+      "grad_norm": 2.716921091079712,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.693,
+      "step": 18
+    },
+    {
+      "epoch": 0.04215196894065446,
+      "grad_norm": 2.667419195175171,
+      "learning_rate": 7.0370370370370375e-06,
+      "loss": 0.6739,
+      "step": 19
+    },
+    {
+      "epoch": 0.044370493621741544,
+      "grad_norm": 3.6542274951934814,
+      "learning_rate": 7.4074074074074075e-06,
+      "loss": 0.7344,
+      "step": 20
+    },
+    {
+      "epoch": 0.04658901830282862,
+      "grad_norm": 3.7599401473999023,
+      "learning_rate": 7.777777777777777e-06,
+      "loss": 0.8118,
+      "step": 21
+    },
+    {
+      "epoch": 0.048807542983915694,
+      "grad_norm": 2.6256203651428223,
+      "learning_rate": 8.14814814814815e-06,
+      "loss": 0.7521,
+      "step": 22
+    },
+    {
+      "epoch": 0.051026067665002776,
+      "grad_norm": 3.9963862895965576,
+      "learning_rate": 8.518518518518519e-06,
+      "loss": 0.9045,
+      "step": 23
+    },
+    {
+      "epoch": 0.05324459234608985,
+      "grad_norm": 2.820268392562866,
+      "learning_rate": 8.88888888888889e-06,
+      "loss": 0.6683,
+      "step": 24
+    },
+    {
+      "epoch": 0.05546311702717693,
+      "grad_norm": 4.444753170013428,
+      "learning_rate": 9.259259259259259e-06,
+      "loss": 0.9003,
+      "step": 25
+    },
+    {
+      "epoch": 0.057681641708264,
+      "grad_norm": 3.3397834300994873,
+      "learning_rate": 9.62962962962963e-06,
+      "loss": 0.6481,
+      "step": 26
+    },
+    {
+      "epoch": 0.059900166389351084,
+      "grad_norm": 3.181568145751953,
+      "learning_rate": 1e-05,
+      "loss": 0.9298,
+      "step": 27
+    },
+    {
+      "epoch": 0.06211869107043816,
+      "grad_norm": 3.4050517082214355,
+      "learning_rate": 1.037037037037037e-05,
+      "loss": 0.6705,
+      "step": 28
+    },
+    {
+      "epoch": 0.06433721575152523,
+      "grad_norm": 3.388437509536743,
+      "learning_rate": 1.074074074074074e-05,
+      "loss": 0.8034,
+      "step": 29
+    },
+    {
+      "epoch": 0.06655574043261231,
+      "grad_norm": 3.5317623615264893,
+      "learning_rate": 1.1111111111111112e-05,
+      "loss": 0.7132,
+      "step": 30
+    },
+    {
+      "epoch": 0.06877426511369938,
+      "grad_norm": 2.980773687362671,
+      "learning_rate": 1.1481481481481482e-05,
+      "loss": 0.7566,
+      "step": 31
+    },
+    {
+      "epoch": 0.07099278979478647,
+      "grad_norm": 3.7015929222106934,
+      "learning_rate": 1.1851851851851853e-05,
+      "loss": 0.899,
+      "step": 32
+    },
+    {
+      "epoch": 0.07321131447587355,
+      "grad_norm": 3.9218454360961914,
+      "learning_rate": 1.2222222222222222e-05,
+      "loss": 0.7822,
+      "step": 33
+    },
+    {
+      "epoch": 0.07542983915696062,
+      "grad_norm": 3.8294360637664795,
+      "learning_rate": 1.2592592592592592e-05,
+      "loss": 0.8198,
+      "step": 34
+    },
+    {
+      "epoch": 0.0776483638380477,
+      "grad_norm": 3.8117635250091553,
+      "learning_rate": 1.2962962962962962e-05,
+      "loss": 0.6697,
+      "step": 35
+    },
+    {
+      "epoch": 0.07986688851913477,
+      "grad_norm": 3.1223931312561035,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.7494,
+      "step": 36
+    },
+    {
+      "epoch": 0.08208541320022185,
+      "grad_norm": 3.1205201148986816,
+      "learning_rate": 1.3703703703703704e-05,
+      "loss": 0.8894,
+      "step": 37
+    },
+    {
+      "epoch": 0.08430393788130892,
+      "grad_norm": 3.58665132522583,
+      "learning_rate": 1.4074074074074075e-05,
+      "loss": 0.8304,
+      "step": 38
+    },
+    {
+      "epoch": 0.08652246256239601,
+      "grad_norm": 3.1765363216400146,
+      "learning_rate": 1.4444444444444444e-05,
+      "loss": 0.6148,
+      "step": 39
+    },
+    {
+      "epoch": 0.08874098724348309,
+      "grad_norm": 3.496825933456421,
+      "learning_rate": 1.4814814814814815e-05,
+      "loss": 0.8399,
+      "step": 40
+    },
+    {
+      "epoch": 0.09095951192457016,
+      "grad_norm": 3.1212782859802246,
+      "learning_rate": 1.5185185185185186e-05,
+      "loss": 0.7935,
+      "step": 41
+    },
+    {
+      "epoch": 0.09317803660565724,
+      "grad_norm": 3.9241397380828857,
+      "learning_rate": 1.5555555555555555e-05,
+      "loss": 0.7515,
+      "step": 42
+    },
+    {
+      "epoch": 0.09539656128674431,
+      "grad_norm": 3.630863666534424,
+      "learning_rate": 1.5925925925925926e-05,
+      "loss": 0.8169,
+      "step": 43
+    },
+    {
+      "epoch": 0.09761508596783139,
+      "grad_norm": 3.252462863922119,
+      "learning_rate": 1.62962962962963e-05,
+      "loss": 0.8349,
+      "step": 44
+    },
+    {
+      "epoch": 0.09983361064891846,
+      "grad_norm": 4.974841117858887,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.8505,
+      "step": 45
+    },
+    {
+      "epoch": 0.10205213533000555,
+      "grad_norm": 4.169401168823242,
+      "learning_rate": 1.7037037037037038e-05,
+      "loss": 0.8311,
+      "step": 46
+    },
+    {
+      "epoch": 0.10427066001109263,
+      "grad_norm": 3.5278749465942383,
+      "learning_rate": 1.740740740740741e-05,
+      "loss": 0.837,
+      "step": 47
+    },
+    {
+      "epoch": 0.1064891846921797,
+      "grad_norm": 3.508436441421509,
+      "learning_rate": 1.777777777777778e-05,
+      "loss": 0.6947,
+      "step": 48
+    },
+    {
+      "epoch": 0.10870770937326678,
+      "grad_norm": 3.6895928382873535,
+      "learning_rate": 1.814814814814815e-05,
+      "loss": 0.906,
+      "step": 49
+    },
+    {
+      "epoch": 0.11092623405435385,
+      "grad_norm": 3.39996600151062,
+      "learning_rate": 1.8518518518518518e-05,
+      "loss": 0.8211,
+      "step": 50
+    },
+    {
+      "epoch": 0.11314475873544093,
+      "grad_norm": 3.421109914779663,
+      "learning_rate": 1.888888888888889e-05,
+      "loss": 0.8386,
+      "step": 51
+    },
+    {
+      "epoch": 0.115363283416528,
+      "grad_norm": 3.0042779445648193,
+      "learning_rate": 1.925925925925926e-05,
+      "loss": 0.8629,
+      "step": 52
+    },
+    {
+      "epoch": 0.11758180809761509,
+      "grad_norm": 4.003059387207031,
+      "learning_rate": 1.962962962962963e-05,
+      "loss": 0.9471,
+      "step": 53
+    },
+    {
+      "epoch": 0.11980033277870217,
+      "grad_norm": 3.4797556400299072,
+      "learning_rate": 2e-05,
+      "loss": 0.965,
+      "step": 54
+    },
+    {
+      "epoch": 0.12201885745978924,
+      "grad_norm": 2.765791177749634,
+      "learning_rate": 2.037037037037037e-05,
+      "loss": 0.6342,
+      "step": 55
+    },
+    {
+      "epoch": 0.12423738214087632,
+      "grad_norm": 3.4333300590515137,
+      "learning_rate": 2.074074074074074e-05,
+      "loss": 0.7253,
+      "step": 56
+    },
+    {
+      "epoch": 0.1264559068219634,
+      "grad_norm": 3.344712972640991,
+      "learning_rate": 2.111111111111111e-05,
+      "loss": 0.7108,
+      "step": 57
+    },
+    {
+      "epoch": 0.12867443150305047,
+      "grad_norm": 3.3873260021209717,
+      "learning_rate": 2.148148148148148e-05,
+      "loss": 0.8208,
+      "step": 58
+    },
+    {
+      "epoch": 0.13089295618413754,
+      "grad_norm": 3.7189722061157227,
+      "learning_rate": 2.1851851851851852e-05,
+      "loss": 0.7697,
+      "step": 59
+    },
+    {
+      "epoch": 0.13311148086522462,
+      "grad_norm": 2.8762967586517334,
+      "learning_rate": 2.2222222222222223e-05,
+      "loss": 0.6172,
+      "step": 60
+    },
+    {
+      "epoch": 0.1353300055463117,
+      "grad_norm": 3.1867575645446777,
+      "learning_rate": 2.2592592592592594e-05,
+      "loss": 0.8589,
+      "step": 61
+    },
+    {
+      "epoch": 0.13754853022739877,
+      "grad_norm": 3.2019455432891846,
+      "learning_rate": 2.2962962962962965e-05,
+      "loss": 0.7937,
+      "step": 62
+    },
+    {
+      "epoch": 0.13976705490848584,
+      "grad_norm": 3.131075382232666,
+      "learning_rate": 2.3333333333333336e-05,
+      "loss": 0.7951,
+      "step": 63
+    },
+    {
+      "epoch": 0.14198557958957295,
+      "grad_norm": 3.212886095046997,
+      "learning_rate": 2.3703703703703707e-05,
+      "loss": 0.7514,
+      "step": 64
+    },
+    {
+      "epoch": 0.14420410427066002,
+      "grad_norm": 2.9653208255767822,
+      "learning_rate": 2.4074074074074074e-05,
+      "loss": 0.8645,
+      "step": 65
+    },
+    {
+      "epoch": 0.1464226289517471,
+      "grad_norm": 3.0649044513702393,
+      "learning_rate": 2.4444444444444445e-05,
+      "loss": 0.7872,
+      "step": 66
+    },
+    {
+      "epoch": 0.14864115363283417,
+      "grad_norm": 2.9362688064575195,
+      "learning_rate": 2.4814814814814816e-05,
+      "loss": 0.8342,
+      "step": 67
+    },
+    {
+      "epoch": 0.15085967831392125,
+      "grad_norm": 2.2857587337493896,
+      "learning_rate": 2.5185185185185183e-05,
+      "loss": 0.5669,
+      "step": 68
+    },
+    {
+      "epoch": 0.15307820299500832,
+      "grad_norm": 3.301044464111328,
+      "learning_rate": 2.5555555555555554e-05,
+      "loss": 0.7959,
+      "step": 69
+    },
+    {
+      "epoch": 0.1552967276760954,
+      "grad_norm": 4.293635845184326,
+      "learning_rate": 2.5925925925925925e-05,
+      "loss": 0.8377,
+      "step": 70
+    },
+    {
+      "epoch": 0.15751525235718247,
+      "grad_norm": 2.7663331031799316,
+      "learning_rate": 2.6296296296296296e-05,
+      "loss": 0.742,
+      "step": 71
+    },
+    {
+      "epoch": 0.15973377703826955,
+      "grad_norm": 3.2686896324157715,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.7594,
+      "step": 72
+    },
+    {
+      "epoch": 0.16195230171935662,
+      "grad_norm": 3.3282413482666016,
+      "learning_rate": 2.7037037037037037e-05,
+      "loss": 0.7416,
+      "step": 73
+    },
+    {
+      "epoch": 0.1641708264004437,
+      "grad_norm": 2.7105166912078857,
+      "learning_rate": 2.7407407407407408e-05,
+      "loss": 0.7617,
+      "step": 74
+    },
+    {
+      "epoch": 0.16638935108153077,
+      "grad_norm": 4.032901763916016,
+      "learning_rate": 2.777777777777778e-05,
+      "loss": 0.7842,
+      "step": 75
+    },
+    {
+      "epoch": 0.16860787576261785,
+      "grad_norm": 2.8986096382141113,
+      "learning_rate": 2.814814814814815e-05,
+      "loss": 1.0779,
+      "step": 76
+    },
+    {
+      "epoch": 0.17082640044370492,
+      "grad_norm": 3.5196783542633057,
+      "learning_rate": 2.851851851851852e-05,
+      "loss": 0.888,
+      "step": 77
+    },
+    {
+      "epoch": 0.17304492512479203,
+      "grad_norm": 2.2407848834991455,
+      "learning_rate": 2.8888888888888888e-05,
+      "loss": 0.733,
+      "step": 78
+    },
+    {
+      "epoch": 0.1752634498058791,
+      "grad_norm": 3.3670687675476074,
+      "learning_rate": 2.925925925925926e-05,
+      "loss": 0.9093,
+      "step": 79
+    },
+    {
+      "epoch": 0.17748197448696618,
+      "grad_norm": 2.894740343093872,
+      "learning_rate": 2.962962962962963e-05,
+      "loss": 0.8175,
+      "step": 80
+    },
+    {
+      "epoch": 0.17970049916805325,
+      "grad_norm": 3.1443374156951904,
+      "learning_rate": 3e-05,
+      "loss": 0.8838,
+      "step": 81
+    },
+    {
+      "epoch": 0.18191902384914033,
+      "grad_norm": 2.384002685546875,
+      "learning_rate": 3.037037037037037e-05,
+      "loss": 0.7301,
+      "step": 82
+    },
+    {
+      "epoch": 0.1841375485302274,
+      "grad_norm": 3.5050556659698486,
+      "learning_rate": 3.074074074074074e-05,
+      "loss": 0.9,
+      "step": 83
+    },
+    {
+      "epoch": 0.18635607321131448,
+      "grad_norm": 2.6660268306732178,
+      "learning_rate": 3.111111111111111e-05,
+      "loss": 0.7188,
+      "step": 84
+    },
+    {
+      "epoch": 0.18857459789240155,
+      "grad_norm": 4.030278205871582,
+      "learning_rate": 3.148148148148148e-05,
+      "loss": 0.7989,
+      "step": 85
+    },
+    {
+      "epoch": 0.19079312257348863,
+      "grad_norm": 3.359692335128784,
+      "learning_rate": 3.185185185185185e-05,
+      "loss": 0.7618,
+      "step": 86
+    },
+    {
+      "epoch": 0.1930116472545757,
+      "grad_norm": 3.2106564044952393,
+      "learning_rate": 3.222222222222223e-05,
+      "loss": 0.8222,
+      "step": 87
+    },
+    {
+      "epoch": 0.19523017193566278,
+      "grad_norm": 3.3968513011932373,
+      "learning_rate": 3.25925925925926e-05,
+      "loss": 0.8334,
+      "step": 88
+    },
+    {
+      "epoch": 0.19744869661674985,
+      "grad_norm": 4.148590564727783,
+      "learning_rate": 3.2962962962962964e-05,
+      "loss": 0.8768,
+      "step": 89
+    },
+    {
+      "epoch": 0.19966722129783693,
+      "grad_norm": 3.382739543914795,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.913,
+      "step": 90
+    },
+    {
+      "epoch": 0.201885745978924,
+      "grad_norm": 3.5297436714172363,
+      "learning_rate": 3.3703703703703706e-05,
+      "loss": 1.0539,
+      "step": 91
+    },
+    {
+      "epoch": 0.2041042706600111,
+      "grad_norm": 2.8654565811157227,
+      "learning_rate": 3.4074074074074077e-05,
+      "loss": 0.886,
+      "step": 92
+    },
+    {
+      "epoch": 0.20632279534109818,
+      "grad_norm": 3.262864351272583,
+      "learning_rate": 3.444444444444445e-05,
+      "loss": 0.9341,
+      "step": 93
+    },
+    {
+      "epoch": 0.20854132002218526,
+      "grad_norm": 2.917818784713745,
+      "learning_rate": 3.481481481481482e-05,
+      "loss": 0.9889,
+      "step": 94
+    },
+    {
+      "epoch": 0.21075984470327233,
+      "grad_norm": 3.4845168590545654,
+      "learning_rate": 3.518518518518519e-05,
+      "loss": 0.9131,
+      "step": 95
+    },
+    {
+      "epoch": 0.2129783693843594,
+      "grad_norm": 3.2672839164733887,
+      "learning_rate": 3.555555555555556e-05,
+      "loss": 1.0542,
+      "step": 96
+    },
+    {
+      "epoch": 0.21519689406544648,
+      "grad_norm": 3.137817621231079,
+      "learning_rate": 3.592592592592593e-05,
+      "loss": 0.7785,
+      "step": 97
+    },
+    {
+      "epoch": 0.21741541874653356,
+      "grad_norm": 2.8803884983062744,
+      "learning_rate": 3.62962962962963e-05,
+      "loss": 0.788,
+      "step": 98
+    },
+    {
+      "epoch": 0.21963394342762063,
+      "grad_norm": 3.895504951477051,
+      "learning_rate": 3.6666666666666666e-05,
+      "loss": 0.9973,
+      "step": 99
+    },
+    {
+      "epoch": 0.2218524681087077,
+      "grad_norm": 3.4226481914520264,
+      "learning_rate": 3.7037037037037037e-05,
+      "loss": 0.9773,
+      "step": 100
+    },
+    {
+      "epoch": 0.22407099278979478,
+      "grad_norm": 3.208644151687622,
+      "learning_rate": 3.740740740740741e-05,
+      "loss": 0.684,
+      "step": 101
+    },
+    {
+      "epoch": 0.22628951747088186,
+      "grad_norm": 2.3386526107788086,
+      "learning_rate": 3.777777777777778e-05,
+      "loss": 0.8159,
+      "step": 102
+    },
+    {
+      "epoch": 0.22850804215196893,
+      "grad_norm": 3.4150688648223877,
+      "learning_rate": 3.814814814814815e-05,
+      "loss": 0.8477,
+      "step": 103
+    },
+    {
+      "epoch": 0.230726566833056,
+      "grad_norm": 3.7421483993530273,
+      "learning_rate": 3.851851851851852e-05,
+      "loss": 0.7845,
+      "step": 104
+    },
+    {
+      "epoch": 0.23294509151414308,
+      "grad_norm": 2.4387078285217285,
+      "learning_rate": 3.888888888888889e-05,
+      "loss": 0.7606,
+      "step": 105
+    },
+    {
+      "epoch": 0.23516361619523019,
+      "grad_norm": 2.6791069507598877,
+      "learning_rate": 3.925925925925926e-05,
+      "loss": 0.7151,
+      "step": 106
+    },
+    {
+      "epoch": 0.23738214087631726,
+      "grad_norm": 3.3553466796875,
+      "learning_rate": 3.962962962962963e-05,
+      "loss": 0.8043,
+      "step": 107
+    },
+    {
+      "epoch": 0.23960066555740434,
+      "grad_norm": 2.8396224975585938,
+      "learning_rate": 4e-05,
+      "loss": 0.8599,
+      "step": 108
+    },
+    {
+      "epoch": 0.2418191902384914,
+      "grad_norm": 3.539052724838257,
+      "learning_rate": 4.0370370370370374e-05,
+      "loss": 1.008,
+      "step": 109
+    },
+    {
+      "epoch": 0.24403771491957849,
+      "grad_norm": 3.3255953788757324,
+      "learning_rate": 4.074074074074074e-05,
+      "loss": 0.945,
+      "step": 110
+    },
+    {
+      "epoch": 0.24625623960066556,
+      "grad_norm": 2.9897146224975586,
+      "learning_rate": 4.111111111111111e-05,
+      "loss": 0.7239,
+      "step": 111
+    },
+    {
+      "epoch": 0.24847476428175264,
+      "grad_norm": 3.553955316543579,
+      "learning_rate": 4.148148148148148e-05,
+      "loss": 0.7633,
+      "step": 112
+    },
+    {
+      "epoch": 0.2506932889628397,
+      "grad_norm": 3.129826784133911,
+      "learning_rate": 4.185185185185185e-05,
+      "loss": 0.8589,
+      "step": 113
+    },
+    {
+      "epoch": 0.2529118136439268,
+      "grad_norm": 3.026575803756714,
+      "learning_rate": 4.222222222222222e-05,
+      "loss": 0.9763,
+      "step": 114
+    },
+    {
+      "epoch": 0.25513033832501386,
+      "grad_norm": 2.42875337600708,
+      "learning_rate": 4.259259259259259e-05,
+      "loss": 0.7803,
+      "step": 115
+    },
+    {
+      "epoch": 0.25734886300610094,
+      "grad_norm": 3.734910726547241,
+      "learning_rate": 4.296296296296296e-05,
+      "loss": 0.9598,
+      "step": 116
+    },
+    {
+      "epoch": 0.259567387687188,
+      "grad_norm": 4.16299295425415,
+      "learning_rate": 4.3333333333333334e-05,
+      "loss": 0.8639,
+      "step": 117
+    },
+    {
+      "epoch": 0.2617859123682751,
+      "grad_norm": 2.8426740169525146,
+      "learning_rate": 4.3703703703703705e-05,
+      "loss": 0.8119,
+      "step": 118
+    },
+    {
+      "epoch": 0.26400443704936216,
+      "grad_norm": 3.0460636615753174,
+      "learning_rate": 4.4074074074074076e-05,
+      "loss": 0.9726,
+      "step": 119
+    },
+    {
+      "epoch": 0.26622296173044924,
+      "grad_norm": 3.437364339828491,
+      "learning_rate": 4.4444444444444447e-05,
+      "loss": 1.1136,
+      "step": 120
+    },
+    {
+      "epoch": 0.2684414864115363,
+      "grad_norm": 3.4276633262634277,
+      "learning_rate": 4.481481481481482e-05,
+      "loss": 0.9224,
+      "step": 121
+    },
+    {
+      "epoch": 0.2706600110926234,
+      "grad_norm": 2.755446195602417,
+      "learning_rate": 4.518518518518519e-05,
+      "loss": 1.0003,
+      "step": 122
+    },
+    {
+      "epoch": 0.27287853577371046,
+      "grad_norm": 2.5793802738189697,
+      "learning_rate": 4.555555555555556e-05,
+      "loss": 0.7428,
+      "step": 123
+    },
+    {
+      "epoch": 0.27509706045479754,
+      "grad_norm": 2.7217273712158203,
+      "learning_rate": 4.592592592592593e-05,
+      "loss": 0.9279,
+      "step": 124
+    },
+    {
+      "epoch": 0.2773155851358846,
+      "grad_norm": 3.0808141231536865,
+      "learning_rate": 4.62962962962963e-05,
+      "loss": 0.9411,
+      "step": 125
+    },
+    {
+      "epoch": 0.2795341098169717,
+      "grad_norm": 2.689789295196533,
+      "learning_rate": 4.666666666666667e-05,
+      "loss": 0.6956,
+      "step": 126
+    },
+    {
+      "epoch": 0.2817526344980588,
+      "grad_norm": 3.1820526123046875,
+      "learning_rate": 4.703703703703704e-05,
+      "loss": 0.7701,
+      "step": 127
+    },
+    {
+      "epoch": 0.2839711591791459,
+      "grad_norm": 3.0511860847473145,
+      "learning_rate": 4.740740740740741e-05,
+      "loss": 0.8727,
+      "step": 128
+    },
+    {
+      "epoch": 0.28618968386023297,
+      "grad_norm": 2.7354345321655273,
+      "learning_rate": 4.7777777777777784e-05,
+      "loss": 0.8789,
+      "step": 129
+    },
+    {
+      "epoch": 0.28840820854132004,
+      "grad_norm": 2.6017563343048096,
+      "learning_rate": 4.814814814814815e-05,
+      "loss": 0.8655,
+      "step": 130
+    },
+    {
+      "epoch": 0.2906267332224071,
+      "grad_norm": 3.055305242538452,
+      "learning_rate": 4.851851851851852e-05,
+      "loss": 0.9921,
+      "step": 131
+    },
+    {
+      "epoch": 0.2928452579034942,
+      "grad_norm": 2.5635929107666016,
+      "learning_rate": 4.888888888888889e-05,
+      "loss": 0.6851,
+      "step": 132
+    },
+    {
+      "epoch": 0.29506378258458127,
+      "grad_norm": 3.0876786708831787,
+      "learning_rate": 4.925925925925926e-05,
+      "loss": 1.0512,
+      "step": 133
+    },
+    {
+      "epoch": 0.29728230726566834,
+      "grad_norm": 2.7007269859313965,
+      "learning_rate": 4.962962962962963e-05,
+      "loss": 0.8667,
+      "step": 134
+    },
+    {
+      "epoch": 0.2995008319467554,
+      "grad_norm": 3.7213735580444336,
+      "learning_rate": 5e-05,
+      "loss": 0.7713,
+      "step": 135
+    },
+    {
+      "epoch": 0.3017193566278425,
+      "grad_norm": 2.7143640518188477,
+      "learning_rate": 4.995884773662552e-05,
+      "loss": 0.7339,
+      "step": 136
+    },
+    {
+      "epoch": 0.30393788130892957,
+      "grad_norm": 2.9997265338897705,
+      "learning_rate": 4.991769547325103e-05,
+      "loss": 1.0332,
+      "step": 137
+    },
+    {
+      "epoch": 0.30615640599001664,
+      "grad_norm": 3.013274669647217,
+      "learning_rate": 4.987654320987655e-05,
+      "loss": 0.8924,
+      "step": 138
+    },
+    {
+      "epoch": 0.3083749306711037,
+      "grad_norm": 2.868612766265869,
+      "learning_rate": 4.983539094650206e-05,
+      "loss": 0.8179,
+      "step": 139
+    },
+    {
+      "epoch": 0.3105934553521908,
+      "grad_norm": 3.173387289047241,
+      "learning_rate": 4.9794238683127575e-05,
+      "loss": 0.9558,
+      "step": 140
+    },
+    {
+      "epoch": 0.31281198003327787,
+      "grad_norm": 3.1992545127868652,
+      "learning_rate": 4.9753086419753084e-05,
+      "loss": 0.7204,
+      "step": 141
+    },
+    {
+      "epoch": 0.31503050471436495,
+      "grad_norm": 3.0312423706054688,
+      "learning_rate": 4.971193415637861e-05,
+      "loss": 0.8439,
+      "step": 142
+    },
+    {
+      "epoch": 0.317249029395452,
+      "grad_norm": 2.8332014083862305,
+      "learning_rate": 4.967078189300412e-05,
+      "loss": 0.875,
+      "step": 143
+    },
+    {
+      "epoch": 0.3194675540765391,
+      "grad_norm": 3.4750418663024902,
+      "learning_rate": 4.962962962962963e-05,
+      "loss": 0.8567,
+      "step": 144
+    },
+    {
+      "epoch": 0.32168607875762617,
+      "grad_norm": 2.9301557540893555,
+      "learning_rate": 4.958847736625515e-05,
+      "loss": 0.8338,
+      "step": 145
+    },
+    {
+      "epoch": 0.32390460343871325,
+      "grad_norm": 2.786050319671631,
+      "learning_rate": 4.9547325102880656e-05,
+      "loss": 0.8408,
+      "step": 146
+    },
+    {
+      "epoch": 0.3261231281198003,
+      "grad_norm": 2.697935104370117,
+      "learning_rate": 4.950617283950618e-05,
+      "loss": 0.6778,
+      "step": 147
+    },
+    {
+      "epoch": 0.3283416528008874,
+      "grad_norm": 2.706058979034424,
+      "learning_rate": 4.946502057613169e-05,
+      "loss": 0.8561,
+      "step": 148
+    },
+    {
+      "epoch": 0.33056017748197447,
+      "grad_norm": 2.853421449661255,
+      "learning_rate": 4.9423868312757204e-05,
+      "loss": 0.9226,
+      "step": 149
+    },
+    {
+      "epoch": 0.33277870216306155,
+      "grad_norm": 2.949120044708252,
+      "learning_rate": 4.938271604938271e-05,
+      "loss": 0.905,
+      "step": 150
+    },
+    {
+      "epoch": 0.3349972268441486,
+      "grad_norm": 2.4828479290008545,
+      "learning_rate": 4.9341563786008236e-05,
+      "loss": 0.869,
+      "step": 151
+    },
+    {
+      "epoch": 0.3372157515252357,
+      "grad_norm": 3.9800920486450195,
+      "learning_rate": 4.930041152263375e-05,
+      "loss": 0.9813,
+      "step": 152
+    },
+    {
+      "epoch": 0.33943427620632277,
+      "grad_norm": 2.8985257148742676,
+      "learning_rate": 4.925925925925926e-05,
+      "loss": 0.985,
+      "step": 153
+    },
+    {
+      "epoch": 0.34165280088740985,
+      "grad_norm": 2.758923053741455,
+      "learning_rate": 4.9218106995884777e-05,
+      "loss": 0.9377,
+      "step": 154
+    },
+    {
+      "epoch": 0.343871325568497,
+      "grad_norm": 3.606003999710083,
+      "learning_rate": 4.9176954732510286e-05,
+      "loss": 1.0176,
+      "step": 155
+    },
+    {
+      "epoch": 0.34608985024958405,
+      "grad_norm": 2.7474193572998047,
+      "learning_rate": 4.913580246913581e-05,
+      "loss": 0.9892,
+      "step": 156
+    },
+    {
+      "epoch": 0.3483083749306711,
+      "grad_norm": 3.412952184677124,
+      "learning_rate": 4.909465020576132e-05,
+      "loss": 0.8605,
+      "step": 157
+    },
+    {
+      "epoch": 0.3505268996117582,
+      "grad_norm": 2.6602165699005127,
+      "learning_rate": 4.905349794238683e-05,
+      "loss": 0.9364,
+      "step": 158
+    },
+    {
+      "epoch": 0.3527454242928453,
+      "grad_norm": 2.8943512439727783,
+      "learning_rate": 4.901234567901235e-05,
+      "loss": 0.7349,
+      "step": 159
+    },
+    {
+      "epoch": 0.35496394897393235,
+      "grad_norm": 3.232520818710327,
+      "learning_rate": 4.8971193415637865e-05,
+      "loss": 0.7483,
+      "step": 160
+    },
+    {
+      "epoch": 0.35718247365501943,
+      "grad_norm": 2.8281142711639404,
+      "learning_rate": 4.893004115226338e-05,
+      "loss": 0.9272,
+      "step": 161
+    },
+    {
+      "epoch": 0.3594009983361065,
+      "grad_norm": 2.8702096939086914,
+      "learning_rate": 4.888888888888889e-05,
+      "loss": 0.9922,
+      "step": 162
+    },
+    {
+      "epoch": 0.3616195230171936,
+      "grad_norm": 2.885303020477295,
+      "learning_rate": 4.8847736625514406e-05,
+      "loss": 0.8412,
+      "step": 163
+    },
+    {
+      "epoch": 0.36383804769828065,
+      "grad_norm": 2.4432108402252197,
+      "learning_rate": 4.8806584362139915e-05,
+      "loss": 0.7666,
+      "step": 164
+    },
+    {
+      "epoch": 0.36605657237936773,
+      "grad_norm": 2.540187358856201,
+      "learning_rate": 4.876543209876544e-05,
+      "loss": 0.7657,
+      "step": 165
+    },
+    {
+      "epoch": 0.3682750970604548,
+      "grad_norm": 3.024909496307373,
+      "learning_rate": 4.872427983539095e-05,
+      "loss": 0.974,
+      "step": 166
+    },
+    {
+      "epoch": 0.3704936217415419,
+      "grad_norm": 2.8133575916290283,
+      "learning_rate": 4.868312757201646e-05,
+      "loss": 0.8997,
+      "step": 167
+    },
+    {
+      "epoch": 0.37271214642262895,
+      "grad_norm": 2.4253928661346436,
+      "learning_rate": 4.864197530864198e-05,
+      "loss": 0.8412,
+      "step": 168
+    },
+    {
+      "epoch": 0.37493067110371603,
+      "grad_norm": 3.4854648113250732,
+      "learning_rate": 4.860082304526749e-05,
+      "loss": 0.7213,
+      "step": 169
+    },
+    {
+      "epoch": 0.3771491957848031,
+      "grad_norm": 2.793919801712036,
+      "learning_rate": 4.855967078189301e-05,
+      "loss": 0.9483,
+      "step": 170
+    },
+    {
+      "epoch": 0.3793677204658902,
+      "grad_norm": 2.8342771530151367,
+      "learning_rate": 4.851851851851852e-05,
+      "loss": 0.96,
+      "step": 171
+    },
+    {
+      "epoch": 0.38158624514697725,
+      "grad_norm": 3.328544855117798,
+      "learning_rate": 4.8477366255144035e-05,
+      "loss": 0.7988,
+      "step": 172
+    },
+    {
+      "epoch": 0.38380476982806433,
+      "grad_norm": 2.5279934406280518,
+      "learning_rate": 4.843621399176955e-05,
+      "loss": 0.9058,
+      "step": 173
+    },
+    {
+      "epoch": 0.3860232945091514,
+      "grad_norm": 3.1248507499694824,
+      "learning_rate": 4.8395061728395067e-05,
+      "loss": 1.009,
+      "step": 174
+    },
+    {
+      "epoch": 0.3882418191902385,
+      "grad_norm": 2.961460828781128,
+      "learning_rate": 4.835390946502058e-05,
+      "loss": 0.9253,
+      "step": 175
+    },
+    {
+      "epoch": 0.39046034387132555,
+      "grad_norm": 2.605442762374878,
+      "learning_rate": 4.831275720164609e-05,
+      "loss": 0.9026,
+      "step": 176
+    },
+    {
+      "epoch": 0.39267886855241263,
+      "grad_norm": 3.4075448513031006,
+      "learning_rate": 4.827160493827161e-05,
+      "loss": 0.9607,
+      "step": 177
+    },
+    {
+      "epoch": 0.3948973932334997,
+      "grad_norm": 2.7340664863586426,
+      "learning_rate": 4.8230452674897116e-05,
+      "loss": 0.8057,
+      "step": 178
+    },
+    {
+      "epoch": 0.3971159179145868,
+      "grad_norm": 2.6199333667755127,
+      "learning_rate": 4.818930041152264e-05,
+      "loss": 0.9926,
+      "step": 179
+    },
+    {
+      "epoch": 0.39933444259567386,
+      "grad_norm": 3.1111562252044678,
+      "learning_rate": 4.814814814814815e-05,
+      "loss": 0.868,
+      "step": 180
+    },
+    {
+      "epoch": 0.40155296727676093,
+      "grad_norm": 3.1567575931549072,
+      "learning_rate": 4.8106995884773664e-05,
+      "loss": 0.9678,
+      "step": 181
+    },
+    {
+      "epoch": 0.403771491957848,
+      "grad_norm": 2.307004928588867,
+      "learning_rate": 4.806584362139918e-05,
+      "loss": 0.7731,
+      "step": 182
+    },
+    {
+      "epoch": 0.4059900166389351,
+      "grad_norm": 2.7701804637908936,
+      "learning_rate": 4.8024691358024696e-05,
+      "loss": 0.7931,
+      "step": 183
+    },
+    {
+      "epoch": 0.4082085413200222,
+      "grad_norm": 2.952460289001465,
+      "learning_rate": 4.798353909465021e-05,
+      "loss": 0.9942,
+      "step": 184
+    },
+    {
+      "epoch": 0.4104270660011093,
+      "grad_norm": 2.1727120876312256,
+      "learning_rate": 4.794238683127572e-05,
+      "loss": 0.6968,
+      "step": 185
+    },
+    {
+      "epoch": 0.41264559068219636,
+      "grad_norm": 3.7534117698669434,
+      "learning_rate": 4.7901234567901237e-05,
+      "loss": 0.9999,
+      "step": 186
+    },
+    {
+      "epoch": 0.41486411536328344,
+      "grad_norm": 3.3388822078704834,
+      "learning_rate": 4.7860082304526746e-05,
+      "loss": 0.9372,
+      "step": 187
+    },
+    {
+      "epoch": 0.4170826400443705,
+      "grad_norm": 2.8070921897888184,
+      "learning_rate": 4.781893004115227e-05,
+      "loss": 0.8618,
+      "step": 188
+    },
+    {
+      "epoch": 0.4193011647254576,
+      "grad_norm": 3.2333202362060547,
+      "learning_rate": 4.7777777777777784e-05,
+      "loss": 1.1612,
+      "step": 189
+    },
+    {
+      "epoch": 0.42151968940654466,
+      "grad_norm": 2.213495969772339,
+      "learning_rate": 4.773662551440329e-05,
+      "loss": 0.8153,
+      "step": 190
+    },
+    {
+      "epoch": 0.42373821408763174,
+      "grad_norm": 2.3913533687591553,
+      "learning_rate": 4.769547325102881e-05,
+      "loss": 0.7256,
+      "step": 191
+    },
+    {
+      "epoch": 0.4259567387687188,
+      "grad_norm": 2.691166639328003,
+      "learning_rate": 4.7654320987654325e-05,
+      "loss": 0.9205,
+      "step": 192
+    },
+    {
+      "epoch": 0.4281752634498059,
+      "grad_norm": 2.31721830368042,
+      "learning_rate": 4.761316872427984e-05,
+      "loss": 0.8315,
+      "step": 193
+    },
+    {
+      "epoch": 0.43039378813089296,
+      "grad_norm": 2.9450769424438477,
+      "learning_rate": 4.757201646090535e-05,
+      "loss": 0.7593,
+      "step": 194
+    },
+    {
+      "epoch": 0.43261231281198004,
+      "grad_norm": 2.5545473098754883,
+      "learning_rate": 4.7530864197530866e-05,
+      "loss": 0.8137,
+      "step": 195
+    },
+    {
+      "epoch": 0.4348308374930671,
+      "grad_norm": 2.7791800498962402,
+      "learning_rate": 4.748971193415638e-05,
+      "loss": 1.1224,
+      "step": 196
+    },
+    {
+      "epoch": 0.4370493621741542,
+      "grad_norm": 2.9285545349121094,
+      "learning_rate": 4.74485596707819e-05,
+      "loss": 0.9109,
+      "step": 197
+    },
+    {
+      "epoch": 0.43926788685524126,
+      "grad_norm": 2.4910876750946045,
+      "learning_rate": 4.740740740740741e-05,
+      "loss": 0.8531,
+      "step": 198
+    },
+    {
+      "epoch": 0.44148641153632834,
+      "grad_norm": 2.7596633434295654,
+      "learning_rate": 4.736625514403292e-05,
+      "loss": 0.9492,
+      "step": 199
+    },
+    {
+      "epoch": 0.4437049362174154,
+      "grad_norm": 2.132411479949951,
+      "learning_rate": 4.732510288065844e-05,
+      "loss": 0.8367,
+      "step": 200
+    },
+    {
+      "epoch": 0.4459234608985025,
+      "grad_norm": 2.6128880977630615,
+      "learning_rate": 4.7283950617283954e-05,
+      "loss": 0.881,
+      "step": 201
+    },
+    {
+      "epoch": 0.44814198557958956,
+      "grad_norm": 2.595998764038086,
+      "learning_rate": 4.724279835390947e-05,
+      "loss": 0.7445,
+      "step": 202
+    },
+    {
+      "epoch": 0.45036051026067664,
+      "grad_norm": 2.4865283966064453,
+      "learning_rate": 4.7201646090534986e-05,
+      "loss": 0.9057,
+      "step": 203
+    },
+    {
+      "epoch": 0.4525790349417637,
+      "grad_norm": 3.1629889011383057,
+      "learning_rate": 4.7160493827160495e-05,
+      "loss": 1.1854,
+      "step": 204
+    },
+    {
+      "epoch": 0.4547975596228508,
+      "grad_norm": 2.9386508464813232,
+      "learning_rate": 4.711934156378601e-05,
+      "loss": 1.0335,
+      "step": 205
+    },
+    {
+      "epoch": 0.45701608430393786,
+      "grad_norm": 2.5068747997283936,
+      "learning_rate": 4.7078189300411527e-05,
+      "loss": 0.7898,
+      "step": 206
+    },
+    {
+      "epoch": 0.45923460898502494,
+      "grad_norm": 2.4851531982421875,
+      "learning_rate": 4.703703703703704e-05,
+      "loss": 0.9578,
+      "step": 207
+    },
+    {
+      "epoch": 0.461453133666112,
+      "grad_norm": 2.7017292976379395,
+      "learning_rate": 4.699588477366255e-05,
+      "loss": 0.833,
+      "step": 208
+    },
+    {
+      "epoch": 0.4636716583471991,
+      "grad_norm": 2.341001510620117,
+      "learning_rate": 4.695473251028807e-05,
+      "loss": 0.7142,
+      "step": 209
+    },
+    {
+      "epoch": 0.46589018302828616,
+      "grad_norm": 2.5884008407592773,
+      "learning_rate": 4.691358024691358e-05,
+      "loss": 0.917,
+      "step": 210
+    },
+    {
+      "epoch": 0.46810870770937324,
+      "grad_norm": 2.3509302139282227,
+      "learning_rate": 4.68724279835391e-05,
+      "loss": 0.7935,
+      "step": 211
+    },
+    {
+      "epoch": 0.47032723239046037,
+      "grad_norm": 2.5616133213043213,
+      "learning_rate": 4.6831275720164615e-05,
+      "loss": 0.8914,
+      "step": 212
+    },
+    {
+      "epoch": 0.47254575707154745,
+      "grad_norm": 3.368060350418091,
+      "learning_rate": 4.6790123456790124e-05,
+      "loss": 0.9117,
+      "step": 213
+    },
+    {
+      "epoch": 0.4747642817526345,
+      "grad_norm": 2.88454270362854,
+      "learning_rate": 4.674897119341564e-05,
+      "loss": 0.8665,
+      "step": 214
+    },
+    {
+      "epoch": 0.4769828064337216,
+      "grad_norm": 2.7611846923828125,
+      "learning_rate": 4.6707818930041156e-05,
+      "loss": 0.8364,
+      "step": 215
+    },
+    {
+      "epoch": 0.47920133111480867,
+      "grad_norm": 2.909740924835205,
+      "learning_rate": 4.666666666666667e-05,
+      "loss": 0.8931,
+      "step": 216
+    },
+    {
+      "epoch": 0.48141985579589575,
+      "grad_norm": 2.7494494915008545,
+      "learning_rate": 4.662551440329218e-05,
+      "loss": 0.8473,
+      "step": 217
+    },
+    {
+      "epoch": 0.4836383804769828,
+      "grad_norm": 2.691772699356079,
+      "learning_rate": 4.6584362139917697e-05,
+      "loss": 0.8426,
+      "step": 218
+    },
+    {
+      "epoch": 0.4858569051580699,
+      "grad_norm": 2.6373000144958496,
+      "learning_rate": 4.654320987654321e-05,
+      "loss": 0.7732,
+      "step": 219
+    },
+    {
+      "epoch": 0.48807542983915697,
+      "grad_norm": 2.3603529930114746,
+      "learning_rate": 4.650205761316873e-05,
+      "loss": 0.8127,
+      "step": 220
+    },
+    {
+      "epoch": 0.49029395452024405,
+      "grad_norm": 2.701234817504883,
+      "learning_rate": 4.6460905349794244e-05,
+      "loss": 0.9838,
+      "step": 221
+    },
+    {
+      "epoch": 0.4925124792013311,
+      "grad_norm": 2.381791353225708,
+      "learning_rate": 4.641975308641975e-05,
+      "loss": 0.8845,
+      "step": 222
+    },
+    {
+      "epoch": 0.4947310038824182,
+      "grad_norm": 1.8874443769454956,
+      "learning_rate": 4.637860082304527e-05,
+      "loss": 0.9486,
+      "step": 223
+    },
+    {
+      "epoch": 0.49694952856350527,
+      "grad_norm": 2.5028183460235596,
+      "learning_rate": 4.6337448559670785e-05,
+      "loss": 0.9471,
+      "step": 224
+    },
+    {
+      "epoch": 0.49916805324459235,
+      "grad_norm": 2.597698926925659,
+      "learning_rate": 4.62962962962963e-05,
+      "loss": 0.7997,
+      "step": 225
+    },
+    {
+      "epoch": 0.5013865779256794,
+      "grad_norm": 2.791551113128662,
+      "learning_rate": 4.625514403292182e-05,
+      "loss": 0.9639,
+      "step": 226
+    },
+    {
+      "epoch": 0.5036051026067665,
+      "grad_norm": 2.609265089035034,
+      "learning_rate": 4.6213991769547326e-05,
+      "loss": 1.1101,
+      "step": 227
+    },
+    {
+      "epoch": 0.5058236272878536,
+      "grad_norm": 2.3793084621429443,
+      "learning_rate": 4.617283950617284e-05,
+      "loss": 0.8327,
+      "step": 228
+    },
+    {
+      "epoch": 0.5080421519689406,
+      "grad_norm": 2.4891560077667236,
+      "learning_rate": 4.613168724279836e-05,
+      "loss": 0.9444,
+      "step": 229
+    },
+    {
+      "epoch": 0.5102606766500277,
+      "grad_norm": 2.418086290359497,
+      "learning_rate": 4.609053497942387e-05,
+      "loss": 0.8786,
+      "step": 230
+    },
+    {
+      "epoch": 0.5124792013311148,
+      "grad_norm": 3.385559558868408,
+      "learning_rate": 4.604938271604938e-05,
+      "loss": 0.7347,
+      "step": 231
+    },
+    {
+      "epoch": 0.5146977260122019,
+      "grad_norm": 2.807567596435547,
+      "learning_rate": 4.60082304526749e-05,
+      "loss": 0.9081,
+      "step": 232
+    },
+    {
+      "epoch": 0.516916250693289,
+      "grad_norm": 2.7700653076171875,
+      "learning_rate": 4.5967078189300414e-05,
+      "loss": 0.9226,
+      "step": 233
+    },
+    {
+      "epoch": 0.519134775374376,
+      "grad_norm": 2.584984302520752,
+      "learning_rate": 4.592592592592593e-05,
+      "loss": 0.7871,
+      "step": 234
+    },
+    {
+      "epoch": 0.5213533000554631,
+      "grad_norm": 3.6548373699188232,
+      "learning_rate": 4.5884773662551446e-05,
+      "loss": 0.7452,
+      "step": 235
+    },
+    {
+      "epoch": 0.5235718247365502,
+      "grad_norm": 2.438746690750122,
+      "learning_rate": 4.5843621399176955e-05,
+      "loss": 0.8832,
+      "step": 236
+    },
+    {
+      "epoch": 0.5257903494176372,
+      "grad_norm": 2.82358717918396,
+      "learning_rate": 4.580246913580247e-05,
+      "loss": 0.9254,
+      "step": 237
+    },
+    {
+      "epoch": 0.5280088740987243,
+      "grad_norm": 2.6896610260009766,
+      "learning_rate": 4.5761316872427987e-05,
+      "loss": 0.9515,
+      "step": 238
+    },
+    {
+      "epoch": 0.5302273987798114,
+      "grad_norm": 2.646381378173828,
+      "learning_rate": 4.57201646090535e-05,
+      "loss": 0.8091,
+      "step": 239
+    },
+    {
+      "epoch": 0.5324459234608985,
+      "grad_norm": 2.35813570022583,
+      "learning_rate": 4.567901234567901e-05,
+      "loss": 0.7459,
+      "step": 240
+    },
+    {
+      "epoch": 0.5346644481419855,
+      "grad_norm": 2.5789496898651123,
+      "learning_rate": 4.563786008230453e-05,
+      "loss": 0.9163,
+      "step": 241
+    },
+    {
+      "epoch": 0.5368829728230726,
+      "grad_norm": 3.0104360580444336,
+      "learning_rate": 4.559670781893004e-05,
+      "loss": 0.8254,
+      "step": 242
+    },
+    {
+      "epoch": 0.5391014975041597,
+      "grad_norm": 2.7714197635650635,
+      "learning_rate": 4.555555555555556e-05,
+      "loss": 0.7869,
+      "step": 243
+    },
+    {
+      "epoch": 0.5413200221852468,
+      "grad_norm": 2.5280089378356934,
+      "learning_rate": 4.5514403292181075e-05,
+      "loss": 0.8252,
+      "step": 244
+    },
+    {
+      "epoch": 0.5435385468663338,
+      "grad_norm": 3.157022714614868,
+      "learning_rate": 4.5473251028806584e-05,
+      "loss": 0.885,
+      "step": 245
+    },
+    {
+      "epoch": 0.5457570715474209,
+      "grad_norm": 2.9378457069396973,
+      "learning_rate": 4.54320987654321e-05,
+      "loss": 0.7652,
+      "step": 246
+    },
+    {
+      "epoch": 0.547975596228508,
+      "grad_norm": 2.1952061653137207,
+      "learning_rate": 4.5390946502057616e-05,
+      "loss": 0.8186,
+      "step": 247
+    },
+    {
+      "epoch": 0.5501941209095951,
+      "grad_norm": 2.1745641231536865,
+      "learning_rate": 4.534979423868313e-05,
+      "loss": 0.6305,
+      "step": 248
+    },
+    {
+      "epoch": 0.5524126455906821,
+      "grad_norm": 2.0844204425811768,
+      "learning_rate": 4.530864197530865e-05,
+      "loss": 0.7635,
+      "step": 249
+    },
+    {
+      "epoch": 0.5546311702717692,
+      "grad_norm": 2.8574438095092773,
+      "learning_rate": 4.5267489711934157e-05,
+      "loss": 0.7288,
+      "step": 250
+    },
+    {
+      "epoch": 0.5568496949528563,
+      "grad_norm": 3.5157852172851562,
+      "learning_rate": 4.522633744855967e-05,
+      "loss": 0.9295,
+      "step": 251
+    },
+    {
+      "epoch": 0.5590682196339434,
+      "grad_norm": 2.7382731437683105,
+      "learning_rate": 4.518518518518519e-05,
+      "loss": 0.8042,
+      "step": 252
+    },
+    {
+      "epoch": 0.5612867443150305,
+      "grad_norm": 2.332974910736084,
+      "learning_rate": 4.5144032921810704e-05,
+      "loss": 0.8973,
+      "step": 253
+    },
+    {
+      "epoch": 0.5635052689961176,
+      "grad_norm": 2.8559820652008057,
+      "learning_rate": 4.510288065843621e-05,
+      "loss": 0.9496,
+      "step": 254
+    },
+    {
+      "epoch": 0.5657237936772047,
+      "grad_norm": 2.241053581237793,
+      "learning_rate": 4.506172839506173e-05,
+      "loss": 0.7976,
+      "step": 255
+    },
+    {
+      "epoch": 0.5679423183582918,
+      "grad_norm": 2.5193064212799072,
+      "learning_rate": 4.5020576131687245e-05,
+      "loss": 0.722,
+      "step": 256
+    },
+    {
+      "epoch": 0.5701608430393789,
+      "grad_norm": 2.3484842777252197,
+      "learning_rate": 4.497942386831276e-05,
+      "loss": 0.7881,
+      "step": 257
+    },
+    {
+      "epoch": 0.5723793677204659,
+      "grad_norm": 3.2392523288726807,
+      "learning_rate": 4.493827160493828e-05,
+      "loss": 0.9439,
+      "step": 258
+    },
+    {
+      "epoch": 0.574597892401553,
+      "grad_norm": 2.176273822784424,
+      "learning_rate": 4.4897119341563786e-05,
+      "loss": 0.8624,
+      "step": 259
+    },
+    {
+      "epoch": 0.5768164170826401,
+      "grad_norm": 2.214510202407837,
+      "learning_rate": 4.48559670781893e-05,
+      "loss": 0.9,
+      "step": 260
+    },
+    {
+      "epoch": 0.5790349417637272,
+      "grad_norm": 3.1795337200164795,
+      "learning_rate": 4.481481481481482e-05,
+      "loss": 1.1315,
+      "step": 261
+    },
+    {
+      "epoch": 0.5812534664448142,
+      "grad_norm": 2.2774133682250977,
+      "learning_rate": 4.477366255144033e-05,
+      "loss": 0.6617,
+      "step": 262
+    },
+    {
+      "epoch": 0.5834719911259013,
+      "grad_norm": 2.3901000022888184,
+      "learning_rate": 4.473251028806584e-05,
+      "loss": 0.7379,
+      "step": 263
+    },
+    {
+      "epoch": 0.5856905158069884,
+      "grad_norm": 2.8346636295318604,
+      "learning_rate": 4.469135802469136e-05,
+      "loss": 0.7712,
+      "step": 264
+    },
+    {
+      "epoch": 0.5879090404880755,
+      "grad_norm": 2.362989664077759,
+      "learning_rate": 4.4650205761316874e-05,
+      "loss": 0.8837,
+      "step": 265
+    },
+    {
+      "epoch": 0.5901275651691625,
+      "grad_norm": 2.596905469894409,
+      "learning_rate": 4.460905349794239e-05,
+      "loss": 0.9551,
+      "step": 266
+    },
+    {
+      "epoch": 0.5923460898502496,
+      "grad_norm": 2.465303659439087,
+      "learning_rate": 4.4567901234567906e-05,
+      "loss": 0.82,
+      "step": 267
+    },
+    {
+      "epoch": 0.5945646145313367,
+      "grad_norm": 2.4091827869415283,
+      "learning_rate": 4.4526748971193415e-05,
+      "loss": 1.0148,
+      "step": 268
+    },
+    {
+      "epoch": 0.5967831392124238,
+      "grad_norm": 2.312535524368286,
+      "learning_rate": 4.448559670781893e-05,
+      "loss": 1.0038,
+      "step": 269
+    },
+    {
+      "epoch": 0.5990016638935108,
+      "grad_norm": 2.774113655090332,
+      "learning_rate": 4.4444444444444447e-05,
+      "loss": 0.9275,
+      "step": 270
+    },
+    {
+      "epoch": 0.6012201885745979,
+      "grad_norm": 2.962479829788208,
+      "learning_rate": 4.440329218106996e-05,
+      "loss": 0.9749,
+      "step": 271
+    },
+    {
+      "epoch": 0.603438713255685,
+      "grad_norm": 2.3545751571655273,
+      "learning_rate": 4.436213991769548e-05,
+      "loss": 0.6651,
+      "step": 272
+    },
+    {
+      "epoch": 0.6056572379367721,
+      "grad_norm": 2.711094379425049,
+      "learning_rate": 4.432098765432099e-05,
+      "loss": 0.9291,
+      "step": 273
+    },
+    {
+      "epoch": 0.6078757626178591,
+      "grad_norm": 2.4724483489990234,
+      "learning_rate": 4.42798353909465e-05,
+      "loss": 0.9587,
+      "step": 274
+    },
+    {
+      "epoch": 0.6100942872989462,
+      "grad_norm": 1.9685262441635132,
+      "learning_rate": 4.423868312757202e-05,
+      "loss": 0.6659,
+      "step": 275
+    },
+    {
+      "epoch": 0.6123128119800333,
+      "grad_norm": 2.3737220764160156,
+      "learning_rate": 4.4197530864197535e-05,
+      "loss": 0.9022,
+      "step": 276
+    },
+    {
+      "epoch": 0.6145313366611204,
+      "grad_norm": 2.2392120361328125,
+      "learning_rate": 4.4156378600823044e-05,
+      "loss": 0.8572,
+      "step": 277
+    },
+    {
+      "epoch": 0.6167498613422074,
+      "grad_norm": 2.8564071655273438,
+      "learning_rate": 4.411522633744856e-05,
+      "loss": 0.7826,
+      "step": 278
+    },
+    {
+      "epoch": 0.6189683860232945,
+      "grad_norm": 2.136309862136841,
+      "learning_rate": 4.4074074074074076e-05,
+      "loss": 0.8055,
+      "step": 279
+    },
+    {
+      "epoch": 0.6211869107043816,
+      "grad_norm": 2.1790919303894043,
+      "learning_rate": 4.403292181069959e-05,
+      "loss": 0.7438,
+      "step": 280
+    },
+    {
+      "epoch": 0.6234054353854687,
+      "grad_norm": 2.3335819244384766,
+      "learning_rate": 4.399176954732511e-05,
+      "loss": 0.7979,
+      "step": 281
+    },
+    {
+      "epoch": 0.6256239600665557,
+      "grad_norm": 3.1455516815185547,
+      "learning_rate": 4.3950617283950617e-05,
+      "loss": 0.7508,
+      "step": 282
+    },
+    {
+      "epoch": 0.6278424847476428,
+      "grad_norm": 1.9712636470794678,
+      "learning_rate": 4.390946502057613e-05,
+      "loss": 0.6198,
+      "step": 283
+    },
+    {
+      "epoch": 0.6300610094287299,
+      "grad_norm": 2.3927738666534424,
+      "learning_rate": 4.386831275720165e-05,
+      "loss": 0.7708,
+      "step": 284
+    },
+    {
+      "epoch": 0.632279534109817,
+      "grad_norm": 2.345599412918091,
+      "learning_rate": 4.3827160493827164e-05,
+      "loss": 0.7693,
+      "step": 285
+    },
+    {
+      "epoch": 0.634498058790904,
+      "grad_norm": 2.6853535175323486,
+      "learning_rate": 4.378600823045268e-05,
+      "loss": 0.7984,
+      "step": 286
+    },
+    {
+      "epoch": 0.6367165834719911,
+      "grad_norm": 2.5671777725219727,
+      "learning_rate": 4.374485596707819e-05,
+      "loss": 0.8955,
+      "step": 287
+    },
+    {
+      "epoch": 0.6389351081530782,
+      "grad_norm": 3.7215967178344727,
+      "learning_rate": 4.3703703703703705e-05,
+      "loss": 0.8439,
+      "step": 288
+    },
+    {
+      "epoch": 0.6411536328341653,
+      "grad_norm": 2.1262567043304443,
+      "learning_rate": 4.366255144032922e-05,
+      "loss": 0.9158,
+      "step": 289
+    },
+    {
+      "epoch": 0.6433721575152523,
+      "grad_norm": 3.3655998706817627,
+      "learning_rate": 4.3621399176954737e-05,
+      "loss": 0.8424,
+      "step": 290
+    },
+    {
+      "epoch": 0.6455906821963394,
+      "grad_norm": 3.093371629714966,
+      "learning_rate": 4.3580246913580246e-05,
+      "loss": 0.9434,
+      "step": 291
+    },
+    {
+      "epoch": 0.6478092068774265,
+      "grad_norm": 2.442232131958008,
+      "learning_rate": 4.353909465020576e-05,
+      "loss": 0.9574,
+      "step": 292
+    },
+    {
+      "epoch": 0.6500277315585136,
+      "grad_norm": 2.7226009368896484,
+      "learning_rate": 4.349794238683128e-05,
+      "loss": 0.8657,
+      "step": 293
+    },
+    {
+      "epoch": 0.6522462562396006,
+      "grad_norm": 2.5951650142669678,
+      "learning_rate": 4.345679012345679e-05,
+      "loss": 0.816,
+      "step": 294
+    },
+    {
+      "epoch": 0.6544647809206877,
+      "grad_norm": 2.3298048973083496,
+      "learning_rate": 4.341563786008231e-05,
+      "loss": 0.6767,
+      "step": 295
+    },
+    {
+      "epoch": 0.6566833056017748,
+      "grad_norm": 2.63897705078125,
+      "learning_rate": 4.337448559670782e-05,
+      "loss": 0.9529,
+      "step": 296
+    },
+    {
+      "epoch": 0.6589018302828619,
+      "grad_norm": 2.2566025257110596,
+      "learning_rate": 4.3333333333333334e-05,
+      "loss": 0.7444,
+      "step": 297
+    },
+    {
+      "epoch": 0.6611203549639489,
+      "grad_norm": 2.4676222801208496,
+      "learning_rate": 4.329218106995885e-05,
+      "loss": 0.7921,
+      "step": 298
+    },
+    {
+      "epoch": 0.663338879645036,
+      "grad_norm": 3.1744277477264404,
+      "learning_rate": 4.3251028806584366e-05,
+      "loss": 1.0412,
+      "step": 299
+    },
+    {
+      "epoch": 0.6655574043261231,
+      "grad_norm": 2.926365852355957,
+      "learning_rate": 4.3209876543209875e-05,
+      "loss": 1.0081,
+      "step": 300
+    },
+    {
+      "epoch": 0.6677759290072102,
+      "grad_norm": 2.833530902862549,
+      "learning_rate": 4.316872427983539e-05,
+      "loss": 0.7298,
+      "step": 301
+    },
+    {
+      "epoch": 0.6699944536882972,
+      "grad_norm": 2.4643208980560303,
+      "learning_rate": 4.3127572016460907e-05,
+      "loss": 0.8966,
+      "step": 302
+    },
+    {
+      "epoch": 0.6722129783693843,
+      "grad_norm": 2.0077173709869385,
+      "learning_rate": 4.308641975308642e-05,
+      "loss": 0.7604,
+      "step": 303
+    },
+    {
+      "epoch": 0.6744315030504714,
+      "grad_norm": 2.417147159576416,
+      "learning_rate": 4.304526748971194e-05,
+      "loss": 0.6983,
+      "step": 304
+    },
+    {
+      "epoch": 0.6766500277315585,
+      "grad_norm": 2.4896717071533203,
+      "learning_rate": 4.300411522633745e-05,
+      "loss": 0.9081,
+      "step": 305
+    },
+    {
+      "epoch": 0.6788685524126455,
+      "grad_norm": 2.648681163787842,
+      "learning_rate": 4.296296296296296e-05,
+      "loss": 0.7394,
+      "step": 306
+    },
+    {
+      "epoch": 0.6810870770937326,
+      "grad_norm": 2.484480857849121,
+      "learning_rate": 4.292181069958848e-05,
+      "loss": 0.8846,
+      "step": 307
+    },
+    {
+      "epoch": 0.6833056017748197,
+      "grad_norm": 2.686354875564575,
+      "learning_rate": 4.2880658436213995e-05,
+      "loss": 0.9162,
+      "step": 308
+    },
+    {
+      "epoch": 0.6855241264559068,
+      "grad_norm": 2.4837875366210938,
+      "learning_rate": 4.283950617283951e-05,
+      "loss": 0.7917,
+      "step": 309
+    },
+    {
+      "epoch": 0.687742651136994,
+      "grad_norm": 2.7384748458862305,
+      "learning_rate": 4.279835390946502e-05,
+      "loss": 0.7378,
+      "step": 310
+    },
+    {
+      "epoch": 0.689961175818081,
+      "grad_norm": 2.743865966796875,
+      "learning_rate": 4.2757201646090536e-05,
+      "loss": 0.9542,
+      "step": 311
+    },
+    {
+      "epoch": 0.6921797004991681,
+      "grad_norm": 2.2879457473754883,
+      "learning_rate": 4.271604938271605e-05,
+      "loss": 0.8427,
+      "step": 312
+    },
+    {
+      "epoch": 0.6943982251802552,
+      "grad_norm": 2.4502720832824707,
+      "learning_rate": 4.267489711934157e-05,
+      "loss": 0.7854,
+      "step": 313
+    },
+    {
+      "epoch": 0.6966167498613423,
+      "grad_norm": 2.4553287029266357,
+      "learning_rate": 4.2633744855967077e-05,
+      "loss": 0.8757,
+      "step": 314
+    },
+    {
+      "epoch": 0.6988352745424293,
+      "grad_norm": 2.112945556640625,
+      "learning_rate": 4.259259259259259e-05,
+      "loss": 0.8321,
+      "step": 315
+    },
+    {
+      "epoch": 0.7010537992235164,
+      "grad_norm": 2.499385118484497,
+      "learning_rate": 4.255144032921811e-05,
+      "loss": 0.7663,
+      "step": 316
+    },
+    {
+      "epoch": 0.7032723239046035,
+      "grad_norm": 2.126103401184082,
+      "learning_rate": 4.2510288065843624e-05,
+      "loss": 0.6955,
+      "step": 317
+    },
+    {
+      "epoch": 0.7054908485856906,
+      "grad_norm": 2.7470061779022217,
+      "learning_rate": 4.246913580246914e-05,
+      "loss": 0.64,
+      "step": 318
+    },
+    {
+      "epoch": 0.7077093732667776,
+      "grad_norm": 2.8587868213653564,
+      "learning_rate": 4.242798353909465e-05,
+      "loss": 0.9511,
+      "step": 319
+    },
+    {
+      "epoch": 0.7099278979478647,
+      "grad_norm": 2.686136245727539,
+      "learning_rate": 4.2386831275720165e-05,
+      "loss": 0.9591,
+      "step": 320
+    },
+    {
+      "epoch": 0.7121464226289518,
+      "grad_norm": 2.4061784744262695,
+      "learning_rate": 4.234567901234568e-05,
+      "loss": 0.7752,
+      "step": 321
+    },
+    {
+      "epoch": 0.7143649473100389,
+      "grad_norm": 2.7960126399993896,
+      "learning_rate": 4.2304526748971197e-05,
+      "loss": 0.9213,
+      "step": 322
+    },
+    {
+      "epoch": 0.7165834719911259,
+      "grad_norm": 2.4860634803771973,
+      "learning_rate": 4.2263374485596706e-05,
+      "loss": 0.8553,
+      "step": 323
+    },
+    {
+      "epoch": 0.718801996672213,
+      "grad_norm": 2.866135358810425,
+      "learning_rate": 4.222222222222222e-05,
+      "loss": 0.9254,
+      "step": 324
+    },
+    {
+      "epoch": 0.7210205213533001,
+      "grad_norm": 2.2684476375579834,
+      "learning_rate": 4.2181069958847744e-05,
+      "loss": 0.662,
+      "step": 325
+    },
+    {
+      "epoch": 0.7232390460343872,
+      "grad_norm": 2.2148725986480713,
+      "learning_rate": 4.213991769547325e-05,
+      "loss": 0.8685,
+      "step": 326
+    },
+    {
+      "epoch": 0.7254575707154742,
+      "grad_norm": 2.287087917327881,
+      "learning_rate": 4.209876543209877e-05,
+      "loss": 0.9582,
+      "step": 327
+    },
+    {
+      "epoch": 0.7276760953965613,
+      "grad_norm": 1.8944892883300781,
+      "learning_rate": 4.205761316872428e-05,
+      "loss": 0.8793,
+      "step": 328
+    },
+    {
+      "epoch": 0.7298946200776484,
+      "grad_norm": 2.4895966053009033,
+      "learning_rate": 4.2016460905349794e-05,
+      "loss": 0.8614,
+      "step": 329
+    },
+    {
+      "epoch": 0.7321131447587355,
+      "grad_norm": 2.600193977355957,
+      "learning_rate": 4.197530864197531e-05,
+      "loss": 0.8825,
+      "step": 330
+    },
+    {
+      "epoch": 0.7343316694398225,
+      "grad_norm": 2.590364456176758,
+      "learning_rate": 4.1934156378600826e-05,
+      "loss": 0.9323,
+      "step": 331
+    },
+    {
+      "epoch": 0.7365501941209096,
+      "grad_norm": 2.645531415939331,
+      "learning_rate": 4.189300411522634e-05,
+      "loss": 0.8635,
+      "step": 332
+    },
+    {
+      "epoch": 0.7387687188019967,
+      "grad_norm": 2.0355777740478516,
+      "learning_rate": 4.185185185185185e-05,
+      "loss": 0.6483,
+      "step": 333
+    },
+    {
+      "epoch": 0.7409872434830838,
+      "grad_norm": 3.033715009689331,
+      "learning_rate": 4.181069958847737e-05,
+      "loss": 0.8722,
+      "step": 334
+    },
+    {
+      "epoch": 0.7432057681641708,
+      "grad_norm": 2.768638849258423,
+      "learning_rate": 4.176954732510288e-05,
+      "loss": 0.8842,
+      "step": 335
+    },
+    {
+      "epoch": 0.7454242928452579,
+      "grad_norm": 2.4029829502105713,
+      "learning_rate": 4.17283950617284e-05,
+      "loss": 0.8903,
+      "step": 336
+    },
+    {
+      "epoch": 0.747642817526345,
+      "grad_norm": 2.8639230728149414,
+      "learning_rate": 4.168724279835391e-05,
+      "loss": 0.9127,
+      "step": 337
+    },
+    {
+      "epoch": 0.7498613422074321,
+      "grad_norm": 2.195549249649048,
+      "learning_rate": 4.164609053497942e-05,
+      "loss": 0.8743,
+      "step": 338
+    },
+    {
+      "epoch": 0.7520798668885191,
+      "grad_norm": 2.6430952548980713,
+      "learning_rate": 4.1604938271604946e-05,
+      "loss": 0.9415,
+      "step": 339
+    },
+    {
+      "epoch": 0.7542983915696062,
+      "grad_norm": 2.2044715881347656,
+      "learning_rate": 4.1563786008230455e-05,
+      "loss": 0.9688,
+      "step": 340
+    },
+    {
+      "epoch": 0.7565169162506933,
+      "grad_norm": 3.5434839725494385,
+      "learning_rate": 4.152263374485597e-05,
+      "loss": 0.9032,
+      "step": 341
+    },
+    {
+      "epoch": 0.7587354409317804,
+      "grad_norm": 2.675981283187866,
+      "learning_rate": 4.148148148148148e-05,
+      "loss": 0.6579,
+      "step": 342
+    },
+    {
+      "epoch": 0.7609539656128674,
+      "grad_norm": 2.359999895095825,
+      "learning_rate": 4.1440329218106996e-05,
+      "loss": 0.6791,
+      "step": 343
+    },
+    {
+      "epoch": 0.7631724902939545,
+      "grad_norm": 3.3212859630584717,
+      "learning_rate": 4.139917695473251e-05,
+      "loss": 0.8944,
+      "step": 344
+    },
+    {
+      "epoch": 0.7653910149750416,
+      "grad_norm": 3.132432460784912,
+      "learning_rate": 4.135802469135803e-05,
+      "loss": 0.9972,
+      "step": 345
+    },
+    {
+      "epoch": 0.7676095396561287,
+      "grad_norm": 2.307504177093506,
+      "learning_rate": 4.1316872427983537e-05,
+      "loss": 0.8389,
+      "step": 346
+    },
+    {
+      "epoch": 0.7698280643372157,
+      "grad_norm": 2.42441987991333,
+      "learning_rate": 4.127572016460905e-05,
+      "loss": 0.6305,
+      "step": 347
+    },
+    {
+      "epoch": 0.7720465890183028,
+      "grad_norm": 2.671686887741089,
+      "learning_rate": 4.1234567901234575e-05,
+      "loss": 0.814,
+      "step": 348
+    },
+    {
+      "epoch": 0.7742651136993899,
+      "grad_norm": 2.4801604747772217,
+      "learning_rate": 4.1193415637860084e-05,
+      "loss": 0.7194,
+      "step": 349
+    },
+    {
+      "epoch": 0.776483638380477,
+      "grad_norm": 3.1407759189605713,
+      "learning_rate": 4.11522633744856e-05,
+      "loss": 1.0266,
+      "step": 350
+    },
+    {
+      "epoch": 0.778702163061564,
+      "grad_norm": 2.1769816875457764,
+      "learning_rate": 4.111111111111111e-05,
+      "loss": 0.7053,
+      "step": 351
+    },
+    {
+      "epoch": 0.7809206877426511,
+      "grad_norm": 2.3974196910858154,
+      "learning_rate": 4.1069958847736625e-05,
+      "loss": 0.8707,
+      "step": 352
+    },
+    {
+      "epoch": 0.7831392124237382,
+      "grad_norm": 2.6001367568969727,
+      "learning_rate": 4.102880658436214e-05,
+      "loss": 0.8598,
+      "step": 353
+    },
+    {
+      "epoch": 0.7853577371048253,
+      "grad_norm": 2.423523426055908,
+      "learning_rate": 4.0987654320987657e-05,
+      "loss": 0.8859,
+      "step": 354
+    },
+    {
+      "epoch": 0.7875762617859123,
+      "grad_norm": 2.952566146850586,
+      "learning_rate": 4.094650205761317e-05,
+      "loss": 1.0418,
+      "step": 355
+    },
+    {
+      "epoch": 0.7897947864669994,
+      "grad_norm": 2.0914344787597656,
+      "learning_rate": 4.090534979423868e-05,
+      "loss": 0.9717,
+      "step": 356
+    },
+    {
+      "epoch": 0.7920133111480865,
+      "grad_norm": 2.5211215019226074,
+      "learning_rate": 4.0864197530864204e-05,
+      "loss": 0.8734,
+      "step": 357
+    },
+    {
+      "epoch": 0.7942318358291736,
+      "grad_norm": 2.3540291786193848,
+      "learning_rate": 4.082304526748971e-05,
+      "loss": 0.8394,
+      "step": 358
+    },
+    {
+      "epoch": 0.7964503605102606,
+      "grad_norm": 2.257180690765381,
+      "learning_rate": 4.078189300411523e-05,
+      "loss": 0.7865,
+      "step": 359
+    },
+    {
+      "epoch": 0.7986688851913477,
+      "grad_norm": 2.5502257347106934,
+      "learning_rate": 4.074074074074074e-05,
+      "loss": 0.9766,
+      "step": 360
+    },
+    {
+      "epoch": 0.8008874098724348,
+      "grad_norm": 2.5117642879486084,
+      "learning_rate": 4.0699588477366254e-05,
+      "loss": 0.8121,
+      "step": 361
+    },
+    {
+      "epoch": 0.8031059345535219,
+      "grad_norm": 2.523190975189209,
+      "learning_rate": 4.065843621399178e-05,
+      "loss": 0.8026,
+      "step": 362
+    },
+    {
+      "epoch": 0.8053244592346089,
+      "grad_norm": 2.1156351566314697,
+      "learning_rate": 4.0617283950617286e-05,
+      "loss": 0.8173,
+      "step": 363
+    },
+    {
+      "epoch": 0.807542983915696,
+      "grad_norm": 2.719038248062134,
+      "learning_rate": 4.05761316872428e-05,
+      "loss": 1.0945,
+      "step": 364
+    },
+    {
+      "epoch": 0.8097615085967831,
+      "grad_norm": 3.233534812927246,
+      "learning_rate": 4.053497942386831e-05,
+      "loss": 1.035,
+      "step": 365
+    },
+    {
+      "epoch": 0.8119800332778702,
+      "grad_norm": 2.4117770195007324,
+      "learning_rate": 4.049382716049383e-05,
+      "loss": 0.7489,
+      "step": 366
+    },
+    {
+      "epoch": 0.8141985579589573,
+      "grad_norm": 2.4701526165008545,
+      "learning_rate": 4.045267489711934e-05,
+      "loss": 1.0449,
+      "step": 367
+    },
+    {
+      "epoch": 0.8164170826400444,
+      "grad_norm": 2.4779155254364014,
+      "learning_rate": 4.041152263374486e-05,
+      "loss": 0.8412,
+      "step": 368
+    },
+    {
+      "epoch": 0.8186356073211315,
+      "grad_norm": 3.3620009422302246,
+      "learning_rate": 4.0370370370370374e-05,
+      "loss": 1.0082,
+      "step": 369
+    },
+    {
+      "epoch": 0.8208541320022186,
+      "grad_norm": 2.447540283203125,
+      "learning_rate": 4.032921810699588e-05,
+      "loss": 0.9453,
+      "step": 370
+    },
+    {
+      "epoch": 0.8230726566833056,
+      "grad_norm": 1.9913147687911987,
+      "learning_rate": 4.0288065843621406e-05,
+      "loss": 0.7515,
+      "step": 371
+    },
+    {
+      "epoch": 0.8252911813643927,
+      "grad_norm": 2.197465419769287,
+      "learning_rate": 4.0246913580246915e-05,
+      "loss": 0.8735,
+      "step": 372
+    },
+    {
+      "epoch": 0.8275097060454798,
+      "grad_norm": 2.7398712635040283,
+      "learning_rate": 4.020576131687243e-05,
+      "loss": 0.8689,
+      "step": 373
+    },
+    {
+      "epoch": 0.8297282307265669,
+      "grad_norm": 2.31486439704895,
+      "learning_rate": 4.016460905349794e-05,
+      "loss": 0.7564,
+      "step": 374
+    },
+    {
+      "epoch": 0.831946755407654,
+      "grad_norm": 2.095136880874634,
+      "learning_rate": 4.012345679012346e-05,
+      "loss": 0.7684,
+      "step": 375
+    },
+    {
+      "epoch": 0.834165280088741,
+      "grad_norm": 2.7112295627593994,
+      "learning_rate": 4.008230452674897e-05,
+      "loss": 0.9873,
+      "step": 376
+    },
+    {
+      "epoch": 0.8363838047698281,
+      "grad_norm": 2.6337904930114746,
+      "learning_rate": 4.004115226337449e-05,
+      "loss": 0.9256,
+      "step": 377
+    },
+    {
+      "epoch": 0.8386023294509152,
+      "grad_norm": 2.7133731842041016,
+      "learning_rate": 4e-05,
+      "loss": 1.1145,
+      "step": 378
+    },
+    {
+      "epoch": 0.8408208541320022,
+      "grad_norm": 2.5339035987854004,
+      "learning_rate": 3.995884773662551e-05,
+      "loss": 0.9136,
+      "step": 379
+    },
+    {
+      "epoch": 0.8430393788130893,
+      "grad_norm": 2.544438123703003,
+      "learning_rate": 3.9917695473251035e-05,
+      "loss": 0.8761,
+      "step": 380
+    },
+    {
+      "epoch": 0.8452579034941764,
+      "grad_norm": 3.183976650238037,
+      "learning_rate": 3.9876543209876544e-05,
+      "loss": 0.7823,
+      "step": 381
+    },
+    {
+      "epoch": 0.8474764281752635,
+      "grad_norm": 2.2459630966186523,
+      "learning_rate": 3.983539094650206e-05,
+      "loss": 0.8034,
+      "step": 382
+    },
+    {
+      "epoch": 0.8496949528563505,
+      "grad_norm": 2.5165297985076904,
+      "learning_rate": 3.979423868312757e-05,
+      "loss": 0.879,
+      "step": 383
+    },
+    {
+      "epoch": 0.8519134775374376,
+      "grad_norm": 2.6023669242858887,
+      "learning_rate": 3.975308641975309e-05,
+      "loss": 0.6989,
+      "step": 384
+    },
+    {
+      "epoch": 0.8541320022185247,
+      "grad_norm": 2.575261116027832,
+      "learning_rate": 3.971193415637861e-05,
+      "loss": 0.8245,
+      "step": 385
+    },
+    {
+      "epoch": 0.8563505268996118,
+      "grad_norm": 2.5048012733459473,
+      "learning_rate": 3.9670781893004117e-05,
+      "loss": 0.8186,
+      "step": 386
+    },
+    {
+      "epoch": 0.8585690515806988,
+      "grad_norm": 2.1843042373657227,
+      "learning_rate": 3.962962962962963e-05,
+      "loss": 1.0923,
+      "step": 387
+    },
+    {
+      "epoch": 0.8607875762617859,
+      "grad_norm": 2.620774745941162,
+      "learning_rate": 3.958847736625514e-05,
+      "loss": 0.8784,
+      "step": 388
+    },
+    {
+      "epoch": 0.863006100942873,
+      "grad_norm": 2.8896753787994385,
+      "learning_rate": 3.9547325102880664e-05,
+      "loss": 0.8651,
+      "step": 389
+    },
+    {
+      "epoch": 0.8652246256239601,
+      "grad_norm": 2.5130579471588135,
+      "learning_rate": 3.950617283950617e-05,
+      "loss": 0.8967,
+      "step": 390
+    },
+    {
+      "epoch": 0.8674431503050472,
+      "grad_norm": 2.700167417526245,
+      "learning_rate": 3.946502057613169e-05,
+      "loss": 0.934,
+      "step": 391
+    },
+    {
+      "epoch": 0.8696616749861342,
+      "grad_norm": 2.4684665203094482,
+      "learning_rate": 3.9423868312757205e-05,
+      "loss": 0.8215,
+      "step": 392
+    },
+    {
+      "epoch": 0.8718801996672213,
+      "grad_norm": 2.275214195251465,
+      "learning_rate": 3.938271604938272e-05,
+      "loss": 0.8292,
+      "step": 393
+    },
+    {
+      "epoch": 0.8740987243483084,
+      "grad_norm": 2.0604209899902344,
+      "learning_rate": 3.934156378600824e-05,
+      "loss": 0.8575,
+      "step": 394
+    },
+    {
+      "epoch": 0.8763172490293955,
+      "grad_norm": 2.17086124420166,
+      "learning_rate": 3.9300411522633746e-05,
+      "loss": 0.7844,
+      "step": 395
+    },
+    {
+      "epoch": 0.8785357737104825,
+      "grad_norm": 2.010873794555664,
+      "learning_rate": 3.925925925925926e-05,
+      "loss": 0.7772,
+      "step": 396
+    },
+    {
+      "epoch": 0.8807542983915696,
+      "grad_norm": 2.4863860607147217,
+      "learning_rate": 3.921810699588477e-05,
+      "loss": 0.8888,
+      "step": 397
+    },
+    {
+      "epoch": 0.8829728230726567,
+      "grad_norm": 2.776113748550415,
+      "learning_rate": 3.917695473251029e-05,
+      "loss": 0.9546,
+      "step": 398
+    },
+    {
+      "epoch": 0.8851913477537438,
+      "grad_norm": 2.6805546283721924,
+      "learning_rate": 3.91358024691358e-05,
+      "loss": 1.0661,
+      "step": 399
+    },
+    {
+      "epoch": 0.8874098724348308,
+      "grad_norm": 2.4641025066375732,
+      "learning_rate": 3.909465020576132e-05,
+      "loss": 0.9155,
+      "step": 400
+    },
+    {
+      "epoch": 0.8896283971159179,
+      "grad_norm": 2.5395474433898926,
+      "learning_rate": 3.9053497942386834e-05,
+      "loss": 0.7259,
+      "step": 401
+    },
+    {
+      "epoch": 0.891846921797005,
+      "grad_norm": 3.0238442420959473,
+      "learning_rate": 3.901234567901234e-05,
+      "loss": 0.8653,
+      "step": 402
+    },
+    {
+      "epoch": 0.894065446478092,
+      "grad_norm": 2.4377074241638184,
+      "learning_rate": 3.8971193415637866e-05,
+      "loss": 0.8289,
+      "step": 403
+    },
+    {
+      "epoch": 0.8962839711591791,
+      "grad_norm": 2.571044921875,
+      "learning_rate": 3.8930041152263375e-05,
+      "loss": 0.861,
+      "step": 404
+    },
+    {
+      "epoch": 0.8985024958402662,
+      "grad_norm": 2.09232497215271,
+      "learning_rate": 3.888888888888889e-05,
+      "loss": 0.7901,
+      "step": 405
+    },
+    {
+      "epoch": 0.9007210205213533,
+      "grad_norm": 2.145253896713257,
+      "learning_rate": 3.88477366255144e-05,
+      "loss": 0.5617,
+      "step": 406
+    },
+    {
+      "epoch": 0.9029395452024404,
+      "grad_norm": 2.66852068901062,
+      "learning_rate": 3.880658436213992e-05,
+      "loss": 0.855,
+      "step": 407
+    },
+    {
+      "epoch": 0.9051580698835274,
+      "grad_norm": 2.8901891708374023,
+      "learning_rate": 3.876543209876544e-05,
+      "loss": 1.114,
+      "step": 408
+    },
+    {
+      "epoch": 0.9073765945646145,
+      "grad_norm": 2.3946845531463623,
+      "learning_rate": 3.872427983539095e-05,
+      "loss": 0.9951,
+      "step": 409
+    },
+    {
+      "epoch": 0.9095951192457016,
+      "grad_norm": 2.5448341369628906,
+      "learning_rate": 3.868312757201646e-05,
+      "loss": 0.9367,
+      "step": 410
+    },
+    {
+      "epoch": 0.9118136439267887,
+      "grad_norm": 2.59200119972229,
+      "learning_rate": 3.864197530864197e-05,
+      "loss": 0.7482,
+      "step": 411
+    },
+    {
+      "epoch": 0.9140321686078757,
+      "grad_norm": 2.454349994659424,
+      "learning_rate": 3.8600823045267495e-05,
+      "loss": 0.8337,
+      "step": 412
+    },
+    {
+      "epoch": 0.9162506932889628,
+      "grad_norm": 2.481480836868286,
+      "learning_rate": 3.8559670781893004e-05,
+      "loss": 0.7191,
+      "step": 413
+    },
+    {
+      "epoch": 0.9184692179700499,
+      "grad_norm": 2.6003763675689697,
+      "learning_rate": 3.851851851851852e-05,
+      "loss": 0.8389,
+      "step": 414
+    },
+    {
+      "epoch": 0.920687742651137,
+      "grad_norm": 2.415048837661743,
+      "learning_rate": 3.8477366255144036e-05,
+      "loss": 0.7386,
+      "step": 415
+    },
+    {
+      "epoch": 0.922906267332224,
+      "grad_norm": 2.107497215270996,
+      "learning_rate": 3.843621399176955e-05,
+      "loss": 0.8014,
+      "step": 416
+    },
+    {
+      "epoch": 0.9251247920133111,
+      "grad_norm": 3.1628942489624023,
+      "learning_rate": 3.839506172839507e-05,
+      "loss": 0.8502,
+      "step": 417
+    },
+    {
+      "epoch": 0.9273433166943982,
+      "grad_norm": 2.4569737911224365,
+      "learning_rate": 3.8353909465020577e-05,
+      "loss": 0.9159,
+      "step": 418
+    },
+    {
+      "epoch": 0.9295618413754853,
+      "grad_norm": 2.7585675716400146,
+      "learning_rate": 3.831275720164609e-05,
+      "loss": 0.8727,
+      "step": 419
+    },
+    {
+      "epoch": 0.9317803660565723,
+      "grad_norm": 2.109832286834717,
+      "learning_rate": 3.82716049382716e-05,
+      "loss": 0.7377,
+      "step": 420
+    },
+    {
+      "epoch": 0.9339988907376594,
+      "grad_norm": 2.898301362991333,
+      "learning_rate": 3.8230452674897124e-05,
+      "loss": 0.8601,
+      "step": 421
+    },
+    {
+      "epoch": 0.9362174154187465,
+      "grad_norm": 2.061516284942627,
+      "learning_rate": 3.818930041152264e-05,
+      "loss": 0.9687,
+      "step": 422
+    },
+    {
+      "epoch": 0.9384359400998337,
+      "grad_norm": 2.579833984375,
+      "learning_rate": 3.814814814814815e-05,
+      "loss": 0.9992,
+      "step": 423
+    },
+    {
+      "epoch": 0.9406544647809207,
+      "grad_norm": 2.400799036026001,
+      "learning_rate": 3.8106995884773665e-05,
+      "loss": 0.9064,
+      "step": 424
+    },
+    {
+      "epoch": 0.9428729894620078,
+      "grad_norm": 2.4964163303375244,
+      "learning_rate": 3.806584362139918e-05,
+      "loss": 0.9076,
+      "step": 425
+    },
+    {
+      "epoch": 0.9450915141430949,
+      "grad_norm": 3.378230094909668,
+      "learning_rate": 3.80246913580247e-05,
+      "loss": 0.968,
+      "step": 426
+    },
+    {
+      "epoch": 0.947310038824182,
+      "grad_norm": 2.4372470378875732,
+      "learning_rate": 3.7983539094650206e-05,
+      "loss": 0.8334,
+      "step": 427
+    },
+    {
+      "epoch": 0.949528563505269,
+      "grad_norm": 2.289325714111328,
+      "learning_rate": 3.794238683127572e-05,
+      "loss": 0.9324,
+      "step": 428
+    },
+    {
+      "epoch": 0.9517470881863561,
+      "grad_norm": 2.3763060569763184,
+      "learning_rate": 3.790123456790123e-05,
+      "loss": 0.8749,
+      "step": 429
+    },
+    {
+      "epoch": 0.9539656128674432,
+      "grad_norm": 2.20355486869812,
+      "learning_rate": 3.786008230452675e-05,
+      "loss": 0.6213,
+      "step": 430
+    },
+    {
+      "epoch": 0.9561841375485303,
+      "grad_norm": 2.462961435317993,
+      "learning_rate": 3.781893004115227e-05,
+      "loss": 0.8655,
+      "step": 431
+    },
+    {
+      "epoch": 0.9584026622296173,
+      "grad_norm": 2.3069040775299072,
+      "learning_rate": 3.777777777777778e-05,
+      "loss": 0.8946,
+      "step": 432
+    },
+    {
+      "epoch": 0.9606211869107044,
+      "grad_norm": 2.9334943294525146,
+      "learning_rate": 3.7736625514403294e-05,
+      "loss": 0.9654,
+      "step": 433
+    },
+    {
+      "epoch": 0.9628397115917915,
+      "grad_norm": 2.500938653945923,
+      "learning_rate": 3.769547325102881e-05,
+      "loss": 1.0374,
+      "step": 434
+    },
+    {
+      "epoch": 0.9650582362728786,
+      "grad_norm": 2.5743601322174072,
+      "learning_rate": 3.7654320987654326e-05,
+      "loss": 0.8145,
+      "step": 435
+    },
+    {
+      "epoch": 0.9672767609539656,
+      "grad_norm": 2.1049318313598633,
+      "learning_rate": 3.7613168724279835e-05,
+      "loss": 0.8196,
+      "step": 436
+    },
+    {
+      "epoch": 0.9694952856350527,
+      "grad_norm": 2.144895315170288,
+      "learning_rate": 3.757201646090535e-05,
+      "loss": 0.7571,
+      "step": 437
+    },
+    {
+      "epoch": 0.9717138103161398,
+      "grad_norm": 2.041776657104492,
+      "learning_rate": 3.7530864197530867e-05,
+      "loss": 0.7129,
+      "step": 438
+    },
+    {
+      "epoch": 0.9739323349972269,
+      "grad_norm": 1.9652653932571411,
+      "learning_rate": 3.748971193415638e-05,
+      "loss": 0.8207,
+      "step": 439
+    },
+    {
+      "epoch": 0.9761508596783139,
+      "grad_norm": 2.853630304336548,
+      "learning_rate": 3.74485596707819e-05,
+      "loss": 1.0877,
+      "step": 440
+    },
+    {
+      "epoch": 0.978369384359401,
+      "grad_norm": 2.438640594482422,
+      "learning_rate": 3.740740740740741e-05,
+      "loss": 0.8602,
+      "step": 441
+    },
+    {
+      "epoch": 0.9805879090404881,
+      "grad_norm": 2.426461696624756,
+      "learning_rate": 3.736625514403292e-05,
+      "loss": 0.7444,
+      "step": 442
+    },
+    {
+      "epoch": 0.9828064337215752,
+      "grad_norm": 2.5666048526763916,
+      "learning_rate": 3.732510288065844e-05,
+      "loss": 1.0394,
+      "step": 443
+    },
+    {
+      "epoch": 0.9850249584026622,
+      "grad_norm": 2.504512071609497,
+      "learning_rate": 3.7283950617283955e-05,
+      "loss": 0.7838,
+      "step": 444
+    },
+    {
+      "epoch": 0.9872434830837493,
+      "grad_norm": 2.106992244720459,
+      "learning_rate": 3.724279835390947e-05,
+      "loss": 0.7671,
+      "step": 445
+    },
+    {
+      "epoch": 0.9894620077648364,
+      "grad_norm": 2.7785143852233887,
+      "learning_rate": 3.720164609053498e-05,
+      "loss": 0.9332,
+      "step": 446
+    },
+    {
+      "epoch": 0.9916805324459235,
+      "grad_norm": 2.3925273418426514,
+      "learning_rate": 3.7160493827160496e-05,
+      "loss": 0.6741,
+      "step": 447
+    },
+    {
+      "epoch": 0.9938990571270105,
+      "grad_norm": 2.2031757831573486,
+      "learning_rate": 3.711934156378601e-05,
+      "loss": 0.8299,
+      "step": 448
+    },
+    {
+      "epoch": 0.9961175818080976,
+      "grad_norm": 2.8807144165039062,
+      "learning_rate": 3.707818930041153e-05,
+      "loss": 0.9496,
+      "step": 449
+    },
+    {
+      "epoch": 0.9983361064891847,
+      "grad_norm": 2.981637716293335,
+      "learning_rate": 3.7037037037037037e-05,
+      "loss": 0.8863,
+      "step": 450
+    },
+    {
+      "epoch": 1.0005546311702718,
+      "grad_norm": 2.1683170795440674,
+      "learning_rate": 3.699588477366255e-05,
+      "loss": 0.9213,
+      "step": 451
+    },
+    {
+      "epoch": 1.0027731558513588,
+      "grad_norm": 2.215830087661743,
+      "learning_rate": 3.695473251028807e-05,
+      "loss": 0.6341,
+      "step": 452
+    },
+    {
+      "epoch": 1.004991680532446,
+      "grad_norm": 1.91196870803833,
+      "learning_rate": 3.6913580246913584e-05,
+      "loss": 0.5355,
+      "step": 453
+    },
+    {
+      "epoch": 1.007210205213533,
+      "grad_norm": 1.5960261821746826,
+      "learning_rate": 3.68724279835391e-05,
+      "loss": 0.5113,
+      "step": 454
+    },
+    {
+      "epoch": 1.00942872989462,
+      "grad_norm": 2.094749927520752,
+      "learning_rate": 3.683127572016461e-05,
+      "loss": 0.5023,
+      "step": 455
+    },
+    {
+      "epoch": 1.0116472545757071,
+      "grad_norm": 2.0780062675476074,
+      "learning_rate": 3.6790123456790125e-05,
+      "loss": 0.6194,
+      "step": 456
+    },
+    {
+      "epoch": 1.0138657792567942,
+      "grad_norm": 2.324847936630249,
+      "learning_rate": 3.674897119341564e-05,
+      "loss": 0.4705,
+      "step": 457
+    },
+    {
+      "epoch": 1.0160843039378813,
+      "grad_norm": 2.164628267288208,
+      "learning_rate": 3.670781893004116e-05,
+      "loss": 0.488,
+      "step": 458
+    },
+    {
+      "epoch": 1.0183028286189684,
+      "grad_norm": 2.1279964447021484,
+      "learning_rate": 3.6666666666666666e-05,
+      "loss": 0.4564,
+      "step": 459
+    },
+    {
+      "epoch": 1.0205213533000554,
+      "grad_norm": 2.1615474224090576,
+      "learning_rate": 3.662551440329218e-05,
+      "loss": 0.3699,
+      "step": 460
+    },
+    {
+      "epoch": 1.0227398779811425,
+      "grad_norm": 2.3255395889282227,
+      "learning_rate": 3.65843621399177e-05,
+      "loss": 0.5279,
+      "step": 461
+    },
+    {
+      "epoch": 1.0249584026622296,
+      "grad_norm": 2.4930872917175293,
+      "learning_rate": 3.654320987654321e-05,
+      "loss": 0.5708,
+      "step": 462
+    },
+    {
+      "epoch": 1.0271769273433167,
+      "grad_norm": 2.4449357986450195,
+      "learning_rate": 3.650205761316873e-05,
+      "loss": 0.4707,
+      "step": 463
+    },
+    {
+      "epoch": 1.0293954520244037,
+      "grad_norm": 2.385634422302246,
+      "learning_rate": 3.646090534979424e-05,
+      "loss": 0.4898,
+      "step": 464
+    },
+    {
+      "epoch": 1.0316139767054908,
+      "grad_norm": 2.704658031463623,
+      "learning_rate": 3.6419753086419754e-05,
+      "loss": 0.5939,
+      "step": 465
+    },
+    {
+      "epoch": 1.033832501386578,
+      "grad_norm": 2.2051949501037598,
+      "learning_rate": 3.637860082304527e-05,
+      "loss": 0.5444,
+      "step": 466
+    },
+    {
+      "epoch": 1.036051026067665,
+      "grad_norm": 2.695496082305908,
+      "learning_rate": 3.6337448559670786e-05,
+      "loss": 0.5788,
+      "step": 467
+    },
+    {
+      "epoch": 1.038269550748752,
+      "grad_norm": 3.695551633834839,
+      "learning_rate": 3.62962962962963e-05,
+      "loss": 0.6287,
+      "step": 468
+    },
+    {
+      "epoch": 1.0404880754298391,
+      "grad_norm": 1.9971530437469482,
+      "learning_rate": 3.625514403292181e-05,
+      "loss": 0.5131,
+      "step": 469
+    },
+    {
+      "epoch": 1.0427066001109262,
+      "grad_norm": 2.118812084197998,
+      "learning_rate": 3.6213991769547327e-05,
+      "loss": 0.5014,
+      "step": 470
+    },
+    {
+      "epoch": 1.0449251247920133,
+      "grad_norm": 2.512852191925049,
+      "learning_rate": 3.617283950617284e-05,
+      "loss": 0.548,
+      "step": 471
+    },
+    {
+      "epoch": 1.0471436494731003,
+      "grad_norm": 2.6009161472320557,
+      "learning_rate": 3.613168724279836e-05,
+      "loss": 0.7103,
+      "step": 472
+    },
+    {
+      "epoch": 1.0493621741541874,
+      "grad_norm": 2.31130051612854,
+      "learning_rate": 3.609053497942387e-05,
+      "loss": 0.5628,
+      "step": 473
+    },
+    {
+      "epoch": 1.0515806988352745,
+      "grad_norm": 2.1873180866241455,
+      "learning_rate": 3.604938271604938e-05,
+      "loss": 0.62,
+      "step": 474
+    },
+    {
+      "epoch": 1.0537992235163616,
+      "grad_norm": 2.3002912998199463,
+      "learning_rate": 3.60082304526749e-05,
+      "loss": 0.4966,
+      "step": 475
+    },
+    {
+      "epoch": 1.0560177481974486,
+      "grad_norm": 2.13053560256958,
+      "learning_rate": 3.5967078189300415e-05,
+      "loss": 0.525,
+      "step": 476
+    },
+    {
+      "epoch": 1.0582362728785357,
+      "grad_norm": 1.9421828985214233,
+      "learning_rate": 3.592592592592593e-05,
+      "loss": 0.3268,
+      "step": 477
+    },
+    {
+      "epoch": 1.0604547975596228,
+      "grad_norm": 2.435682773590088,
+      "learning_rate": 3.588477366255144e-05,
+      "loss": 0.4733,
+      "step": 478
+    },
+    {
+      "epoch": 1.0626733222407099,
+      "grad_norm": 2.0943732261657715,
+      "learning_rate": 3.5843621399176956e-05,
+      "loss": 0.485,
+      "step": 479
+    },
+    {
+      "epoch": 1.064891846921797,
+      "grad_norm": 2.999981641769409,
+      "learning_rate": 3.580246913580247e-05,
+      "loss": 0.6419,
+      "step": 480
+    },
+    {
+      "epoch": 1.067110371602884,
+      "grad_norm": 2.2813262939453125,
+      "learning_rate": 3.576131687242799e-05,
+      "loss": 0.4722,
+      "step": 481
+    },
+    {
+      "epoch": 1.069328896283971,
+      "grad_norm": 2.4672632217407227,
+      "learning_rate": 3.5720164609053497e-05,
+      "loss": 0.6396,
+      "step": 482
+    },
+    {
+      "epoch": 1.0715474209650582,
+      "grad_norm": 2.2799105644226074,
+      "learning_rate": 3.567901234567901e-05,
+      "loss": 0.6887,
+      "step": 483
+    },
+    {
+      "epoch": 1.0737659456461452,
+      "grad_norm": 2.7640743255615234,
+      "learning_rate": 3.563786008230453e-05,
+      "loss": 0.5466,
+      "step": 484
+    },
+    {
+      "epoch": 1.0759844703272323,
+      "grad_norm": 2.2956655025482178,
+      "learning_rate": 3.5596707818930044e-05,
+      "loss": 0.4739,
+      "step": 485
+    },
+    {
+      "epoch": 1.0782029950083194,
+      "grad_norm": 1.8849239349365234,
+      "learning_rate": 3.555555555555556e-05,
+      "loss": 0.5482,
+      "step": 486
+    },
+    {
+      "epoch": 1.0804215196894065,
+      "grad_norm": 2.890810966491699,
+      "learning_rate": 3.551440329218107e-05,
+      "loss": 0.4971,
+      "step": 487
+    },
+    {
+      "epoch": 1.0826400443704935,
+      "grad_norm": 2.282778263092041,
+      "learning_rate": 3.5473251028806585e-05,
+      "loss": 0.5902,
+      "step": 488
+    },
+    {
+      "epoch": 1.0848585690515806,
+      "grad_norm": 2.5978939533233643,
+      "learning_rate": 3.54320987654321e-05,
+      "loss": 0.7649,
+      "step": 489
+    },
+    {
+      "epoch": 1.0870770937326677,
+      "grad_norm": 2.3558173179626465,
+      "learning_rate": 3.539094650205762e-05,
+      "loss": 0.6972,
+      "step": 490
+    },
+    {
+      "epoch": 1.0892956184137548,
+      "grad_norm": 2.738722324371338,
+      "learning_rate": 3.534979423868313e-05,
+      "loss": 0.6411,
+      "step": 491
+    },
+    {
+      "epoch": 1.0915141430948418,
+      "grad_norm": 1.8174808025360107,
+      "learning_rate": 3.530864197530864e-05,
+      "loss": 0.5553,
+      "step": 492
+    },
+    {
+      "epoch": 1.093732667775929,
+      "grad_norm": 2.407663106918335,
+      "learning_rate": 3.526748971193416e-05,
+      "loss": 0.6264,
+      "step": 493
+    },
+    {
+      "epoch": 1.095951192457016,
+      "grad_norm": 2.8249285221099854,
+      "learning_rate": 3.522633744855967e-05,
+      "loss": 0.5956,
+      "step": 494
+    },
+    {
+      "epoch": 1.098169717138103,
+      "grad_norm": 2.1922669410705566,
+      "learning_rate": 3.518518518518519e-05,
+      "loss": 0.4324,
+      "step": 495
+    },
+    {
+      "epoch": 1.1003882418191901,
+      "grad_norm": 2.0410611629486084,
+      "learning_rate": 3.51440329218107e-05,
+      "loss": 0.6668,
+      "step": 496
+    },
+    {
+      "epoch": 1.1026067665002772,
+      "grad_norm": 2.4528214931488037,
+      "learning_rate": 3.5102880658436214e-05,
+      "loss": 0.6589,
+      "step": 497
+    },
+    {
+      "epoch": 1.1048252911813643,
+      "grad_norm": 2.1697845458984375,
+      "learning_rate": 3.506172839506173e-05,
+      "loss": 0.5475,
+      "step": 498
+    },
+    {
+      "epoch": 1.1070438158624514,
+      "grad_norm": 2.2146642208099365,
+      "learning_rate": 3.5020576131687246e-05,
+      "loss": 0.5463,
+      "step": 499
+    },
+    {
+      "epoch": 1.1092623405435384,
+      "grad_norm": 2.1385345458984375,
+      "learning_rate": 3.497942386831276e-05,
+      "loss": 0.5204,
+      "step": 500
+    },
+    {
+      "epoch": 1.1114808652246255,
+      "grad_norm": 2.4533028602600098,
+      "learning_rate": 3.493827160493827e-05,
+      "loss": 0.4911,
+      "step": 501
+    },
+    {
+      "epoch": 1.1136993899057126,
+      "grad_norm": 2.3409910202026367,
+      "learning_rate": 3.4897119341563787e-05,
+      "loss": 0.5352,
+      "step": 502
+    },
+    {
+      "epoch": 1.1159179145867997,
+      "grad_norm": 1.9345369338989258,
+      "learning_rate": 3.48559670781893e-05,
+      "loss": 0.6193,
+      "step": 503
+    },
+    {
+      "epoch": 1.1181364392678868,
+      "grad_norm": 2.357593059539795,
+      "learning_rate": 3.481481481481482e-05,
+      "loss": 0.5113,
+      "step": 504
+    },
+    {
+      "epoch": 1.1203549639489738,
+      "grad_norm": 2.2779459953308105,
+      "learning_rate": 3.4773662551440334e-05,
+      "loss": 0.568,
+      "step": 505
+    },
+    {
+      "epoch": 1.122573488630061,
+      "grad_norm": 1.9965463876724243,
+      "learning_rate": 3.473251028806584e-05,
+      "loss": 0.5249,
+      "step": 506
+    },
+    {
+      "epoch": 1.124792013311148,
+      "grad_norm": 2.304286241531372,
+      "learning_rate": 3.469135802469136e-05,
+      "loss": 0.5414,
+      "step": 507
+    },
+    {
+      "epoch": 1.127010537992235,
+      "grad_norm": 2.1614229679107666,
+      "learning_rate": 3.4650205761316875e-05,
+      "loss": 0.4441,
+      "step": 508
+    },
+    {
+      "epoch": 1.1292290626733221,
+      "grad_norm": 1.9792081117630005,
+      "learning_rate": 3.460905349794239e-05,
+      "loss": 0.4752,
+      "step": 509
+    },
+    {
+      "epoch": 1.1314475873544092,
+      "grad_norm": 2.12204909324646,
+      "learning_rate": 3.45679012345679e-05,
+      "loss": 0.5364,
+      "step": 510
+    },
+    {
+      "epoch": 1.1336661120354963,
+      "grad_norm": 2.640716075897217,
+      "learning_rate": 3.4526748971193416e-05,
+      "loss": 0.6408,
+      "step": 511
+    },
+    {
+      "epoch": 1.1358846367165834,
+      "grad_norm": 2.9724271297454834,
+      "learning_rate": 3.448559670781893e-05,
+      "loss": 0.7697,
+      "step": 512
+    },
+    {
+      "epoch": 1.1381031613976704,
+      "grad_norm": 2.3501429557800293,
+      "learning_rate": 3.444444444444445e-05,
+      "loss": 0.6144,
+      "step": 513
+    },
+    {
+      "epoch": 1.1403216860787575,
+      "grad_norm": 2.7037365436553955,
+      "learning_rate": 3.440329218106996e-05,
+      "loss": 0.4197,
+      "step": 514
+    },
+    {
+      "epoch": 1.1425402107598448,
+      "grad_norm": 2.9580609798431396,
+      "learning_rate": 3.436213991769547e-05,
+      "loss": 0.7042,
+      "step": 515
+    },
+    {
+      "epoch": 1.1447587354409319,
+      "grad_norm": 2.224079132080078,
+      "learning_rate": 3.432098765432099e-05,
+      "loss": 0.6375,
+      "step": 516
+    },
+    {
+      "epoch": 1.146977260122019,
+      "grad_norm": 2.622692823410034,
+      "learning_rate": 3.4279835390946504e-05,
+      "loss": 0.6024,
+      "step": 517
+    },
+    {
+      "epoch": 1.149195784803106,
+      "grad_norm": 2.5141777992248535,
+      "learning_rate": 3.423868312757202e-05,
+      "loss": 0.5751,
+      "step": 518
+    },
+    {
+      "epoch": 1.151414309484193,
+      "grad_norm": 2.2731761932373047,
+      "learning_rate": 3.419753086419753e-05,
+      "loss": 0.629,
+      "step": 519
+    },
+    {
+      "epoch": 1.1536328341652802,
+      "grad_norm": 2.190129518508911,
+      "learning_rate": 3.4156378600823045e-05,
+      "loss": 0.4999,
+      "step": 520
+    },
+    {
+      "epoch": 1.1558513588463672,
+      "grad_norm": 1.7967180013656616,
+      "learning_rate": 3.411522633744856e-05,
+      "loss": 0.596,
+      "step": 521
+    },
+    {
+      "epoch": 1.1580698835274543,
+      "grad_norm": 2.0199036598205566,
+      "learning_rate": 3.4074074074074077e-05,
+      "loss": 0.5649,
+      "step": 522
+    },
+    {
+      "epoch": 1.1602884082085414,
+      "grad_norm": 2.3646140098571777,
+      "learning_rate": 3.403292181069959e-05,
+      "loss": 0.509,
+      "step": 523
+    },
+    {
+      "epoch": 1.1625069328896285,
+      "grad_norm": 1.9126660823822021,
+      "learning_rate": 3.39917695473251e-05,
+      "loss": 0.4659,
+      "step": 524
+    },
+    {
+      "epoch": 1.1647254575707155,
+      "grad_norm": 2.4993457794189453,
+      "learning_rate": 3.395061728395062e-05,
+      "loss": 0.6414,
+      "step": 525
+    },
+    {
+      "epoch": 1.1669439822518026,
+      "grad_norm": 2.0620944499969482,
+      "learning_rate": 3.390946502057613e-05,
+      "loss": 0.5715,
+      "step": 526
+    },
+    {
+      "epoch": 1.1691625069328897,
+      "grad_norm": 1.9987257719039917,
+      "learning_rate": 3.386831275720165e-05,
+      "loss": 0.3991,
+      "step": 527
+    },
+    {
+      "epoch": 1.1713810316139768,
+      "grad_norm": 2.2777981758117676,
+      "learning_rate": 3.3827160493827165e-05,
+      "loss": 0.5506,
+      "step": 528
+    },
+    {
+      "epoch": 1.1735995562950639,
+      "grad_norm": 2.412094831466675,
+      "learning_rate": 3.3786008230452674e-05,
+      "loss": 0.5987,
+      "step": 529
+    },
+    {
+      "epoch": 1.175818080976151,
+      "grad_norm": 2.3012962341308594,
+      "learning_rate": 3.374485596707819e-05,
+      "loss": 0.5073,
+      "step": 530
+    },
+    {
+      "epoch": 1.178036605657238,
+      "grad_norm": 2.798724889755249,
+      "learning_rate": 3.3703703703703706e-05,
+      "loss": 0.5089,
+      "step": 531
+    },
+    {
+      "epoch": 1.180255130338325,
+      "grad_norm": 2.2754406929016113,
+      "learning_rate": 3.366255144032922e-05,
+      "loss": 0.5536,
+      "step": 532
+    },
+    {
+      "epoch": 1.1824736550194122,
+      "grad_norm": 2.744678020477295,
+      "learning_rate": 3.362139917695473e-05,
+      "loss": 0.5167,
+      "step": 533
+    },
+    {
+      "epoch": 1.1846921797004992,
+      "grad_norm": 1.9823955297470093,
+      "learning_rate": 3.3580246913580247e-05,
+      "loss": 0.4435,
+      "step": 534
+    },
+    {
+      "epoch": 1.1869107043815863,
+      "grad_norm": 2.0235283374786377,
+      "learning_rate": 3.353909465020576e-05,
+      "loss": 0.5414,
+      "step": 535
+    },
+    {
+      "epoch": 1.1891292290626734,
+      "grad_norm": 3.27765154838562,
+      "learning_rate": 3.349794238683128e-05,
+      "loss": 0.4984,
+      "step": 536
+    },
+    {
+      "epoch": 1.1913477537437605,
+      "grad_norm": 2.383869171142578,
+      "learning_rate": 3.3456790123456794e-05,
+      "loss": 0.4967,
+      "step": 537
+    },
+    {
+      "epoch": 1.1935662784248475,
+      "grad_norm": 2.182147741317749,
+      "learning_rate": 3.34156378600823e-05,
+      "loss": 0.622,
+      "step": 538
+    },
+    {
+      "epoch": 1.1957848031059346,
+      "grad_norm": 1.9554792642593384,
+      "learning_rate": 3.337448559670782e-05,
+      "loss": 0.471,
+      "step": 539
+    },
+    {
+      "epoch": 1.1980033277870217,
+      "grad_norm": 2.321263313293457,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.5341,
+      "step": 540
+    },
+    {
+      "epoch": 1.2002218524681088,
+      "grad_norm": 1.936584234237671,
+      "learning_rate": 3.329218106995885e-05,
+      "loss": 0.497,
+      "step": 541
+    },
+    {
+      "epoch": 1.2024403771491958,
+      "grad_norm": 2.515679121017456,
+      "learning_rate": 3.325102880658436e-05,
+      "loss": 0.5462,
+      "step": 542
+    },
+    {
+      "epoch": 1.204658901830283,
+      "grad_norm": 2.189818859100342,
+      "learning_rate": 3.3209876543209876e-05,
+      "loss": 0.5073,
+      "step": 543
+    },
+    {
+      "epoch": 1.20687742651137,
+      "grad_norm": 1.9511988162994385,
+      "learning_rate": 3.316872427983539e-05,
+      "loss": 0.4862,
+      "step": 544
+    },
+    {
+      "epoch": 1.209095951192457,
+      "grad_norm": 2.0378596782684326,
+      "learning_rate": 3.312757201646091e-05,
+      "loss": 0.4735,
+      "step": 545
+    },
+    {
+      "epoch": 1.2113144758735441,
+      "grad_norm": 2.080709218978882,
+      "learning_rate": 3.308641975308642e-05,
+      "loss": 0.43,
+      "step": 546
+    },
+    {
+      "epoch": 1.2135330005546312,
+      "grad_norm": 2.454526901245117,
+      "learning_rate": 3.304526748971193e-05,
+      "loss": 0.574,
+      "step": 547
+    },
+    {
+      "epoch": 1.2157515252357183,
+      "grad_norm": 2.9881608486175537,
+      "learning_rate": 3.300411522633745e-05,
+      "loss": 0.4997,
+      "step": 548
+    },
+    {
+      "epoch": 1.2179700499168054,
+      "grad_norm": 2.3299403190612793,
+      "learning_rate": 3.2962962962962964e-05,
+      "loss": 0.6229,
+      "step": 549
+    },
+    {
+      "epoch": 1.2201885745978924,
+      "grad_norm": 2.5587589740753174,
+      "learning_rate": 3.292181069958848e-05,
+      "loss": 0.6606,
+      "step": 550
+    },
+    {
+      "epoch": 1.2224070992789795,
+      "grad_norm": 2.4784085750579834,
+      "learning_rate": 3.2880658436213996e-05,
+      "loss": 0.5553,
+      "step": 551
+    },
+    {
+      "epoch": 1.2246256239600666,
+      "grad_norm": 3.380730390548706,
+      "learning_rate": 3.2839506172839505e-05,
+      "loss": 0.5832,
+      "step": 552
+    },
+    {
+      "epoch": 1.2268441486411537,
+      "grad_norm": 2.1143057346343994,
+      "learning_rate": 3.279835390946502e-05,
+      "loss": 0.5168,
+      "step": 553
+    },
+    {
+      "epoch": 1.2290626733222407,
+      "grad_norm": 2.1933341026306152,
+      "learning_rate": 3.2757201646090537e-05,
+      "loss": 0.6098,
+      "step": 554
+    },
+    {
+      "epoch": 1.2312811980033278,
+      "grad_norm": 2.4897682666778564,
+      "learning_rate": 3.271604938271605e-05,
+      "loss": 0.6705,
+      "step": 555
+    },
+    {
+      "epoch": 1.2334997226844149,
+      "grad_norm": 1.8998596668243408,
+      "learning_rate": 3.267489711934156e-05,
+      "loss": 0.6223,
+      "step": 556
+    },
+    {
+      "epoch": 1.235718247365502,
+      "grad_norm": 2.208927631378174,
+      "learning_rate": 3.263374485596708e-05,
+      "loss": 0.5151,
+      "step": 557
+    },
+    {
+      "epoch": 1.237936772046589,
+      "grad_norm": 2.3265204429626465,
+      "learning_rate": 3.25925925925926e-05,
+      "loss": 0.4392,
+      "step": 558
+    },
+    {
+      "epoch": 1.240155296727676,
+      "grad_norm": 1.9276748895645142,
+      "learning_rate": 3.255144032921811e-05,
+      "loss": 0.5483,
+      "step": 559
+    },
+    {
+      "epoch": 1.2423738214087632,
+      "grad_norm": 2.193643808364868,
+      "learning_rate": 3.2510288065843625e-05,
+      "loss": 0.5374,
+      "step": 560
+    },
+    {
+      "epoch": 1.2445923460898503,
+      "grad_norm": 2.1251633167266846,
+      "learning_rate": 3.2469135802469134e-05,
+      "loss": 0.4523,
+      "step": 561
+    },
+    {
+      "epoch": 1.2468108707709373,
+      "grad_norm": 2.3506054878234863,
+      "learning_rate": 3.242798353909465e-05,
+      "loss": 0.531,
+      "step": 562
+    },
+    {
+      "epoch": 1.2490293954520244,
+      "grad_norm": 2.507739782333374,
+      "learning_rate": 3.2386831275720166e-05,
+      "loss": 0.6662,
+      "step": 563
+    },
+    {
+      "epoch": 1.2512479201331115,
+      "grad_norm": 2.6904561519622803,
+      "learning_rate": 3.234567901234568e-05,
+      "loss": 0.5682,
+      "step": 564
+    },
+    {
+      "epoch": 1.2534664448141986,
+      "grad_norm": 2.2025396823883057,
+      "learning_rate": 3.230452674897119e-05,
+      "loss": 0.5238,
+      "step": 565
+    },
+    {
+      "epoch": 1.2556849694952856,
+      "grad_norm": 2.7115561962127686,
+      "learning_rate": 3.2263374485596707e-05,
+      "loss": 0.6556,
+      "step": 566
+    },
+    {
+      "epoch": 1.2579034941763727,
+      "grad_norm": 2.2397167682647705,
+      "learning_rate": 3.222222222222223e-05,
+      "loss": 0.6283,
+      "step": 567
+    },
+    {
+      "epoch": 1.2601220188574598,
+      "grad_norm": 2.273963689804077,
+      "learning_rate": 3.218106995884774e-05,
+      "loss": 0.6581,
+      "step": 568
+    },
+    {
+      "epoch": 1.2623405435385469,
+      "grad_norm": 1.8225631713867188,
+      "learning_rate": 3.2139917695473254e-05,
+      "loss": 0.5347,
+      "step": 569
+    },
+    {
+      "epoch": 1.264559068219634,
+      "grad_norm": 1.7965829372406006,
+      "learning_rate": 3.209876543209876e-05,
+      "loss": 0.4678,
+      "step": 570
+    },
+    {
+      "epoch": 1.266777592900721,
+      "grad_norm": 2.197831869125366,
+      "learning_rate": 3.205761316872428e-05,
+      "loss": 0.5374,
+      "step": 571
+    },
+    {
+      "epoch": 1.268996117581808,
+      "grad_norm": 2.780158042907715,
+      "learning_rate": 3.2016460905349795e-05,
+      "loss": 0.5889,
+      "step": 572
+    },
+    {
+      "epoch": 1.2712146422628952,
+      "grad_norm": 2.588772773742676,
+      "learning_rate": 3.197530864197531e-05,
+      "loss": 0.6548,
+      "step": 573
+    },
+    {
+      "epoch": 1.2734331669439822,
+      "grad_norm": 2.1758575439453125,
+      "learning_rate": 3.193415637860083e-05,
+      "loss": 0.541,
+      "step": 574
+    },
+    {
+      "epoch": 1.2756516916250693,
+      "grad_norm": 2.276061534881592,
+      "learning_rate": 3.1893004115226336e-05,
+      "loss": 0.6665,
+      "step": 575
+    },
+    {
+      "epoch": 1.2778702163061564,
+      "grad_norm": 2.0960500240325928,
+      "learning_rate": 3.185185185185185e-05,
+      "loss": 0.4968,
+      "step": 576
+    },
+    {
+      "epoch": 1.2800887409872435,
+      "grad_norm": 2.1840274333953857,
+      "learning_rate": 3.181069958847737e-05,
+      "loss": 0.5605,
+      "step": 577
+    },
+    {
+      "epoch": 1.2823072656683305,
+      "grad_norm": 1.9419673681259155,
+      "learning_rate": 3.176954732510288e-05,
+      "loss": 0.5851,
+      "step": 578
+    },
+    {
+      "epoch": 1.2845257903494176,
+      "grad_norm": 1.9319486618041992,
+      "learning_rate": 3.172839506172839e-05,
+      "loss": 0.4939,
+      "step": 579
+    },
+    {
+      "epoch": 1.2867443150305047,
+      "grad_norm": 2.6804351806640625,
+      "learning_rate": 3.168724279835391e-05,
+      "loss": 0.6921,
+      "step": 580
+    },
+    {
+      "epoch": 1.2889628397115918,
+      "grad_norm": 2.292356252670288,
+      "learning_rate": 3.164609053497943e-05,
+      "loss": 0.6893,
+      "step": 581
+    },
+    {
+      "epoch": 1.2911813643926788,
+      "grad_norm": 1.867220401763916,
+      "learning_rate": 3.160493827160494e-05,
+      "loss": 0.5086,
+      "step": 582
+    },
+    {
+      "epoch": 1.293399889073766,
+      "grad_norm": 2.229401111602783,
+      "learning_rate": 3.1563786008230456e-05,
+      "loss": 0.4348,
+      "step": 583
+    },
+    {
+      "epoch": 1.295618413754853,
+      "grad_norm": 2.4550673961639404,
+      "learning_rate": 3.1522633744855965e-05,
+      "loss": 0.5262,
+      "step": 584
+    },
+    {
+      "epoch": 1.29783693843594,
+      "grad_norm": 2.5540525913238525,
+      "learning_rate": 3.148148148148148e-05,
+      "loss": 0.6108,
+      "step": 585
+    },
+    {
+      "epoch": 1.3000554631170271,
+      "grad_norm": 1.9748079776763916,
+      "learning_rate": 3.1440329218106997e-05,
+      "loss": 0.391,
+      "step": 586
+    },
+    {
+      "epoch": 1.3022739877981142,
+      "grad_norm": 2.3239924907684326,
+      "learning_rate": 3.139917695473251e-05,
+      "loss": 0.4509,
+      "step": 587
+    },
+    {
+      "epoch": 1.3044925124792013,
+      "grad_norm": 2.4383327960968018,
+      "learning_rate": 3.135802469135803e-05,
+      "loss": 0.6071,
+      "step": 588
+    },
+    {
+      "epoch": 1.3067110371602884,
+      "grad_norm": 2.252150774002075,
+      "learning_rate": 3.131687242798354e-05,
+      "loss": 0.5366,
+      "step": 589
+    },
+    {
+      "epoch": 1.3089295618413754,
+      "grad_norm": 2.6697232723236084,
+      "learning_rate": 3.127572016460906e-05,
+      "loss": 0.4877,
+      "step": 590
+    },
+    {
+      "epoch": 1.3111480865224625,
+      "grad_norm": 1.8694210052490234,
+      "learning_rate": 3.123456790123457e-05,
+      "loss": 0.4301,
+      "step": 591
+    },
+    {
+      "epoch": 1.3133666112035496,
+      "grad_norm": 2.960336685180664,
+      "learning_rate": 3.1193415637860085e-05,
+      "loss": 0.4767,
+      "step": 592
+    },
+    {
+      "epoch": 1.3155851358846367,
+      "grad_norm": 1.981068730354309,
+      "learning_rate": 3.1152263374485594e-05,
+      "loss": 0.5759,
+      "step": 593
+    },
+    {
+      "epoch": 1.3178036605657237,
+      "grad_norm": 2.4632904529571533,
+      "learning_rate": 3.111111111111111e-05,
+      "loss": 0.7422,
+      "step": 594
+    },
+    {
+      "epoch": 1.3200221852468108,
+      "grad_norm": 2.4641497135162354,
+      "learning_rate": 3.1069958847736626e-05,
+      "loss": 0.7962,
+      "step": 595
+    },
+    {
+      "epoch": 1.3222407099278979,
+      "grad_norm": 2.0334317684173584,
+      "learning_rate": 3.102880658436214e-05,
+      "loss": 0.4725,
+      "step": 596
+    },
+    {
+      "epoch": 1.324459234608985,
+      "grad_norm": 2.2943968772888184,
+      "learning_rate": 3.098765432098766e-05,
+      "loss": 0.4844,
+      "step": 597
+    },
+    {
+      "epoch": 1.326677759290072,
+      "grad_norm": 2.264467716217041,
+      "learning_rate": 3.0946502057613167e-05,
+      "loss": 0.4704,
+      "step": 598
+    },
+    {
+      "epoch": 1.328896283971159,
+      "grad_norm": 1.8216358423233032,
+      "learning_rate": 3.090534979423869e-05,
+      "loss": 0.3846,
+      "step": 599
+    },
+    {
+      "epoch": 1.3311148086522462,
+      "grad_norm": 2.1953861713409424,
+      "learning_rate": 3.08641975308642e-05,
+      "loss": 0.6035,
+      "step": 600
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 2.55543851852417,
+      "learning_rate": 3.0823045267489714e-05,
+      "loss": 0.6432,
+      "step": 601
+    },
+    {
+      "epoch": 1.3355518580144203,
+      "grad_norm": 2.383530855178833,
+      "learning_rate": 3.078189300411522e-05,
+      "loss": 0.6083,
+      "step": 602
+    },
+    {
+      "epoch": 1.3377703826955074,
+      "grad_norm": 2.0405237674713135,
+      "learning_rate": 3.074074074074074e-05,
+      "loss": 0.5081,
+      "step": 603
+    },
+    {
+      "epoch": 1.3399889073765945,
+      "grad_norm": 2.6011359691619873,
+      "learning_rate": 3.069958847736626e-05,
+      "loss": 0.5601,
+      "step": 604
+    },
+    {
+      "epoch": 1.3422074320576816,
+      "grad_norm": 2.63419508934021,
+      "learning_rate": 3.065843621399177e-05,
+      "loss": 0.6334,
+      "step": 605
+    },
+    {
+      "epoch": 1.3444259567387689,
+      "grad_norm": 2.2986178398132324,
+      "learning_rate": 3.061728395061729e-05,
+      "loss": 0.656,
+      "step": 606
+    },
+    {
+      "epoch": 1.346644481419856,
+      "grad_norm": 2.2699239253997803,
+      "learning_rate": 3.0576131687242796e-05,
+      "loss": 0.7082,
+      "step": 607
+    },
+    {
+      "epoch": 1.348863006100943,
+      "grad_norm": 2.233299732208252,
+      "learning_rate": 3.053497942386832e-05,
+      "loss": 0.6025,
+      "step": 608
+    },
+    {
+      "epoch": 1.35108153078203,
+      "grad_norm": 2.1833865642547607,
+      "learning_rate": 3.0493827160493827e-05,
+      "loss": 0.6846,
+      "step": 609
+    },
+    {
+      "epoch": 1.3533000554631172,
+      "grad_norm": 2.3082120418548584,
+      "learning_rate": 3.0452674897119343e-05,
+      "loss": 0.5177,
+      "step": 610
+    },
+    {
+      "epoch": 1.3555185801442042,
+      "grad_norm": 2.026233434677124,
+      "learning_rate": 3.041152263374486e-05,
+      "loss": 0.4083,
+      "step": 611
+    },
+    {
+      "epoch": 1.3577371048252913,
+      "grad_norm": 3.010104179382324,
+      "learning_rate": 3.037037037037037e-05,
+      "loss": 0.5326,
+      "step": 612
+    },
+    {
+      "epoch": 1.3599556295063784,
+      "grad_norm": 2.501295566558838,
+      "learning_rate": 3.0329218106995887e-05,
+      "loss": 0.506,
+      "step": 613
+    },
+    {
+      "epoch": 1.3621741541874655,
+      "grad_norm": 2.170513153076172,
+      "learning_rate": 3.02880658436214e-05,
+      "loss": 0.5247,
+      "step": 614
+    },
+    {
+      "epoch": 1.3643926788685525,
+      "grad_norm": 2.0497632026672363,
+      "learning_rate": 3.0246913580246916e-05,
+      "loss": 0.5088,
+      "step": 615
+    },
+    {
+      "epoch": 1.3666112035496396,
+      "grad_norm": 2.317243814468384,
+      "learning_rate": 3.0205761316872428e-05,
+      "loss": 0.5164,
+      "step": 616
+    },
+    {
+      "epoch": 1.3688297282307267,
+      "grad_norm": 2.3622162342071533,
+      "learning_rate": 3.0164609053497944e-05,
+      "loss": 0.633,
+      "step": 617
+    },
+    {
+      "epoch": 1.3710482529118138,
+      "grad_norm": 2.370572328567505,
+      "learning_rate": 3.012345679012346e-05,
+      "loss": 0.5974,
+      "step": 618
+    },
+    {
+      "epoch": 1.3732667775929008,
+      "grad_norm": 2.453535318374634,
+      "learning_rate": 3.0082304526748972e-05,
+      "loss": 0.5206,
+      "step": 619
+    },
+    {
+      "epoch": 1.375485302273988,
+      "grad_norm": 2.3509390354156494,
+      "learning_rate": 3.0041152263374488e-05,
+      "loss": 0.5254,
+      "step": 620
+    },
+    {
+      "epoch": 1.377703826955075,
+      "grad_norm": 2.2925517559051514,
+      "learning_rate": 3e-05,
+      "loss": 0.6652,
+      "step": 621
+    },
+    {
+      "epoch": 1.379922351636162,
+      "grad_norm": 2.0061333179473877,
+      "learning_rate": 2.9958847736625517e-05,
+      "loss": 0.4235,
+      "step": 622
+    },
+    {
+      "epoch": 1.3821408763172491,
+      "grad_norm": 2.3848578929901123,
+      "learning_rate": 2.991769547325103e-05,
+      "loss": 0.5378,
+      "step": 623
+    },
+    {
+      "epoch": 1.3843594009983362,
+      "grad_norm": 2.3057801723480225,
+      "learning_rate": 2.9876543209876545e-05,
+      "loss": 0.5414,
+      "step": 624
+    },
+    {
+      "epoch": 1.3865779256794233,
+      "grad_norm": 2.2040059566497803,
+      "learning_rate": 2.9835390946502057e-05,
+      "loss": 0.4418,
+      "step": 625
+    },
+    {
+      "epoch": 1.3887964503605104,
+      "grad_norm": 2.2582526206970215,
+      "learning_rate": 2.9794238683127573e-05,
+      "loss": 0.4206,
+      "step": 626
+    },
+    {
+      "epoch": 1.3910149750415974,
+      "grad_norm": 2.142595052719116,
+      "learning_rate": 2.975308641975309e-05,
+      "loss": 0.5301,
+      "step": 627
+    },
+    {
+      "epoch": 1.3932334997226845,
+      "grad_norm": 2.2664132118225098,
+      "learning_rate": 2.97119341563786e-05,
+      "loss": 0.4419,
+      "step": 628
+    },
+    {
+      "epoch": 1.3954520244037716,
+      "grad_norm": 2.385953426361084,
+      "learning_rate": 2.9670781893004117e-05,
+      "loss": 0.5739,
+      "step": 629
+    },
+    {
+      "epoch": 1.3976705490848587,
+      "grad_norm": 2.38935923576355,
+      "learning_rate": 2.962962962962963e-05,
+      "loss": 0.5094,
+      "step": 630
+    },
+    {
+      "epoch": 1.3998890737659457,
+      "grad_norm": 2.4804189205169678,
+      "learning_rate": 2.9588477366255146e-05,
+      "loss": 0.5193,
+      "step": 631
+    },
+    {
+      "epoch": 1.4021075984470328,
+      "grad_norm": 2.221989154815674,
+      "learning_rate": 2.9547325102880658e-05,
+      "loss": 0.4448,
+      "step": 632
+    },
+    {
+      "epoch": 1.4043261231281199,
+      "grad_norm": 2.127974510192871,
+      "learning_rate": 2.9506172839506174e-05,
+      "loss": 0.6063,
+      "step": 633
+    },
+    {
+      "epoch": 1.406544647809207,
+      "grad_norm": 2.711513042449951,
+      "learning_rate": 2.946502057613169e-05,
+      "loss": 0.4422,
+      "step": 634
+    },
+    {
+      "epoch": 1.408763172490294,
+      "grad_norm": 2.342075824737549,
+      "learning_rate": 2.9423868312757202e-05,
+      "loss": 0.5081,
+      "step": 635
+    },
+    {
+      "epoch": 1.410981697171381,
+      "grad_norm": 2.2848403453826904,
+      "learning_rate": 2.9382716049382718e-05,
+      "loss": 0.5714,
+      "step": 636
+    },
+    {
+      "epoch": 1.4132002218524682,
+      "grad_norm": 1.9042919874191284,
+      "learning_rate": 2.934156378600823e-05,
+      "loss": 0.4983,
+      "step": 637
+    },
+    {
+      "epoch": 1.4154187465335553,
+      "grad_norm": 2.3013734817504883,
+      "learning_rate": 2.9300411522633747e-05,
+      "loss": 0.5759,
+      "step": 638
+    },
+    {
+      "epoch": 1.4176372712146423,
+      "grad_norm": 2.320420503616333,
+      "learning_rate": 2.925925925925926e-05,
+      "loss": 0.6178,
+      "step": 639
+    },
+    {
+      "epoch": 1.4198557958957294,
+      "grad_norm": 2.3049232959747314,
+      "learning_rate": 2.9218106995884775e-05,
+      "loss": 0.4559,
+      "step": 640
+    },
+    {
+      "epoch": 1.4220743205768165,
+      "grad_norm": 2.201328992843628,
+      "learning_rate": 2.917695473251029e-05,
+      "loss": 0.5373,
+      "step": 641
+    },
+    {
+      "epoch": 1.4242928452579036,
+      "grad_norm": 2.2685835361480713,
+      "learning_rate": 2.9135802469135803e-05,
+      "loss": 0.5245,
+      "step": 642
+    },
+    {
+      "epoch": 1.4265113699389906,
+      "grad_norm": 2.8394269943237305,
+      "learning_rate": 2.909465020576132e-05,
+      "loss": 0.4163,
+      "step": 643
+    },
+    {
+      "epoch": 1.4287298946200777,
+      "grad_norm": 2.971799850463867,
+      "learning_rate": 2.905349794238683e-05,
+      "loss": 0.6166,
+      "step": 644
+    },
+    {
+      "epoch": 1.4309484193011648,
+      "grad_norm": 2.367595672607422,
+      "learning_rate": 2.9012345679012347e-05,
+      "loss": 0.5524,
+      "step": 645
+    },
+    {
+      "epoch": 1.4331669439822519,
+      "grad_norm": 2.3127245903015137,
+      "learning_rate": 2.897119341563786e-05,
+      "loss": 0.488,
+      "step": 646
+    },
+    {
+      "epoch": 1.435385468663339,
+      "grad_norm": 2.0375916957855225,
+      "learning_rate": 2.8930041152263376e-05,
+      "loss": 0.5157,
+      "step": 647
+    },
+    {
+      "epoch": 1.437603993344426,
+      "grad_norm": 2.4097347259521484,
+      "learning_rate": 2.8888888888888888e-05,
+      "loss": 0.4808,
+      "step": 648
+    },
+    {
+      "epoch": 1.439822518025513,
+      "grad_norm": 1.9461358785629272,
+      "learning_rate": 2.8847736625514404e-05,
+      "loss": 0.5892,
+      "step": 649
+    },
+    {
+      "epoch": 1.4420410427066002,
+      "grad_norm": 2.614422082901001,
+      "learning_rate": 2.880658436213992e-05,
+      "loss": 0.4664,
+      "step": 650
+    },
+    {
+      "epoch": 1.4442595673876872,
+      "grad_norm": 3.0740487575531006,
+      "learning_rate": 2.8765432098765432e-05,
+      "loss": 0.5847,
+      "step": 651
+    },
+    {
+      "epoch": 1.4464780920687743,
+      "grad_norm": 2.4231059551239014,
+      "learning_rate": 2.8724279835390948e-05,
+      "loss": 0.4945,
+      "step": 652
+    },
+    {
+      "epoch": 1.4486966167498614,
+      "grad_norm": 2.3467047214508057,
+      "learning_rate": 2.868312757201646e-05,
+      "loss": 0.3876,
+      "step": 653
+    },
+    {
+      "epoch": 1.4509151414309485,
+      "grad_norm": 2.38425612449646,
+      "learning_rate": 2.8641975308641977e-05,
+      "loss": 0.4394,
+      "step": 654
+    },
+    {
+      "epoch": 1.4531336661120355,
+      "grad_norm": 2.0597290992736816,
+      "learning_rate": 2.860082304526749e-05,
+      "loss": 0.5451,
+      "step": 655
+    },
+    {
+      "epoch": 1.4553521907931226,
+      "grad_norm": 2.4608826637268066,
+      "learning_rate": 2.8559670781893005e-05,
+      "loss": 0.459,
+      "step": 656
+    },
+    {
+      "epoch": 1.4575707154742097,
+      "grad_norm": 2.0731842517852783,
+      "learning_rate": 2.851851851851852e-05,
+      "loss": 0.472,
+      "step": 657
+    },
+    {
+      "epoch": 1.4597892401552968,
+      "grad_norm": 2.123661518096924,
+      "learning_rate": 2.8477366255144033e-05,
+      "loss": 0.5794,
+      "step": 658
+    },
+    {
+      "epoch": 1.4620077648363838,
+      "grad_norm": 2.8792505264282227,
+      "learning_rate": 2.843621399176955e-05,
+      "loss": 0.7506,
+      "step": 659
+    },
+    {
+      "epoch": 1.464226289517471,
+      "grad_norm": 2.357522964477539,
+      "learning_rate": 2.839506172839506e-05,
+      "loss": 0.418,
+      "step": 660
+    },
+    {
+      "epoch": 1.466444814198558,
+      "grad_norm": 2.579900026321411,
+      "learning_rate": 2.8353909465020577e-05,
+      "loss": 0.767,
+      "step": 661
+    },
+    {
+      "epoch": 1.468663338879645,
+      "grad_norm": 2.586224317550659,
+      "learning_rate": 2.831275720164609e-05,
+      "loss": 0.6251,
+      "step": 662
+    },
+    {
+      "epoch": 1.4708818635607321,
+      "grad_norm": 1.990934133529663,
+      "learning_rate": 2.8271604938271606e-05,
+      "loss": 0.5035,
+      "step": 663
+    },
+    {
+      "epoch": 1.4731003882418192,
+      "grad_norm": 2.8094747066497803,
+      "learning_rate": 2.823045267489712e-05,
+      "loss": 0.6209,
+      "step": 664
+    },
+    {
+      "epoch": 1.4753189129229063,
+      "grad_norm": 2.1593875885009766,
+      "learning_rate": 2.8189300411522634e-05,
+      "loss": 0.6086,
+      "step": 665
+    },
+    {
+      "epoch": 1.4775374376039934,
+      "grad_norm": 2.2435381412506104,
+      "learning_rate": 2.814814814814815e-05,
+      "loss": 0.6808,
+      "step": 666
+    },
+    {
+      "epoch": 1.4797559622850804,
+      "grad_norm": 2.229222297668457,
+      "learning_rate": 2.8106995884773662e-05,
+      "loss": 0.5321,
+      "step": 667
+    },
+    {
+      "epoch": 1.4819744869661675,
+      "grad_norm": 2.4216480255126953,
+      "learning_rate": 2.8065843621399178e-05,
+      "loss": 0.4173,
+      "step": 668
+    },
+    {
+      "epoch": 1.4841930116472546,
+      "grad_norm": 2.7127418518066406,
+      "learning_rate": 2.802469135802469e-05,
+      "loss": 0.6565,
+      "step": 669
+    },
+    {
+      "epoch": 1.4864115363283417,
+      "grad_norm": 2.2433009147644043,
+      "learning_rate": 2.7983539094650207e-05,
+      "loss": 0.5443,
+      "step": 670
+    },
+    {
+      "epoch": 1.4886300610094287,
+      "grad_norm": 2.01070237159729,
+      "learning_rate": 2.7942386831275726e-05,
+      "loss": 0.537,
+      "step": 671
+    },
+    {
+      "epoch": 1.4908485856905158,
+      "grad_norm": 2.3232204914093018,
+      "learning_rate": 2.7901234567901235e-05,
+      "loss": 0.5571,
+      "step": 672
+    },
+    {
+      "epoch": 1.493067110371603,
+      "grad_norm": 2.1172633171081543,
+      "learning_rate": 2.786008230452675e-05,
+      "loss": 0.5299,
+      "step": 673
+    },
+    {
+      "epoch": 1.49528563505269,
+      "grad_norm": 2.5957720279693604,
+      "learning_rate": 2.7818930041152263e-05,
+      "loss": 0.538,
+      "step": 674
+    },
+    {
+      "epoch": 1.497504159733777,
+      "grad_norm": 2.1347787380218506,
+      "learning_rate": 2.777777777777778e-05,
+      "loss": 0.508,
+      "step": 675
+    },
+    {
+      "epoch": 1.4997226844148641,
+      "grad_norm": 2.3274364471435547,
+      "learning_rate": 2.773662551440329e-05,
+      "loss": 0.5115,
+      "step": 676
+    },
+    {
+      "epoch": 1.5019412090959512,
+      "grad_norm": 2.144840717315674,
+      "learning_rate": 2.7695473251028807e-05,
+      "loss": 0.6448,
+      "step": 677
+    },
+    {
+      "epoch": 1.5041597337770383,
+      "grad_norm": 2.1130123138427734,
+      "learning_rate": 2.765432098765432e-05,
+      "loss": 0.3943,
+      "step": 678
+    },
+    {
+      "epoch": 1.5063782584581253,
+      "grad_norm": 1.9813129901885986,
+      "learning_rate": 2.7613168724279836e-05,
+      "loss": 0.4844,
+      "step": 679
+    },
+    {
+      "epoch": 1.5085967831392124,
+      "grad_norm": 2.163133144378662,
+      "learning_rate": 2.757201646090535e-05,
+      "loss": 0.5309,
+      "step": 680
+    },
+    {
+      "epoch": 1.5108153078202995,
+      "grad_norm": 2.154939889907837,
+      "learning_rate": 2.7530864197530864e-05,
+      "loss": 0.6394,
+      "step": 681
+    },
+    {
+      "epoch": 1.5130338325013866,
+      "grad_norm": 3.9356112480163574,
+      "learning_rate": 2.748971193415638e-05,
+      "loss": 0.7976,
+      "step": 682
+    },
+    {
+      "epoch": 1.5152523571824736,
+      "grad_norm": 2.2219185829162598,
+      "learning_rate": 2.7448559670781892e-05,
+      "loss": 0.4995,
+      "step": 683
+    },
+    {
+      "epoch": 1.5174708818635607,
+      "grad_norm": 2.471142292022705,
+      "learning_rate": 2.7407407407407408e-05,
+      "loss": 0.4821,
+      "step": 684
+    },
+    {
+      "epoch": 1.5196894065446478,
+      "grad_norm": 2.1839277744293213,
+      "learning_rate": 2.736625514403292e-05,
+      "loss": 0.5461,
+      "step": 685
+    },
+    {
+      "epoch": 1.5219079312257349,
+      "grad_norm": 2.7166779041290283,
+      "learning_rate": 2.7325102880658437e-05,
+      "loss": 0.4248,
+      "step": 686
+    },
+    {
+      "epoch": 1.524126455906822,
+      "grad_norm": 2.260599136352539,
+      "learning_rate": 2.7283950617283956e-05,
+      "loss": 0.6288,
+      "step": 687
+    },
+    {
+      "epoch": 1.526344980587909,
+      "grad_norm": 1.9971833229064941,
+      "learning_rate": 2.7242798353909465e-05,
+      "loss": 0.7029,
+      "step": 688
+    },
+    {
+      "epoch": 1.528563505268996,
+      "grad_norm": 2.441941261291504,
+      "learning_rate": 2.720164609053498e-05,
+      "loss": 0.5543,
+      "step": 689
+    },
+    {
+      "epoch": 1.5307820299500832,
+      "grad_norm": 2.0393805503845215,
+      "learning_rate": 2.7160493827160493e-05,
+      "loss": 0.5281,
+      "step": 690
+    },
+    {
+      "epoch": 1.5330005546311702,
+      "grad_norm": 2.2915632724761963,
+      "learning_rate": 2.711934156378601e-05,
+      "loss": 0.6204,
+      "step": 691
+    },
+    {
+      "epoch": 1.5352190793122573,
+      "grad_norm": 2.4581801891326904,
+      "learning_rate": 2.707818930041152e-05,
+      "loss": 0.6234,
+      "step": 692
+    },
+    {
+      "epoch": 1.5374376039933444,
+      "grad_norm": 2.1464343070983887,
+      "learning_rate": 2.7037037037037037e-05,
+      "loss": 0.6345,
+      "step": 693
+    },
+    {
+      "epoch": 1.5396561286744315,
+      "grad_norm": 2.238211154937744,
+      "learning_rate": 2.6995884773662557e-05,
+      "loss": 0.5511,
+      "step": 694
+    },
+    {
+      "epoch": 1.5418746533555185,
+      "grad_norm": 2.24833083152771,
+      "learning_rate": 2.6954732510288066e-05,
+      "loss": 0.4675,
+      "step": 695
+    },
+    {
+      "epoch": 1.5440931780366056,
+      "grad_norm": 2.141153573989868,
+      "learning_rate": 2.6913580246913585e-05,
+      "loss": 0.5204,
+      "step": 696
+    },
+    {
+      "epoch": 1.5463117027176927,
+      "grad_norm": 2.204092502593994,
+      "learning_rate": 2.6872427983539094e-05,
+      "loss": 0.5056,
+      "step": 697
+    },
+    {
+      "epoch": 1.5485302273987798,
+      "grad_norm": 2.2650368213653564,
+      "learning_rate": 2.683127572016461e-05,
+      "loss": 0.5906,
+      "step": 698
+    },
+    {
+      "epoch": 1.5507487520798668,
+      "grad_norm": 2.1435670852661133,
+      "learning_rate": 2.6790123456790122e-05,
+      "loss": 0.516,
+      "step": 699
+    },
+    {
+      "epoch": 1.552967276760954,
+      "grad_norm": 2.091292381286621,
+      "learning_rate": 2.6748971193415638e-05,
+      "loss": 0.415,
+      "step": 700
+    },
+    {
+      "epoch": 1.555185801442041,
+      "grad_norm": 2.5808703899383545,
+      "learning_rate": 2.6707818930041158e-05,
+      "loss": 0.7601,
+      "step": 701
+    },
+    {
+      "epoch": 1.557404326123128,
+      "grad_norm": 3.2317769527435303,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.554,
+      "step": 702
+    },
+    {
+      "epoch": 1.5596228508042151,
+      "grad_norm": 3.186840295791626,
+      "learning_rate": 2.6625514403292186e-05,
+      "loss": 0.8055,
+      "step": 703
+    },
+    {
+      "epoch": 1.5618413754853022,
+      "grad_norm": 2.101637125015259,
+      "learning_rate": 2.6584362139917695e-05,
+      "loss": 0.6431,
+      "step": 704
+    },
+    {
+      "epoch": 1.5640599001663893,
+      "grad_norm": 2.6127970218658447,
+      "learning_rate": 2.654320987654321e-05,
+      "loss": 0.5502,
+      "step": 705
+    },
+    {
+      "epoch": 1.5662784248474764,
+      "grad_norm": 2.3150572776794434,
+      "learning_rate": 2.6502057613168723e-05,
+      "loss": 0.5467,
+      "step": 706
+    },
+    {
+      "epoch": 1.5684969495285634,
+      "grad_norm": 2.100574254989624,
+      "learning_rate": 2.646090534979424e-05,
+      "loss": 0.469,
+      "step": 707
+    },
+    {
+      "epoch": 1.5707154742096505,
+      "grad_norm": 1.8744125366210938,
+      "learning_rate": 2.641975308641975e-05,
+      "loss": 0.4006,
+      "step": 708
+    },
+    {
+      "epoch": 1.5729339988907376,
+      "grad_norm": 2.5175564289093018,
+      "learning_rate": 2.6378600823045267e-05,
+      "loss": 0.6104,
+      "step": 709
+    },
+    {
+      "epoch": 1.5751525235718247,
+      "grad_norm": 2.0834975242614746,
+      "learning_rate": 2.6337448559670787e-05,
+      "loss": 0.5141,
+      "step": 710
+    },
+    {
+      "epoch": 1.5773710482529117,
+      "grad_norm": 2.564955949783325,
+      "learning_rate": 2.6296296296296296e-05,
+      "loss": 0.9185,
+      "step": 711
+    },
+    {
+      "epoch": 1.5795895729339988,
+      "grad_norm": 1.9477674961090088,
+      "learning_rate": 2.6255144032921815e-05,
+      "loss": 0.4835,
+      "step": 712
+    },
+    {
+      "epoch": 1.581808097615086,
+      "grad_norm": 1.986958384513855,
+      "learning_rate": 2.6213991769547324e-05,
+      "loss": 0.3706,
+      "step": 713
+    },
+    {
+      "epoch": 1.584026622296173,
+      "grad_norm": 2.3932137489318848,
+      "learning_rate": 2.617283950617284e-05,
+      "loss": 0.6249,
+      "step": 714
+    },
+    {
+      "epoch": 1.58624514697726,
+      "grad_norm": 3.311056137084961,
+      "learning_rate": 2.6131687242798352e-05,
+      "loss": 0.5419,
+      "step": 715
+    },
+    {
+      "epoch": 1.5884636716583471,
+      "grad_norm": 2.0535778999328613,
+      "learning_rate": 2.6090534979423868e-05,
+      "loss": 0.6342,
+      "step": 716
+    },
+    {
+      "epoch": 1.5906821963394342,
+      "grad_norm": 1.9541871547698975,
+      "learning_rate": 2.6049382716049388e-05,
+      "loss": 0.5037,
+      "step": 717
+    },
+    {
+      "epoch": 1.5929007210205213,
+      "grad_norm": 2.7204055786132812,
+      "learning_rate": 2.6008230452674897e-05,
+      "loss": 0.5085,
+      "step": 718
+    },
+    {
+      "epoch": 1.5951192457016083,
+      "grad_norm": 2.867917776107788,
+      "learning_rate": 2.5967078189300416e-05,
+      "loss": 0.6184,
+      "step": 719
+    },
+    {
+      "epoch": 1.5973377703826954,
+      "grad_norm": 2.399196147918701,
+      "learning_rate": 2.5925925925925925e-05,
+      "loss": 0.6127,
+      "step": 720
+    },
+    {
+      "epoch": 1.5995562950637825,
+      "grad_norm": 2.023618698120117,
+      "learning_rate": 2.5884773662551444e-05,
+      "loss": 0.6236,
+      "step": 721
+    },
+    {
+      "epoch": 1.6017748197448696,
+      "grad_norm": 2.267808675765991,
+      "learning_rate": 2.5843621399176953e-05,
+      "loss": 0.589,
+      "step": 722
+    },
+    {
+      "epoch": 1.6039933444259566,
+      "grad_norm": 2.9022340774536133,
+      "learning_rate": 2.580246913580247e-05,
+      "loss": 0.456,
+      "step": 723
+    },
+    {
+      "epoch": 1.6062118691070437,
+      "grad_norm": 2.0798535346984863,
+      "learning_rate": 2.576131687242799e-05,
+      "loss": 0.7572,
+      "step": 724
+    },
+    {
+      "epoch": 1.6084303937881308,
+      "grad_norm": 2.280350923538208,
+      "learning_rate": 2.5720164609053497e-05,
+      "loss": 0.5026,
+      "step": 725
+    },
+    {
+      "epoch": 1.6106489184692179,
+      "grad_norm": 2.2222673892974854,
+      "learning_rate": 2.5679012345679017e-05,
+      "loss": 0.5569,
+      "step": 726
+    },
+    {
+      "epoch": 1.612867443150305,
+      "grad_norm": 2.628732681274414,
+      "learning_rate": 2.5637860082304526e-05,
+      "loss": 0.5438,
+      "step": 727
+    },
+    {
+      "epoch": 1.615085967831392,
+      "grad_norm": 2.152247190475464,
+      "learning_rate": 2.5596707818930045e-05,
+      "loss": 0.5582,
+      "step": 728
+    },
+    {
+      "epoch": 1.617304492512479,
+      "grad_norm": 2.117852210998535,
+      "learning_rate": 2.5555555555555554e-05,
+      "loss": 0.5781,
+      "step": 729
+    },
+    {
+      "epoch": 1.6195230171935662,
+      "grad_norm": 2.2085587978363037,
+      "learning_rate": 2.551440329218107e-05,
+      "loss": 0.4884,
+      "step": 730
+    },
+    {
+      "epoch": 1.6217415418746532,
+      "grad_norm": 2.141461133956909,
+      "learning_rate": 2.5473251028806582e-05,
+      "loss": 0.5026,
+      "step": 731
+    },
+    {
+      "epoch": 1.6239600665557403,
+      "grad_norm": 2.445343017578125,
+      "learning_rate": 2.5432098765432098e-05,
+      "loss": 0.5365,
+      "step": 732
+    },
+    {
+      "epoch": 1.6261785912368274,
+      "grad_norm": 2.256298303604126,
+      "learning_rate": 2.5390946502057617e-05,
+      "loss": 0.4766,
+      "step": 733
+    },
+    {
+      "epoch": 1.6283971159179145,
+      "grad_norm": 2.4963643550872803,
+      "learning_rate": 2.5349794238683127e-05,
+      "loss": 0.4736,
+      "step": 734
+    },
+    {
+      "epoch": 1.6306156405990015,
+      "grad_norm": 2.300337076187134,
+      "learning_rate": 2.5308641975308646e-05,
+      "loss": 0.5262,
+      "step": 735
+    },
+    {
+      "epoch": 1.6328341652800886,
+      "grad_norm": 2.2634944915771484,
+      "learning_rate": 2.5267489711934155e-05,
+      "loss": 0.52,
+      "step": 736
+    },
+    {
+      "epoch": 1.6350526899611757,
+      "grad_norm": 2.1593151092529297,
+      "learning_rate": 2.5226337448559674e-05,
+      "loss": 0.4586,
+      "step": 737
+    },
+    {
+      "epoch": 1.6372712146422628,
+      "grad_norm": 2.8624863624572754,
+      "learning_rate": 2.5185185185185183e-05,
+      "loss": 0.5027,
+      "step": 738
+    },
+    {
+      "epoch": 1.6394897393233498,
+      "grad_norm": 2.2401516437530518,
+      "learning_rate": 2.51440329218107e-05,
+      "loss": 0.6091,
+      "step": 739
+    },
+    {
+      "epoch": 1.641708264004437,
+      "grad_norm": 2.1691482067108154,
+      "learning_rate": 2.510288065843622e-05,
+      "loss": 0.556,
+      "step": 740
+    },
+    {
+      "epoch": 1.643926788685524,
+      "grad_norm": 2.5343592166900635,
+      "learning_rate": 2.5061728395061727e-05,
+      "loss": 0.5829,
+      "step": 741
+    },
+    {
+      "epoch": 1.646145313366611,
+      "grad_norm": 2.05989670753479,
+      "learning_rate": 2.5020576131687247e-05,
+      "loss": 0.3896,
+      "step": 742
+    },
+    {
+      "epoch": 1.6483638380476981,
+      "grad_norm": 2.512794256210327,
+      "learning_rate": 2.497942386831276e-05,
+      "loss": 0.5481,
+      "step": 743
+    },
+    {
+      "epoch": 1.6505823627287852,
+      "grad_norm": 2.6164472103118896,
+      "learning_rate": 2.4938271604938275e-05,
+      "loss": 0.5334,
+      "step": 744
+    },
+    {
+      "epoch": 1.6528008874098723,
+      "grad_norm": 2.5660600662231445,
+      "learning_rate": 2.4897119341563787e-05,
+      "loss": 0.5787,
+      "step": 745
+    },
+    {
+      "epoch": 1.6550194120909594,
+      "grad_norm": 2.3578574657440186,
+      "learning_rate": 2.4855967078189303e-05,
+      "loss": 0.5666,
+      "step": 746
+    },
+    {
+      "epoch": 1.6572379367720464,
+      "grad_norm": 2.3784055709838867,
+      "learning_rate": 2.4814814814814816e-05,
+      "loss": 0.5992,
+      "step": 747
+    },
+    {
+      "epoch": 1.6594564614531335,
+      "grad_norm": 2.2994232177734375,
+      "learning_rate": 2.4773662551440328e-05,
+      "loss": 0.5696,
+      "step": 748
+    },
+    {
+      "epoch": 1.6616749861342206,
+      "grad_norm": 2.0111184120178223,
+      "learning_rate": 2.4732510288065844e-05,
+      "loss": 0.4792,
+      "step": 749
+    },
+    {
+      "epoch": 1.6638935108153077,
+      "grad_norm": 1.8092752695083618,
+      "learning_rate": 2.4691358024691357e-05,
+      "loss": 0.4149,
+      "step": 750
+    },
+    {
+      "epoch": 1.6661120354963947,
+      "grad_norm": 1.9316834211349487,
+      "learning_rate": 2.4650205761316876e-05,
+      "loss": 0.4148,
+      "step": 751
+    },
+    {
+      "epoch": 1.6683305601774818,
+      "grad_norm": 1.943949818611145,
+      "learning_rate": 2.4609053497942388e-05,
+      "loss": 0.4426,
+      "step": 752
+    },
+    {
+      "epoch": 1.670549084858569,
+      "grad_norm": 2.4481136798858643,
+      "learning_rate": 2.4567901234567904e-05,
+      "loss": 0.4519,
+      "step": 753
+    },
+    {
+      "epoch": 1.672767609539656,
+      "grad_norm": 2.349368095397949,
+      "learning_rate": 2.4526748971193417e-05,
+      "loss": 0.5271,
+      "step": 754
+    },
+    {
+      "epoch": 1.674986134220743,
+      "grad_norm": 2.1127991676330566,
+      "learning_rate": 2.4485596707818932e-05,
+      "loss": 0.5645,
+      "step": 755
+    },
+    {
+      "epoch": 1.6772046589018301,
+      "grad_norm": 2.5234291553497314,
+      "learning_rate": 2.4444444444444445e-05,
+      "loss": 0.7111,
+      "step": 756
+    },
+    {
+      "epoch": 1.6794231835829172,
+      "grad_norm": 2.2956089973449707,
+      "learning_rate": 2.4403292181069957e-05,
+      "loss": 0.5296,
+      "step": 757
+    },
+    {
+      "epoch": 1.6816417082640043,
+      "grad_norm": 2.2203214168548584,
+      "learning_rate": 2.4362139917695477e-05,
+      "loss": 0.5465,
+      "step": 758
+    },
+    {
+      "epoch": 1.6838602329450914,
+      "grad_norm": 2.4541451930999756,
+      "learning_rate": 2.432098765432099e-05,
+      "loss": 0.5561,
+      "step": 759
+    },
+    {
+      "epoch": 1.6860787576261784,
+      "grad_norm": 2.473926067352295,
+      "learning_rate": 2.4279835390946505e-05,
+      "loss": 0.4814,
+      "step": 760
+    },
+    {
+      "epoch": 1.6882972823072657,
+      "grad_norm": 2.0853326320648193,
+      "learning_rate": 2.4238683127572017e-05,
+      "loss": 0.5931,
+      "step": 761
+    },
+    {
+      "epoch": 1.6905158069883528,
+      "grad_norm": 2.2979753017425537,
+      "learning_rate": 2.4197530864197533e-05,
+      "loss": 0.4835,
+      "step": 762
+    },
+    {
+      "epoch": 1.6927343316694399,
+      "grad_norm": 2.5644450187683105,
+      "learning_rate": 2.4156378600823046e-05,
+      "loss": 0.7892,
+      "step": 763
+    },
+    {
+      "epoch": 1.694952856350527,
+      "grad_norm": 2.850330352783203,
+      "learning_rate": 2.4115226337448558e-05,
+      "loss": 0.5174,
+      "step": 764
+    },
+    {
+      "epoch": 1.697171381031614,
+      "grad_norm": 2.3317348957061768,
+      "learning_rate": 2.4074074074074074e-05,
+      "loss": 0.5761,
+      "step": 765
+    },
+    {
+      "epoch": 1.699389905712701,
+      "grad_norm": 2.2404422760009766,
+      "learning_rate": 2.403292181069959e-05,
+      "loss": 0.5547,
+      "step": 766
+    },
+    {
+      "epoch": 1.7016084303937882,
+      "grad_norm": 2.2664194107055664,
+      "learning_rate": 2.3991769547325106e-05,
+      "loss": 0.5428,
+      "step": 767
+    },
+    {
+      "epoch": 1.7038269550748752,
+      "grad_norm": 1.9951173067092896,
+      "learning_rate": 2.3950617283950618e-05,
+      "loss": 0.4851,
+      "step": 768
+    },
+    {
+      "epoch": 1.7060454797559623,
+      "grad_norm": 2.8596034049987793,
+      "learning_rate": 2.3909465020576134e-05,
+      "loss": 0.5051,
+      "step": 769
+    },
+    {
+      "epoch": 1.7082640044370494,
+      "grad_norm": 2.380742311477661,
+      "learning_rate": 2.3868312757201647e-05,
+      "loss": 0.5906,
+      "step": 770
+    },
+    {
+      "epoch": 1.7104825291181365,
+      "grad_norm": 1.9263993501663208,
+      "learning_rate": 2.3827160493827162e-05,
+      "loss": 0.5275,
+      "step": 771
+    },
+    {
+      "epoch": 1.7127010537992235,
+      "grad_norm": 2.614659547805786,
+      "learning_rate": 2.3786008230452675e-05,
+      "loss": 0.4473,
+      "step": 772
+    },
+    {
+      "epoch": 1.7149195784803106,
+      "grad_norm": 2.3900511264801025,
+      "learning_rate": 2.374485596707819e-05,
+      "loss": 0.5736,
+      "step": 773
+    },
+    {
+      "epoch": 1.7171381031613977,
+      "grad_norm": 2.6308071613311768,
+      "learning_rate": 2.3703703703703707e-05,
+      "loss": 0.6588,
+      "step": 774
+    },
+    {
+      "epoch": 1.7193566278424848,
+      "grad_norm": 2.2880427837371826,
+      "learning_rate": 2.366255144032922e-05,
+      "loss": 0.468,
+      "step": 775
+    },
+    {
+      "epoch": 1.7215751525235718,
+      "grad_norm": 2.0532939434051514,
+      "learning_rate": 2.3621399176954735e-05,
+      "loss": 0.5559,
+      "step": 776
+    },
+    {
+      "epoch": 1.723793677204659,
+      "grad_norm": 2.1995911598205566,
+      "learning_rate": 2.3580246913580247e-05,
+      "loss": 0.7173,
+      "step": 777
+    },
+    {
+      "epoch": 1.726012201885746,
+      "grad_norm": 2.1935765743255615,
+      "learning_rate": 2.3539094650205763e-05,
+      "loss": 0.6418,
+      "step": 778
+    },
+    {
+      "epoch": 1.728230726566833,
+      "grad_norm": 2.4513421058654785,
+      "learning_rate": 2.3497942386831276e-05,
+      "loss": 0.6503,
+      "step": 779
+    },
+    {
+      "epoch": 1.7304492512479202,
+      "grad_norm": 2.5517115592956543,
+      "learning_rate": 2.345679012345679e-05,
+      "loss": 0.5889,
+      "step": 780
+    },
+    {
+      "epoch": 1.7326677759290072,
+      "grad_norm": 2.096762180328369,
+      "learning_rate": 2.3415637860082307e-05,
+      "loss": 0.5276,
+      "step": 781
+    },
+    {
+      "epoch": 1.7348863006100943,
+      "grad_norm": 2.4286603927612305,
+      "learning_rate": 2.337448559670782e-05,
+      "loss": 0.5376,
+      "step": 782
+    },
+    {
+      "epoch": 1.7371048252911814,
+      "grad_norm": 2.3076093196868896,
+      "learning_rate": 2.3333333333333336e-05,
+      "loss": 0.527,
+      "step": 783
+    },
+    {
+      "epoch": 1.7393233499722685,
+      "grad_norm": 2.6694893836975098,
+      "learning_rate": 2.3292181069958848e-05,
+      "loss": 0.5376,
+      "step": 784
+    },
+    {
+      "epoch": 1.7415418746533555,
+      "grad_norm": 2.15592360496521,
+      "learning_rate": 2.3251028806584364e-05,
+      "loss": 0.5455,
+      "step": 785
+    },
+    {
+      "epoch": 1.7437603993344426,
+      "grad_norm": 2.186290979385376,
+      "learning_rate": 2.3209876543209877e-05,
+      "loss": 0.5706,
+      "step": 786
+    },
+    {
+      "epoch": 1.7459789240155297,
+      "grad_norm": 2.2007365226745605,
+      "learning_rate": 2.3168724279835392e-05,
+      "loss": 0.4906,
+      "step": 787
+    },
+    {
+      "epoch": 1.7481974486966168,
+      "grad_norm": 2.4092190265655518,
+      "learning_rate": 2.312757201646091e-05,
+      "loss": 0.7892,
+      "step": 788
+    },
+    {
+      "epoch": 1.7504159733777038,
+      "grad_norm": 1.9239741563796997,
+      "learning_rate": 2.308641975308642e-05,
+      "loss": 0.4777,
+      "step": 789
+    },
+    {
+      "epoch": 1.752634498058791,
+      "grad_norm": 2.0306801795959473,
+      "learning_rate": 2.3045267489711937e-05,
+      "loss": 0.4376,
+      "step": 790
+    },
+    {
+      "epoch": 1.754853022739878,
+      "grad_norm": 2.378998041152954,
+      "learning_rate": 2.300411522633745e-05,
+      "loss": 0.5328,
+      "step": 791
+    },
+    {
+      "epoch": 1.757071547420965,
+      "grad_norm": 1.846678614616394,
+      "learning_rate": 2.2962962962962965e-05,
+      "loss": 0.369,
+      "step": 792
+    },
+    {
+      "epoch": 1.7592900721020521,
+      "grad_norm": 1.9798860549926758,
+      "learning_rate": 2.2921810699588477e-05,
+      "loss": 0.44,
+      "step": 793
+    },
+    {
+      "epoch": 1.7615085967831392,
+      "grad_norm": 2.4514482021331787,
+      "learning_rate": 2.2880658436213993e-05,
+      "loss": 0.4997,
+      "step": 794
+    },
+    {
+      "epoch": 1.7637271214642263,
+      "grad_norm": 2.056880235671997,
+      "learning_rate": 2.2839506172839506e-05,
+      "loss": 0.5525,
+      "step": 795
+    },
+    {
+      "epoch": 1.7659456461453134,
+      "grad_norm": 2.419100046157837,
+      "learning_rate": 2.279835390946502e-05,
+      "loss": 0.6464,
+      "step": 796
+    },
+    {
+      "epoch": 1.7681641708264004,
+      "grad_norm": 2.366624116897583,
+      "learning_rate": 2.2757201646090537e-05,
+      "loss": 0.5389,
+      "step": 797
+    },
+    {
+      "epoch": 1.7703826955074875,
+      "grad_norm": 2.2852563858032227,
+      "learning_rate": 2.271604938271605e-05,
+      "loss": 0.6083,
+      "step": 798
+    },
+    {
+      "epoch": 1.7726012201885746,
+      "grad_norm": 2.3267345428466797,
+      "learning_rate": 2.2674897119341566e-05,
+      "loss": 0.5681,
+      "step": 799
+    },
+    {
+      "epoch": 1.7748197448696617,
+      "grad_norm": 2.4281551837921143,
+      "learning_rate": 2.2633744855967078e-05,
+      "loss": 0.4536,
+      "step": 800
+    },
+    {
+      "epoch": 1.7770382695507487,
+      "grad_norm": 2.25105357170105,
+      "learning_rate": 2.2592592592592594e-05,
+      "loss": 0.5241,
+      "step": 801
+    },
+    {
+      "epoch": 1.7792567942318358,
+      "grad_norm": 2.1971495151519775,
+      "learning_rate": 2.2551440329218107e-05,
+      "loss": 0.4471,
+      "step": 802
+    },
+    {
+      "epoch": 1.7814753189129229,
+      "grad_norm": 2.0844979286193848,
+      "learning_rate": 2.2510288065843622e-05,
+      "loss": 0.4902,
+      "step": 803
+    },
+    {
+      "epoch": 1.78369384359401,
+      "grad_norm": 2.262972354888916,
+      "learning_rate": 2.246913580246914e-05,
+      "loss": 0.5954,
+      "step": 804
+    },
+    {
+      "epoch": 1.785912368275097,
+      "grad_norm": 1.9526909589767456,
+      "learning_rate": 2.242798353909465e-05,
+      "loss": 0.4281,
+      "step": 805
+    },
+    {
+      "epoch": 1.788130892956184,
+      "grad_norm": 1.9824551343917847,
+      "learning_rate": 2.2386831275720167e-05,
+      "loss": 0.5253,
+      "step": 806
+    },
+    {
+      "epoch": 1.7903494176372712,
+      "grad_norm": 2.468169927597046,
+      "learning_rate": 2.234567901234568e-05,
+      "loss": 0.4205,
+      "step": 807
+    },
+    {
+      "epoch": 1.7925679423183583,
+      "grad_norm": 2.1783759593963623,
+      "learning_rate": 2.2304526748971195e-05,
+      "loss": 0.6124,
+      "step": 808
+    },
+    {
+      "epoch": 1.7947864669994453,
+      "grad_norm": 2.028745651245117,
+      "learning_rate": 2.2263374485596707e-05,
+      "loss": 0.4541,
+      "step": 809
+    },
+    {
+      "epoch": 1.7970049916805324,
+      "grad_norm": 1.8194847106933594,
+      "learning_rate": 2.2222222222222223e-05,
+      "loss": 0.4581,
+      "step": 810
+    },
+    {
+      "epoch": 1.7992235163616195,
+      "grad_norm": 2.0109636783599854,
+      "learning_rate": 2.218106995884774e-05,
+      "loss": 0.5453,
+      "step": 811
+    },
+    {
+      "epoch": 1.8014420410427066,
+      "grad_norm": 2.049104928970337,
+      "learning_rate": 2.213991769547325e-05,
+      "loss": 0.6151,
+      "step": 812
+    },
+    {
+      "epoch": 1.8036605657237936,
+      "grad_norm": 1.7753888368606567,
+      "learning_rate": 2.2098765432098767e-05,
+      "loss": 0.625,
+      "step": 813
+    },
+    {
+      "epoch": 1.8058790904048807,
+      "grad_norm": 2.4805421829223633,
+      "learning_rate": 2.205761316872428e-05,
+      "loss": 0.4992,
+      "step": 814
+    },
+    {
+      "epoch": 1.8080976150859678,
+      "grad_norm": 2.201561689376831,
+      "learning_rate": 2.2016460905349796e-05,
+      "loss": 0.4424,
+      "step": 815
+    },
+    {
+      "epoch": 1.8103161397670549,
+      "grad_norm": 2.4569079875946045,
+      "learning_rate": 2.1975308641975308e-05,
+      "loss": 0.7054,
+      "step": 816
+    },
+    {
+      "epoch": 1.8125346644481422,
+      "grad_norm": 2.7239222526550293,
+      "learning_rate": 2.1934156378600824e-05,
+      "loss": 0.5362,
+      "step": 817
+    },
+    {
+      "epoch": 1.8147531891292292,
+      "grad_norm": 1.9250047206878662,
+      "learning_rate": 2.189300411522634e-05,
+      "loss": 0.5817,
+      "step": 818
+    },
+    {
+      "epoch": 1.8169717138103163,
+      "grad_norm": 2.73197865486145,
+      "learning_rate": 2.1851851851851852e-05,
+      "loss": 0.4832,
+      "step": 819
+    },
+    {
+      "epoch": 1.8191902384914034,
+      "grad_norm": 2.7882120609283447,
+      "learning_rate": 2.1810699588477368e-05,
+      "loss": 0.5813,
+      "step": 820
+    },
+    {
+      "epoch": 1.8214087631724905,
+      "grad_norm": 2.2308642864227295,
+      "learning_rate": 2.176954732510288e-05,
+      "loss": 0.5375,
+      "step": 821
+    },
+    {
+      "epoch": 1.8236272878535775,
+      "grad_norm": 2.769709825515747,
+      "learning_rate": 2.1728395061728397e-05,
+      "loss": 0.4928,
+      "step": 822
+    },
+    {
+      "epoch": 1.8258458125346646,
+      "grad_norm": 2.3876256942749023,
+      "learning_rate": 2.168724279835391e-05,
+      "loss": 0.4558,
+      "step": 823
+    },
+    {
+      "epoch": 1.8280643372157517,
+      "grad_norm": 2.717535972595215,
+      "learning_rate": 2.1646090534979425e-05,
+      "loss": 0.6041,
+      "step": 824
+    },
+    {
+      "epoch": 1.8302828618968388,
+      "grad_norm": 2.755963087081909,
+      "learning_rate": 2.1604938271604937e-05,
+      "loss": 0.6511,
+      "step": 825
+    },
+    {
+      "epoch": 1.8325013865779258,
+      "grad_norm": 2.809054136276245,
+      "learning_rate": 2.1563786008230453e-05,
+      "loss": 0.5582,
+      "step": 826
+    },
+    {
+      "epoch": 1.834719911259013,
+      "grad_norm": 2.4799842834472656,
+      "learning_rate": 2.152263374485597e-05,
+      "loss": 0.6776,
+      "step": 827
+    },
+    {
+      "epoch": 1.8369384359401,
+      "grad_norm": 2.3058645725250244,
+      "learning_rate": 2.148148148148148e-05,
+      "loss": 0.4912,
+      "step": 828
+    },
+    {
+      "epoch": 1.839156960621187,
+      "grad_norm": 2.3522088527679443,
+      "learning_rate": 2.1440329218106997e-05,
+      "loss": 0.6117,
+      "step": 829
+    },
+    {
+      "epoch": 1.8413754853022741,
+      "grad_norm": 2.2719180583953857,
+      "learning_rate": 2.139917695473251e-05,
+      "loss": 0.5303,
+      "step": 830
+    },
+    {
+      "epoch": 1.8435940099833612,
+      "grad_norm": 2.2038021087646484,
+      "learning_rate": 2.1358024691358026e-05,
+      "loss": 0.5457,
+      "step": 831
+    },
+    {
+      "epoch": 1.8458125346644483,
+      "grad_norm": 2.165065050125122,
+      "learning_rate": 2.1316872427983538e-05,
+      "loss": 0.561,
+      "step": 832
+    },
+    {
+      "epoch": 1.8480310593455354,
+      "grad_norm": 2.5802323818206787,
+      "learning_rate": 2.1275720164609054e-05,
+      "loss": 0.6211,
+      "step": 833
+    },
+    {
+      "epoch": 1.8502495840266224,
+      "grad_norm": 1.9461251497268677,
+      "learning_rate": 2.123456790123457e-05,
+      "loss": 0.4731,
+      "step": 834
+    },
+    {
+      "epoch": 1.8524681087077095,
+      "grad_norm": 2.741570472717285,
+      "learning_rate": 2.1193415637860082e-05,
+      "loss": 0.5517,
+      "step": 835
+    },
+    {
+      "epoch": 1.8546866333887966,
+      "grad_norm": 2.5907652378082275,
+      "learning_rate": 2.1152263374485598e-05,
+      "loss": 0.6719,
+      "step": 836
+    },
+    {
+      "epoch": 1.8569051580698837,
+      "grad_norm": 2.476729393005371,
+      "learning_rate": 2.111111111111111e-05,
+      "loss": 0.5074,
+      "step": 837
+    },
+    {
+      "epoch": 1.8591236827509707,
+      "grad_norm": 2.4137723445892334,
+      "learning_rate": 2.1069958847736627e-05,
+      "loss": 0.4588,
+      "step": 838
+    },
+    {
+      "epoch": 1.8613422074320578,
+      "grad_norm": 2.0940053462982178,
+      "learning_rate": 2.102880658436214e-05,
+      "loss": 0.4599,
+      "step": 839
+    },
+    {
+      "epoch": 1.8635607321131449,
+      "grad_norm": 2.091827392578125,
+      "learning_rate": 2.0987654320987655e-05,
+      "loss": 0.5653,
+      "step": 840
+    },
+    {
+      "epoch": 1.865779256794232,
+      "grad_norm": 2.5018434524536133,
+      "learning_rate": 2.094650205761317e-05,
+      "loss": 0.5909,
+      "step": 841
+    },
+    {
+      "epoch": 1.867997781475319,
+      "grad_norm": 2.298170566558838,
+      "learning_rate": 2.0905349794238687e-05,
+      "loss": 0.6819,
+      "step": 842
+    },
+    {
+      "epoch": 1.870216306156406,
+      "grad_norm": 2.1202032566070557,
+      "learning_rate": 2.08641975308642e-05,
+      "loss": 0.4525,
+      "step": 843
+    },
+    {
+      "epoch": 1.8724348308374932,
+      "grad_norm": 2.8783698081970215,
+      "learning_rate": 2.082304526748971e-05,
+      "loss": 0.6115,
+      "step": 844
+    },
+    {
+      "epoch": 1.8746533555185803,
+      "grad_norm": 1.9839705228805542,
+      "learning_rate": 2.0781893004115227e-05,
+      "loss": 0.5669,
+      "step": 845
+    },
+    {
+      "epoch": 1.8768718801996673,
+      "grad_norm": 2.3747284412384033,
+      "learning_rate": 2.074074074074074e-05,
+      "loss": 0.5719,
+      "step": 846
+    },
+    {
+      "epoch": 1.8790904048807544,
+      "grad_norm": 2.025235891342163,
+      "learning_rate": 2.0699588477366256e-05,
+      "loss": 0.5006,
+      "step": 847
+    },
+    {
+      "epoch": 1.8813089295618415,
+      "grad_norm": 2.277076482772827,
+      "learning_rate": 2.0658436213991768e-05,
+      "loss": 0.5027,
+      "step": 848
+    },
+    {
+      "epoch": 1.8835274542429286,
+      "grad_norm": 2.2696785926818848,
+      "learning_rate": 2.0617283950617287e-05,
+      "loss": 0.478,
+      "step": 849
+    },
+    {
+      "epoch": 1.8857459789240156,
+      "grad_norm": 2.482240676879883,
+      "learning_rate": 2.05761316872428e-05,
+      "loss": 0.5909,
+      "step": 850
+    },
+    {
+      "epoch": 1.8879645036051027,
+      "grad_norm": 2.4486875534057617,
+      "learning_rate": 2.0534979423868312e-05,
+      "loss": 0.5768,
+      "step": 851
+    },
+    {
+      "epoch": 1.8901830282861898,
+      "grad_norm": 2.698031425476074,
+      "learning_rate": 2.0493827160493828e-05,
+      "loss": 0.5326,
+      "step": 852
+    },
+    {
+      "epoch": 1.8924015529672769,
+      "grad_norm": 3.2022013664245605,
+      "learning_rate": 2.045267489711934e-05,
+      "loss": 0.4471,
+      "step": 853
+    },
+    {
+      "epoch": 1.894620077648364,
+      "grad_norm": 2.440455913543701,
+      "learning_rate": 2.0411522633744857e-05,
+      "loss": 0.5527,
+      "step": 854
+    },
+    {
+      "epoch": 1.896838602329451,
+      "grad_norm": 2.098097801208496,
+      "learning_rate": 2.037037037037037e-05,
+      "loss": 0.8188,
+      "step": 855
+    },
+    {
+      "epoch": 1.899057127010538,
+      "grad_norm": 1.5712358951568604,
+      "learning_rate": 2.032921810699589e-05,
+      "loss": 0.4915,
+      "step": 856
+    },
+    {
+      "epoch": 1.9012756516916252,
+      "grad_norm": 2.1927616596221924,
+      "learning_rate": 2.02880658436214e-05,
+      "loss": 0.4531,
+      "step": 857
+    },
+    {
+      "epoch": 1.9034941763727122,
+      "grad_norm": 2.3700389862060547,
+      "learning_rate": 2.0246913580246917e-05,
+      "loss": 0.5468,
+      "step": 858
+    },
+    {
+      "epoch": 1.9057127010537993,
+      "grad_norm": 2.5128328800201416,
+      "learning_rate": 2.020576131687243e-05,
+      "loss": 0.7061,
+      "step": 859
+    },
+    {
+      "epoch": 1.9079312257348864,
+      "grad_norm": 2.3993642330169678,
+      "learning_rate": 2.016460905349794e-05,
+      "loss": 0.55,
+      "step": 860
+    },
+    {
+      "epoch": 1.9101497504159735,
+      "grad_norm": 1.7821239233016968,
+      "learning_rate": 2.0123456790123457e-05,
+      "loss": 0.784,
+      "step": 861
+    },
+    {
+      "epoch": 1.9123682750970605,
+      "grad_norm": 2.4146625995635986,
+      "learning_rate": 2.008230452674897e-05,
+      "loss": 0.459,
+      "step": 862
+    },
+    {
+      "epoch": 1.9145867997781476,
+      "grad_norm": 1.7972438335418701,
+      "learning_rate": 2.0041152263374486e-05,
+      "loss": 0.4539,
+      "step": 863
+    },
+    {
+      "epoch": 1.9168053244592347,
+      "grad_norm": 2.704371690750122,
+      "learning_rate": 2e-05,
+      "loss": 0.4765,
+      "step": 864
+    },
+    {
+      "epoch": 1.9190238491403218,
+      "grad_norm": 1.8903868198394775,
+      "learning_rate": 1.9958847736625517e-05,
+      "loss": 0.5274,
+      "step": 865
+    },
+    {
+      "epoch": 1.9212423738214088,
+      "grad_norm": 2.0917727947235107,
+      "learning_rate": 1.991769547325103e-05,
+      "loss": 0.4607,
+      "step": 866
+    },
+    {
+      "epoch": 1.923460898502496,
+      "grad_norm": 2.17440128326416,
+      "learning_rate": 1.9876543209876546e-05,
+      "loss": 0.4302,
+      "step": 867
+    },
+    {
+      "epoch": 1.925679423183583,
+      "grad_norm": 2.062361240386963,
+      "learning_rate": 1.9835390946502058e-05,
+      "loss": 0.6525,
+      "step": 868
+    },
+    {
+      "epoch": 1.92789794786467,
+      "grad_norm": 2.202864170074463,
+      "learning_rate": 1.979423868312757e-05,
+      "loss": 0.5557,
+      "step": 869
+    },
+    {
+      "epoch": 1.9301164725457571,
+      "grad_norm": 2.7855234146118164,
+      "learning_rate": 1.9753086419753087e-05,
+      "loss": 0.7359,
+      "step": 870
+    },
+    {
+      "epoch": 1.9323349972268442,
+      "grad_norm": 2.3044323921203613,
+      "learning_rate": 1.9711934156378602e-05,
+      "loss": 0.5439,
+      "step": 871
+    },
+    {
+      "epoch": 1.9345535219079313,
+      "grad_norm": 2.5152721405029297,
+      "learning_rate": 1.967078189300412e-05,
+      "loss": 0.4861,
+      "step": 872
+    },
+    {
+      "epoch": 1.9367720465890184,
+      "grad_norm": 2.160235643386841,
+      "learning_rate": 1.962962962962963e-05,
+      "loss": 0.4491,
+      "step": 873
+    },
+    {
+      "epoch": 1.9389905712701054,
+      "grad_norm": 2.569850444793701,
+      "learning_rate": 1.9588477366255147e-05,
+      "loss": 0.6163,
+      "step": 874
+    },
+    {
+      "epoch": 1.9412090959511925,
+      "grad_norm": 2.450078248977661,
+      "learning_rate": 1.954732510288066e-05,
+      "loss": 0.7032,
+      "step": 875
+    },
+    {
+      "epoch": 1.9434276206322796,
+      "grad_norm": 1.7826166152954102,
+      "learning_rate": 1.950617283950617e-05,
+      "loss": 0.4314,
+      "step": 876
+    },
+    {
+      "epoch": 1.9456461453133667,
+      "grad_norm": 1.922245740890503,
+      "learning_rate": 1.9465020576131687e-05,
+      "loss": 0.4445,
+      "step": 877
+    },
+    {
+      "epoch": 1.9478646699944537,
+      "grad_norm": 2.058699369430542,
+      "learning_rate": 1.94238683127572e-05,
+      "loss": 0.5031,
+      "step": 878
+    },
+    {
+      "epoch": 1.9500831946755408,
+      "grad_norm": 2.8088717460632324,
+      "learning_rate": 1.938271604938272e-05,
+      "loss": 0.5642,
+      "step": 879
+    },
+    {
+      "epoch": 1.9523017193566279,
+      "grad_norm": 2.087885856628418,
+      "learning_rate": 1.934156378600823e-05,
+      "loss": 0.5024,
+      "step": 880
+    },
+    {
+      "epoch": 1.954520244037715,
+      "grad_norm": 1.8678486347198486,
+      "learning_rate": 1.9300411522633747e-05,
+      "loss": 0.4926,
+      "step": 881
+    },
+    {
+      "epoch": 1.956738768718802,
+      "grad_norm": 2.080338716506958,
+      "learning_rate": 1.925925925925926e-05,
+      "loss": 0.5446,
+      "step": 882
+    },
+    {
+      "epoch": 1.958957293399889,
+      "grad_norm": 2.4471797943115234,
+      "learning_rate": 1.9218106995884776e-05,
+      "loss": 0.6323,
+      "step": 883
+    },
+    {
+      "epoch": 1.9611758180809762,
+      "grad_norm": 2.5934202671051025,
+      "learning_rate": 1.9176954732510288e-05,
+      "loss": 0.5565,
+      "step": 884
+    },
+    {
+      "epoch": 1.9633943427620633,
+      "grad_norm": 2.2351367473602295,
+      "learning_rate": 1.91358024691358e-05,
+      "loss": 0.4213,
+      "step": 885
+    },
+    {
+      "epoch": 1.9656128674431503,
+      "grad_norm": 2.36057186126709,
+      "learning_rate": 1.909465020576132e-05,
+      "loss": 0.5571,
+      "step": 886
+    },
+    {
+      "epoch": 1.9678313921242374,
+      "grad_norm": 2.364346742630005,
+      "learning_rate": 1.9053497942386832e-05,
+      "loss": 0.6327,
+      "step": 887
+    },
+    {
+      "epoch": 1.9700499168053245,
+      "grad_norm": 2.468201160430908,
+      "learning_rate": 1.901234567901235e-05,
+      "loss": 0.4615,
+      "step": 888
+    },
+    {
+      "epoch": 1.9722684414864116,
+      "grad_norm": 1.8754422664642334,
+      "learning_rate": 1.897119341563786e-05,
+      "loss": 0.4483,
+      "step": 889
+    },
+    {
+      "epoch": 1.9744869661674986,
+      "grad_norm": 5.303345203399658,
+      "learning_rate": 1.8930041152263377e-05,
+      "loss": 0.448,
+      "step": 890
+    },
+    {
+      "epoch": 1.9767054908485857,
+      "grad_norm": 2.6242637634277344,
+      "learning_rate": 1.888888888888889e-05,
+      "loss": 0.5398,
+      "step": 891
+    },
+    {
+      "epoch": 1.9789240155296728,
+      "grad_norm": 2.45288348197937,
+      "learning_rate": 1.8847736625514405e-05,
+      "loss": 0.4455,
+      "step": 892
+    },
+    {
+      "epoch": 1.9811425402107599,
+      "grad_norm": 2.2496774196624756,
+      "learning_rate": 1.8806584362139917e-05,
+      "loss": 0.4848,
+      "step": 893
+    },
+    {
+      "epoch": 1.983361064891847,
+      "grad_norm": 1.8748193979263306,
+      "learning_rate": 1.8765432098765433e-05,
+      "loss": 0.4773,
+      "step": 894
+    },
+    {
+      "epoch": 1.985579589572934,
+      "grad_norm": 2.149263620376587,
+      "learning_rate": 1.872427983539095e-05,
+      "loss": 0.6121,
+      "step": 895
+    },
+    {
+      "epoch": 1.987798114254021,
+      "grad_norm": 2.207383155822754,
+      "learning_rate": 1.868312757201646e-05,
+      "loss": 0.5167,
+      "step": 896
+    },
+    {
+      "epoch": 1.9900166389351082,
+      "grad_norm": 2.371633529663086,
+      "learning_rate": 1.8641975308641977e-05,
+      "loss": 0.3867,
+      "step": 897
+    },
+    {
+      "epoch": 1.9922351636161952,
+      "grad_norm": 2.1871438026428223,
+      "learning_rate": 1.860082304526749e-05,
+      "loss": 0.4338,
+      "step": 898
+    },
+    {
+      "epoch": 1.9944536882972823,
+      "grad_norm": 2.183847665786743,
+      "learning_rate": 1.8559670781893006e-05,
+      "loss": 0.4778,
+      "step": 899
+    },
+    {
+      "epoch": 1.9966722129783694,
+      "grad_norm": 2.296398639678955,
+      "learning_rate": 1.8518518518518518e-05,
+      "loss": 0.5332,
+      "step": 900
+    },
+    {
+      "epoch": 1.9988907376594565,
+      "grad_norm": 1.7822459936141968,
+      "learning_rate": 1.8477366255144034e-05,
+      "loss": 0.4996,
+      "step": 901
+    },
+    {
+      "epoch": 2.0011092623405435,
+      "grad_norm": 1.7398662567138672,
+      "learning_rate": 1.843621399176955e-05,
+      "loss": 0.411,
+      "step": 902
+    },
+    {
+      "epoch": 2.0033277870216306,
+      "grad_norm": 1.474639892578125,
+      "learning_rate": 1.8395061728395062e-05,
+      "loss": 0.2516,
+      "step": 903
+    },
+    {
+      "epoch": 2.0055463117027177,
+      "grad_norm": 1.709380030632019,
+      "learning_rate": 1.835390946502058e-05,
+      "loss": 0.2983,
+      "step": 904
+    },
+    {
+      "epoch": 2.0077648363838048,
+      "grad_norm": 1.5500868558883667,
+      "learning_rate": 1.831275720164609e-05,
+      "loss": 0.3113,
+      "step": 905
+    },
+    {
+      "epoch": 2.009983361064892,
+      "grad_norm": 2.681804656982422,
+      "learning_rate": 1.8271604938271607e-05,
+      "loss": 0.2662,
+      "step": 906
+    },
+    {
+      "epoch": 2.012201885745979,
+      "grad_norm": 1.512695550918579,
+      "learning_rate": 1.823045267489712e-05,
+      "loss": 0.3172,
+      "step": 907
+    },
+    {
+      "epoch": 2.014420410427066,
+      "grad_norm": 1.6730470657348633,
+      "learning_rate": 1.8189300411522635e-05,
+      "loss": 0.3043,
+      "step": 908
+    },
+    {
+      "epoch": 2.016638935108153,
+      "grad_norm": 2.432608127593994,
+      "learning_rate": 1.814814814814815e-05,
+      "loss": 0.2912,
+      "step": 909
+    },
+    {
+      "epoch": 2.01885745978924,
+      "grad_norm": 2.2701520919799805,
+      "learning_rate": 1.8106995884773663e-05,
+      "loss": 0.2653,
+      "step": 910
+    },
+    {
+      "epoch": 2.021075984470327,
+      "grad_norm": 2.3988521099090576,
+      "learning_rate": 1.806584362139918e-05,
+      "loss": 0.3276,
+      "step": 911
+    },
+    {
+      "epoch": 2.0232945091514143,
+      "grad_norm": 2.1256165504455566,
+      "learning_rate": 1.802469135802469e-05,
+      "loss": 0.3052,
+      "step": 912
+    },
+    {
+      "epoch": 2.0255130338325014,
+      "grad_norm": 2.6119213104248047,
+      "learning_rate": 1.7983539094650207e-05,
+      "loss": 0.2983,
+      "step": 913
+    },
+    {
+      "epoch": 2.0277315585135884,
+      "grad_norm": 2.2927403450012207,
+      "learning_rate": 1.794238683127572e-05,
+      "loss": 0.2765,
+      "step": 914
+    },
+    {
+      "epoch": 2.0299500831946755,
+      "grad_norm": 1.9266024827957153,
+      "learning_rate": 1.7901234567901236e-05,
+      "loss": 0.2173,
+      "step": 915
+    },
+    {
+      "epoch": 2.0321686078757626,
+      "grad_norm": 2.5150351524353027,
+      "learning_rate": 1.7860082304526748e-05,
+      "loss": 0.2915,
+      "step": 916
+    },
+    {
+      "epoch": 2.0343871325568497,
+      "grad_norm": 2.887460708618164,
+      "learning_rate": 1.7818930041152264e-05,
+      "loss": 0.2398,
+      "step": 917
+    },
+    {
+      "epoch": 2.0366056572379367,
+      "grad_norm": 2.7421505451202393,
+      "learning_rate": 1.777777777777778e-05,
+      "loss": 0.2076,
+      "step": 918
+    },
+    {
+      "epoch": 2.038824181919024,
+      "grad_norm": 1.9839962720870972,
+      "learning_rate": 1.7736625514403292e-05,
+      "loss": 0.2724,
+      "step": 919
+    },
+    {
+      "epoch": 2.041042706600111,
+      "grad_norm": 1.998276710510254,
+      "learning_rate": 1.769547325102881e-05,
+      "loss": 0.2836,
+      "step": 920
+    },
+    {
+      "epoch": 2.043261231281198,
+      "grad_norm": 1.722551941871643,
+      "learning_rate": 1.765432098765432e-05,
+      "loss": 0.2726,
+      "step": 921
+    },
+    {
+      "epoch": 2.045479755962285,
+      "grad_norm": 1.9660111665725708,
+      "learning_rate": 1.7613168724279837e-05,
+      "loss": 0.2698,
+      "step": 922
+    },
+    {
+      "epoch": 2.047698280643372,
+      "grad_norm": 1.5865098237991333,
+      "learning_rate": 1.757201646090535e-05,
+      "loss": 0.3552,
+      "step": 923
+    },
+    {
+      "epoch": 2.049916805324459,
+      "grad_norm": 1.5076119899749756,
+      "learning_rate": 1.7530864197530865e-05,
+      "loss": 0.2328,
+      "step": 924
+    },
+    {
+      "epoch": 2.0521353300055463,
+      "grad_norm": 1.9024195671081543,
+      "learning_rate": 1.748971193415638e-05,
+      "loss": 0.2897,
+      "step": 925
+    },
+    {
+      "epoch": 2.0543538546866333,
+      "grad_norm": 1.7773208618164062,
+      "learning_rate": 1.7448559670781893e-05,
+      "loss": 0.3322,
+      "step": 926
+    },
+    {
+      "epoch": 2.0565723793677204,
+      "grad_norm": 1.8001863956451416,
+      "learning_rate": 1.740740740740741e-05,
+      "loss": 0.3277,
+      "step": 927
+    },
+    {
+      "epoch": 2.0587909040488075,
+      "grad_norm": 2.646543264389038,
+      "learning_rate": 1.736625514403292e-05,
+      "loss": 0.246,
+      "step": 928
+    },
+    {
+      "epoch": 2.0610094287298946,
+      "grad_norm": 2.057410478591919,
+      "learning_rate": 1.7325102880658437e-05,
+      "loss": 0.2992,
+      "step": 929
+    },
+    {
+      "epoch": 2.0632279534109816,
+      "grad_norm": 1.8453617095947266,
+      "learning_rate": 1.728395061728395e-05,
+      "loss": 0.302,
+      "step": 930
+    },
+    {
+      "epoch": 2.0654464780920687,
+      "grad_norm": 2.30307936668396,
+      "learning_rate": 1.7242798353909466e-05,
+      "loss": 0.4441,
+      "step": 931
+    },
+    {
+      "epoch": 2.067665002773156,
+      "grad_norm": 1.759211778640747,
+      "learning_rate": 1.720164609053498e-05,
+      "loss": 0.2547,
+      "step": 932
+    },
+    {
+      "epoch": 2.069883527454243,
+      "grad_norm": 1.8802341222763062,
+      "learning_rate": 1.7160493827160494e-05,
+      "loss": 0.335,
+      "step": 933
+    },
+    {
+      "epoch": 2.07210205213533,
+      "grad_norm": 2.7769479751586914,
+      "learning_rate": 1.711934156378601e-05,
+      "loss": 0.2965,
+      "step": 934
+    },
+    {
+      "epoch": 2.074320576816417,
+      "grad_norm": 1.8168123960494995,
+      "learning_rate": 1.7078189300411522e-05,
+      "loss": 0.2441,
+      "step": 935
+    },
+    {
+      "epoch": 2.076539101497504,
+      "grad_norm": 2.5596420764923096,
+      "learning_rate": 1.7037037037037038e-05,
+      "loss": 0.3567,
+      "step": 936
+    },
+    {
+      "epoch": 2.078757626178591,
+      "grad_norm": 2.1905336380004883,
+      "learning_rate": 1.699588477366255e-05,
+      "loss": 0.3142,
+      "step": 937
+    },
+    {
+      "epoch": 2.0809761508596782,
+      "grad_norm": 2.0169525146484375,
+      "learning_rate": 1.6954732510288067e-05,
+      "loss": 0.2529,
+      "step": 938
+    },
+    {
+      "epoch": 2.0831946755407653,
+      "grad_norm": 1.856572151184082,
+      "learning_rate": 1.6913580246913582e-05,
+      "loss": 0.2455,
+      "step": 939
+    },
+    {
+      "epoch": 2.0854132002218524,
+      "grad_norm": 1.9868372678756714,
+      "learning_rate": 1.6872427983539095e-05,
+      "loss": 0.2817,
+      "step": 940
+    },
+    {
+      "epoch": 2.0876317249029395,
+      "grad_norm": 2.2969791889190674,
+      "learning_rate": 1.683127572016461e-05,
+      "loss": 0.3072,
+      "step": 941
+    },
+    {
+      "epoch": 2.0898502495840265,
+      "grad_norm": 2.1292519569396973,
+      "learning_rate": 1.6790123456790123e-05,
+      "loss": 0.2551,
+      "step": 942
+    },
+    {
+      "epoch": 2.0920687742651136,
+      "grad_norm": 2.3186488151550293,
+      "learning_rate": 1.674897119341564e-05,
+      "loss": 0.2568,
+      "step": 943
+    },
+    {
+      "epoch": 2.0942872989462007,
+      "grad_norm": 1.943613052368164,
+      "learning_rate": 1.670781893004115e-05,
+      "loss": 0.2535,
+      "step": 944
+    },
+    {
+      "epoch": 2.0965058236272878,
+      "grad_norm": 2.176027774810791,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.3905,
+      "step": 945
+    },
+    {
+      "epoch": 2.098724348308375,
+      "grad_norm": 2.1666107177734375,
+      "learning_rate": 1.662551440329218e-05,
+      "loss": 0.2266,
+      "step": 946
+    },
+    {
+      "epoch": 2.100942872989462,
+      "grad_norm": 2.120121479034424,
+      "learning_rate": 1.6584362139917696e-05,
+      "loss": 0.3026,
+      "step": 947
+    },
+    {
+      "epoch": 2.103161397670549,
+      "grad_norm": 2.0900471210479736,
+      "learning_rate": 1.654320987654321e-05,
+      "loss": 0.262,
+      "step": 948
+    },
+    {
+      "epoch": 2.105379922351636,
+      "grad_norm": 2.304457426071167,
+      "learning_rate": 1.6502057613168724e-05,
+      "loss": 0.3153,
+      "step": 949
+    },
+    {
+      "epoch": 2.107598447032723,
+      "grad_norm": 1.828019380569458,
+      "learning_rate": 1.646090534979424e-05,
+      "loss": 0.2944,
+      "step": 950
+    },
+    {
+      "epoch": 2.10981697171381,
+      "grad_norm": 1.9051620960235596,
+      "learning_rate": 1.6419753086419752e-05,
+      "loss": 0.2534,
+      "step": 951
+    },
+    {
+      "epoch": 2.1120354963948973,
+      "grad_norm": 2.434159278869629,
+      "learning_rate": 1.6378600823045268e-05,
+      "loss": 0.2942,
+      "step": 952
+    },
+    {
+      "epoch": 2.1142540210759844,
+      "grad_norm": 1.933440089225769,
+      "learning_rate": 1.633744855967078e-05,
+      "loss": 0.2616,
+      "step": 953
+    },
+    {
+      "epoch": 2.1164725457570714,
+      "grad_norm": 1.9921914339065552,
+      "learning_rate": 1.62962962962963e-05,
+      "loss": 0.3649,
+      "step": 954
+    },
+    {
+      "epoch": 2.1186910704381585,
+      "grad_norm": 1.8174148797988892,
+      "learning_rate": 1.6255144032921812e-05,
+      "loss": 0.2113,
+      "step": 955
+    },
+    {
+      "epoch": 2.1209095951192456,
+      "grad_norm": 2.906693458557129,
+      "learning_rate": 1.6213991769547325e-05,
+      "loss": 0.3268,
+      "step": 956
+    },
+    {
+      "epoch": 2.1231281198003327,
+      "grad_norm": 2.1203179359436035,
+      "learning_rate": 1.617283950617284e-05,
+      "loss": 0.3948,
+      "step": 957
+    },
+    {
+      "epoch": 2.1253466444814197,
+      "grad_norm": 2.3752427101135254,
+      "learning_rate": 1.6131687242798353e-05,
+      "loss": 0.2607,
+      "step": 958
+    },
+    {
+      "epoch": 2.127565169162507,
+      "grad_norm": 2.027738094329834,
+      "learning_rate": 1.609053497942387e-05,
+      "loss": 0.1981,
+      "step": 959
+    },
+    {
+      "epoch": 2.129783693843594,
+      "grad_norm": 2.155134439468384,
+      "learning_rate": 1.604938271604938e-05,
+      "loss": 0.4048,
+      "step": 960
+    },
+    {
+      "epoch": 2.132002218524681,
+      "grad_norm": 1.917803406715393,
+      "learning_rate": 1.6008230452674897e-05,
+      "loss": 0.2008,
+      "step": 961
+    },
+    {
+      "epoch": 2.134220743205768,
+      "grad_norm": 2.046168804168701,
+      "learning_rate": 1.5967078189300413e-05,
+      "loss": 0.4992,
+      "step": 962
+    },
+    {
+      "epoch": 2.136439267886855,
+      "grad_norm": 1.845747470855713,
+      "learning_rate": 1.5925925925925926e-05,
+      "loss": 0.2735,
+      "step": 963
+    },
+    {
+      "epoch": 2.138657792567942,
+      "grad_norm": 1.7518247365951538,
+      "learning_rate": 1.588477366255144e-05,
+      "loss": 0.2361,
+      "step": 964
+    },
+    {
+      "epoch": 2.1408763172490293,
+      "grad_norm": 2.0311660766601562,
+      "learning_rate": 1.5843621399176954e-05,
+      "loss": 0.3858,
+      "step": 965
+    },
+    {
+      "epoch": 2.1430948419301163,
+      "grad_norm": 1.9792689085006714,
+      "learning_rate": 1.580246913580247e-05,
+      "loss": 0.2878,
+      "step": 966
+    },
+    {
+      "epoch": 2.1453133666112034,
+      "grad_norm": 2.5637118816375732,
+      "learning_rate": 1.5761316872427982e-05,
+      "loss": 0.3197,
+      "step": 967
+    },
+    {
+      "epoch": 2.1475318912922905,
+      "grad_norm": 1.9614261388778687,
+      "learning_rate": 1.5720164609053498e-05,
+      "loss": 0.237,
+      "step": 968
+    },
+    {
+      "epoch": 2.1497504159733776,
+      "grad_norm": 1.9394991397857666,
+      "learning_rate": 1.5679012345679014e-05,
+      "loss": 0.3239,
+      "step": 969
+    },
+    {
+      "epoch": 2.1519689406544646,
+      "grad_norm": 2.0722339153289795,
+      "learning_rate": 1.563786008230453e-05,
+      "loss": 0.3203,
+      "step": 970
+    },
+    {
+      "epoch": 2.1541874653355517,
+      "grad_norm": 1.9220918416976929,
+      "learning_rate": 1.5596707818930042e-05,
+      "loss": 0.186,
+      "step": 971
+    },
+    {
+      "epoch": 2.156405990016639,
+      "grad_norm": 2.311063528060913,
+      "learning_rate": 1.5555555555555555e-05,
+      "loss": 0.2549,
+      "step": 972
+    },
+    {
+      "epoch": 2.158624514697726,
+      "grad_norm": 1.8842520713806152,
+      "learning_rate": 1.551440329218107e-05,
+      "loss": 0.2681,
+      "step": 973
+    },
+    {
+      "epoch": 2.160843039378813,
+      "grad_norm": 2.0722477436065674,
+      "learning_rate": 1.5473251028806583e-05,
+      "loss": 0.3289,
+      "step": 974
+    },
+    {
+      "epoch": 2.1630615640599,
+      "grad_norm": 1.8954784870147705,
+      "learning_rate": 1.54320987654321e-05,
+      "loss": 0.3623,
+      "step": 975
+    },
+    {
+      "epoch": 2.165280088740987,
+      "grad_norm": 2.076784133911133,
+      "learning_rate": 1.539094650205761e-05,
+      "loss": 0.2836,
+      "step": 976
+    },
+    {
+      "epoch": 2.167498613422074,
+      "grad_norm": 1.72483229637146,
+      "learning_rate": 1.534979423868313e-05,
+      "loss": 0.2313,
+      "step": 977
+    },
+    {
+      "epoch": 2.1697171381031612,
+      "grad_norm": 1.7990970611572266,
+      "learning_rate": 1.5308641975308643e-05,
+      "loss": 0.2359,
+      "step": 978
+    },
+    {
+      "epoch": 2.1719356627842483,
+      "grad_norm": 2.604351282119751,
+      "learning_rate": 1.526748971193416e-05,
+      "loss": 0.5566,
+      "step": 979
+    },
+    {
+      "epoch": 2.1741541874653354,
+      "grad_norm": 2.1897687911987305,
+      "learning_rate": 1.5226337448559672e-05,
+      "loss": 0.2901,
+      "step": 980
+    },
+    {
+      "epoch": 2.1763727121464225,
+      "grad_norm": 1.8529057502746582,
+      "learning_rate": 1.5185185185185186e-05,
+      "loss": 0.2149,
+      "step": 981
+    },
+    {
+      "epoch": 2.1785912368275095,
+      "grad_norm": 1.777066707611084,
+      "learning_rate": 1.51440329218107e-05,
+      "loss": 0.215,
+      "step": 982
+    },
+    {
+      "epoch": 2.1808097615085966,
+      "grad_norm": 2.7177364826202393,
+      "learning_rate": 1.5102880658436214e-05,
+      "loss": 0.4134,
+      "step": 983
+    },
+    {
+      "epoch": 2.1830282861896837,
+      "grad_norm": 3.1439621448516846,
+      "learning_rate": 1.506172839506173e-05,
+      "loss": 0.2827,
+      "step": 984
+    },
+    {
+      "epoch": 2.1852468108707708,
+      "grad_norm": 2.0024325847625732,
+      "learning_rate": 1.5020576131687244e-05,
+      "loss": 0.256,
+      "step": 985
+    },
+    {
+      "epoch": 2.187465335551858,
+      "grad_norm": 1.835290789604187,
+      "learning_rate": 1.4979423868312758e-05,
+      "loss": 0.3087,
+      "step": 986
+    },
+    {
+      "epoch": 2.189683860232945,
+      "grad_norm": 1.8491712808609009,
+      "learning_rate": 1.4938271604938272e-05,
+      "loss": 0.1991,
+      "step": 987
+    },
+    {
+      "epoch": 2.191902384914032,
+      "grad_norm": 2.4228475093841553,
+      "learning_rate": 1.4897119341563787e-05,
+      "loss": 0.1858,
+      "step": 988
+    },
+    {
+      "epoch": 2.194120909595119,
+      "grad_norm": 1.8492109775543213,
+      "learning_rate": 1.48559670781893e-05,
+      "loss": 0.2023,
+      "step": 989
+    },
+    {
+      "epoch": 2.196339434276206,
+      "grad_norm": 2.1545073986053467,
+      "learning_rate": 1.4814814814814815e-05,
+      "loss": 0.2055,
+      "step": 990
+    },
+    {
+      "epoch": 2.1985579589572932,
+      "grad_norm": 2.01643967628479,
+      "learning_rate": 1.4773662551440329e-05,
+      "loss": 0.2132,
+      "step": 991
+    },
+    {
+      "epoch": 2.2007764836383803,
+      "grad_norm": 1.647180199623108,
+      "learning_rate": 1.4732510288065845e-05,
+      "loss": 0.2536,
+      "step": 992
+    },
+    {
+      "epoch": 2.2029950083194674,
+      "grad_norm": 2.725740909576416,
+      "learning_rate": 1.4691358024691359e-05,
+      "loss": 0.5276,
+      "step": 993
+    },
+    {
+      "epoch": 2.2052135330005544,
+      "grad_norm": 1.853236436843872,
+      "learning_rate": 1.4650205761316873e-05,
+      "loss": 0.1208,
+      "step": 994
+    },
+    {
+      "epoch": 2.2074320576816415,
+      "grad_norm": 2.1177146434783936,
+      "learning_rate": 1.4609053497942387e-05,
+      "loss": 0.2338,
+      "step": 995
+    },
+    {
+      "epoch": 2.2096505823627286,
+      "grad_norm": 3.1426072120666504,
+      "learning_rate": 1.4567901234567902e-05,
+      "loss": 0.2563,
+      "step": 996
+    },
+    {
+      "epoch": 2.2118691070438157,
+      "grad_norm": 2.189643144607544,
+      "learning_rate": 1.4526748971193416e-05,
+      "loss": 0.2643,
+      "step": 997
+    },
+    {
+      "epoch": 2.2140876317249027,
+      "grad_norm": 2.056523084640503,
+      "learning_rate": 1.448559670781893e-05,
+      "loss": 0.2493,
+      "step": 998
+    },
+    {
+      "epoch": 2.21630615640599,
+      "grad_norm": 2.7572882175445557,
+      "learning_rate": 1.4444444444444444e-05,
+      "loss": 0.2069,
+      "step": 999
+    },
+    {
+      "epoch": 2.218524681087077,
+      "grad_norm": 2.0325121879577637,
+      "learning_rate": 1.440329218106996e-05,
+      "loss": 0.2234,
+      "step": 1000
+    },
+    {
+      "epoch": 2.220743205768164,
+      "grad_norm": 3.5439984798431396,
+      "learning_rate": 1.4362139917695474e-05,
+      "loss": 0.2955,
+      "step": 1001
+    },
+    {
+      "epoch": 2.222961730449251,
+      "grad_norm": 1.9860895872116089,
+      "learning_rate": 1.4320987654320988e-05,
+      "loss": 0.3912,
+      "step": 1002
+    },
+    {
+      "epoch": 2.225180255130338,
+      "grad_norm": 2.1854023933410645,
+      "learning_rate": 1.4279835390946502e-05,
+      "loss": 0.3151,
+      "step": 1003
+    },
+    {
+      "epoch": 2.227398779811425,
+      "grad_norm": 1.5809242725372314,
+      "learning_rate": 1.4238683127572017e-05,
+      "loss": 0.1834,
+      "step": 1004
+    },
+    {
+      "epoch": 2.2296173044925123,
+      "grad_norm": 2.717968463897705,
+      "learning_rate": 1.419753086419753e-05,
+      "loss": 0.3497,
+      "step": 1005
+    },
+    {
+      "epoch": 2.2318358291735994,
+      "grad_norm": 3.0287833213806152,
+      "learning_rate": 1.4156378600823045e-05,
+      "loss": 0.3089,
+      "step": 1006
+    },
+    {
+      "epoch": 2.2340543538546864,
+      "grad_norm": 2.352271556854248,
+      "learning_rate": 1.411522633744856e-05,
+      "loss": 0.248,
+      "step": 1007
+    },
+    {
+      "epoch": 2.2362728785357735,
+      "grad_norm": 2.320093870162964,
+      "learning_rate": 1.4074074074074075e-05,
+      "loss": 0.3595,
+      "step": 1008
+    },
+    {
+      "epoch": 2.2384914032168606,
+      "grad_norm": 1.760396957397461,
+      "learning_rate": 1.4032921810699589e-05,
+      "loss": 0.3612,
+      "step": 1009
+    },
+    {
+      "epoch": 2.2407099278979477,
+      "grad_norm": 1.8629406690597534,
+      "learning_rate": 1.3991769547325103e-05,
+      "loss": 0.3022,
+      "step": 1010
+    },
+    {
+      "epoch": 2.2429284525790347,
+      "grad_norm": 1.9859365224838257,
+      "learning_rate": 1.3950617283950617e-05,
+      "loss": 0.2364,
+      "step": 1011
+    },
+    {
+      "epoch": 2.245146977260122,
+      "grad_norm": 1.6879619359970093,
+      "learning_rate": 1.3909465020576132e-05,
+      "loss": 0.2738,
+      "step": 1012
+    },
+    {
+      "epoch": 2.247365501941209,
+      "grad_norm": 1.8565163612365723,
+      "learning_rate": 1.3868312757201646e-05,
+      "loss": 0.2701,
+      "step": 1013
+    },
+    {
+      "epoch": 2.249584026622296,
+      "grad_norm": 2.0068273544311523,
+      "learning_rate": 1.382716049382716e-05,
+      "loss": 0.3036,
+      "step": 1014
+    },
+    {
+      "epoch": 2.251802551303383,
+      "grad_norm": 1.9134773015975952,
+      "learning_rate": 1.3786008230452676e-05,
+      "loss": 0.3478,
+      "step": 1015
+    },
+    {
+      "epoch": 2.25402107598447,
+      "grad_norm": 1.8970298767089844,
+      "learning_rate": 1.374485596707819e-05,
+      "loss": 0.2888,
+      "step": 1016
+    },
+    {
+      "epoch": 2.256239600665557,
+      "grad_norm": 1.8005971908569336,
+      "learning_rate": 1.3703703703703704e-05,
+      "loss": 0.4605,
+      "step": 1017
+    },
+    {
+      "epoch": 2.2584581253466443,
+      "grad_norm": 2.723443031311035,
+      "learning_rate": 1.3662551440329218e-05,
+      "loss": 0.461,
+      "step": 1018
+    },
+    {
+      "epoch": 2.2606766500277313,
+      "grad_norm": 2.1651062965393066,
+      "learning_rate": 1.3621399176954732e-05,
+      "loss": 0.3674,
+      "step": 1019
+    },
+    {
+      "epoch": 2.2628951747088184,
+      "grad_norm": 2.2053840160369873,
+      "learning_rate": 1.3580246913580247e-05,
+      "loss": 0.41,
+      "step": 1020
+    },
+    {
+      "epoch": 2.2651136993899055,
+      "grad_norm": 2.4134435653686523,
+      "learning_rate": 1.353909465020576e-05,
+      "loss": 0.2861,
+      "step": 1021
+    },
+    {
+      "epoch": 2.2673322240709926,
+      "grad_norm": 2.266929864883423,
+      "learning_rate": 1.3497942386831278e-05,
+      "loss": 0.1981,
+      "step": 1022
+    },
+    {
+      "epoch": 2.2695507487520796,
+      "grad_norm": 2.2010092735290527,
+      "learning_rate": 1.3456790123456793e-05,
+      "loss": 0.2707,
+      "step": 1023
+    },
+    {
+      "epoch": 2.2717692734331667,
+      "grad_norm": 1.9794020652770996,
+      "learning_rate": 1.3415637860082305e-05,
+      "loss": 0.2013,
+      "step": 1024
+    },
+    {
+      "epoch": 2.2739877981142538,
+      "grad_norm": 1.7918622493743896,
+      "learning_rate": 1.3374485596707819e-05,
+      "loss": 0.3686,
+      "step": 1025
+    },
+    {
+      "epoch": 2.276206322795341,
+      "grad_norm": 2.400038003921509,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.1697,
+      "step": 1026
+    },
+    {
+      "epoch": 2.278424847476428,
+      "grad_norm": 2.077394962310791,
+      "learning_rate": 1.3292181069958847e-05,
+      "loss": 0.3135,
+      "step": 1027
+    },
+    {
+      "epoch": 2.280643372157515,
+      "grad_norm": 2.4591562747955322,
+      "learning_rate": 1.3251028806584362e-05,
+      "loss": 0.2076,
+      "step": 1028
+    },
+    {
+      "epoch": 2.2828618968386025,
+      "grad_norm": 1.9759244918823242,
+      "learning_rate": 1.3209876543209876e-05,
+      "loss": 0.3174,
+      "step": 1029
+    },
+    {
+      "epoch": 2.2850804215196896,
+      "grad_norm": 2.3431131839752197,
+      "learning_rate": 1.3168724279835393e-05,
+      "loss": 0.3327,
+      "step": 1030
+    },
+    {
+      "epoch": 2.2872989462007767,
+      "grad_norm": 2.1458325386047363,
+      "learning_rate": 1.3127572016460907e-05,
+      "loss": 0.2378,
+      "step": 1031
+    },
+    {
+      "epoch": 2.2895174708818637,
+      "grad_norm": 2.2659451961517334,
+      "learning_rate": 1.308641975308642e-05,
+      "loss": 0.3283,
+      "step": 1032
+    },
+    {
+      "epoch": 2.291735995562951,
+      "grad_norm": 1.693854570388794,
+      "learning_rate": 1.3045267489711934e-05,
+      "loss": 0.1923,
+      "step": 1033
+    },
+    {
+      "epoch": 2.293954520244038,
+      "grad_norm": 2.6155357360839844,
+      "learning_rate": 1.3004115226337448e-05,
+      "loss": 0.253,
+      "step": 1034
+    },
+    {
+      "epoch": 2.296173044925125,
+      "grad_norm": 1.8327398300170898,
+      "learning_rate": 1.2962962962962962e-05,
+      "loss": 0.2346,
+      "step": 1035
+    },
+    {
+      "epoch": 2.298391569606212,
+      "grad_norm": 1.925473928451538,
+      "learning_rate": 1.2921810699588477e-05,
+      "loss": 0.2803,
+      "step": 1036
+    },
+    {
+      "epoch": 2.300610094287299,
+      "grad_norm": 1.7880765199661255,
+      "learning_rate": 1.2880658436213994e-05,
+      "loss": 0.2341,
+      "step": 1037
+    },
+    {
+      "epoch": 2.302828618968386,
+      "grad_norm": 2.33659291267395,
+      "learning_rate": 1.2839506172839508e-05,
+      "loss": 0.4249,
+      "step": 1038
+    },
+    {
+      "epoch": 2.3050471436494733,
+      "grad_norm": 2.0593621730804443,
+      "learning_rate": 1.2798353909465022e-05,
+      "loss": 0.283,
+      "step": 1039
+    },
+    {
+      "epoch": 2.3072656683305603,
+      "grad_norm": 1.8160078525543213,
+      "learning_rate": 1.2757201646090535e-05,
+      "loss": 0.2896,
+      "step": 1040
+    },
+    {
+      "epoch": 2.3094841930116474,
+      "grad_norm": 2.1237878799438477,
+      "learning_rate": 1.2716049382716049e-05,
+      "loss": 0.4059,
+      "step": 1041
+    },
+    {
+      "epoch": 2.3117027176927345,
+      "grad_norm": 2.0628535747528076,
+      "learning_rate": 1.2674897119341563e-05,
+      "loss": 0.2464,
+      "step": 1042
+    },
+    {
+      "epoch": 2.3139212423738216,
+      "grad_norm": 2.153855800628662,
+      "learning_rate": 1.2633744855967077e-05,
+      "loss": 0.2368,
+      "step": 1043
+    },
+    {
+      "epoch": 2.3161397670549086,
+      "grad_norm": 2.2237131595611572,
+      "learning_rate": 1.2592592592592592e-05,
+      "loss": 0.3176,
+      "step": 1044
+    },
+    {
+      "epoch": 2.3183582917359957,
+      "grad_norm": 1.823378562927246,
+      "learning_rate": 1.255144032921811e-05,
+      "loss": 0.2566,
+      "step": 1045
+    },
+    {
+      "epoch": 2.320576816417083,
+      "grad_norm": 2.092829942703247,
+      "learning_rate": 1.2510288065843623e-05,
+      "loss": 0.2823,
+      "step": 1046
+    },
+    {
+      "epoch": 2.32279534109817,
+      "grad_norm": 2.187436819076538,
+      "learning_rate": 1.2469135802469137e-05,
+      "loss": 0.2828,
+      "step": 1047
+    },
+    {
+      "epoch": 2.325013865779257,
+      "grad_norm": 2.2013797760009766,
+      "learning_rate": 1.2427983539094652e-05,
+      "loss": 0.315,
+      "step": 1048
+    },
+    {
+      "epoch": 2.327232390460344,
+      "grad_norm": 2.749352216720581,
+      "learning_rate": 1.2386831275720164e-05,
+      "loss": 0.287,
+      "step": 1049
+    },
+    {
+      "epoch": 2.329450915141431,
+      "grad_norm": 2.638491153717041,
+      "learning_rate": 1.2345679012345678e-05,
+      "loss": 0.2764,
+      "step": 1050
+    },
+    {
+      "epoch": 2.331669439822518,
+      "grad_norm": 2.3354508876800537,
+      "learning_rate": 1.2304526748971194e-05,
+      "loss": 0.2831,
+      "step": 1051
+    },
+    {
+      "epoch": 2.3338879645036053,
+      "grad_norm": 2.3840153217315674,
+      "learning_rate": 1.2263374485596708e-05,
+      "loss": 0.1756,
+      "step": 1052
+    },
+    {
+      "epoch": 2.3361064891846923,
+      "grad_norm": 2.46549391746521,
+      "learning_rate": 1.2222222222222222e-05,
+      "loss": 0.3982,
+      "step": 1053
+    },
+    {
+      "epoch": 2.3383250138657794,
+      "grad_norm": 2.558627128601074,
+      "learning_rate": 1.2181069958847738e-05,
+      "loss": 0.2539,
+      "step": 1054
+    },
+    {
+      "epoch": 2.3405435385468665,
+      "grad_norm": 1.7755157947540283,
+      "learning_rate": 1.2139917695473252e-05,
+      "loss": 0.3147,
+      "step": 1055
+    },
+    {
+      "epoch": 2.3427620632279536,
+      "grad_norm": 2.2119107246398926,
+      "learning_rate": 1.2098765432098767e-05,
+      "loss": 0.3604,
+      "step": 1056
+    },
+    {
+      "epoch": 2.3449805879090406,
+      "grad_norm": 1.9261703491210938,
+      "learning_rate": 1.2057613168724279e-05,
+      "loss": 0.2131,
+      "step": 1057
+    },
+    {
+      "epoch": 2.3471991125901277,
+      "grad_norm": 2.3002161979675293,
+      "learning_rate": 1.2016460905349795e-05,
+      "loss": 0.2993,
+      "step": 1058
+    },
+    {
+      "epoch": 2.3494176372712148,
+      "grad_norm": 2.7202792167663574,
+      "learning_rate": 1.1975308641975309e-05,
+      "loss": 0.2613,
+      "step": 1059
+    },
+    {
+      "epoch": 2.351636161952302,
+      "grad_norm": 2.1841354370117188,
+      "learning_rate": 1.1934156378600823e-05,
+      "loss": 0.2397,
+      "step": 1060
+    },
+    {
+      "epoch": 2.353854686633389,
+      "grad_norm": 2.695584774017334,
+      "learning_rate": 1.1893004115226337e-05,
+      "loss": 0.2907,
+      "step": 1061
+    },
+    {
+      "epoch": 2.356073211314476,
+      "grad_norm": 2.078781843185425,
+      "learning_rate": 1.1851851851851853e-05,
+      "loss": 0.3042,
+      "step": 1062
+    },
+    {
+      "epoch": 2.358291735995563,
+      "grad_norm": 1.9624292850494385,
+      "learning_rate": 1.1810699588477367e-05,
+      "loss": 0.2471,
+      "step": 1063
+    },
+    {
+      "epoch": 2.36051026067665,
+      "grad_norm": 2.8841054439544678,
+      "learning_rate": 1.1769547325102882e-05,
+      "loss": 0.3181,
+      "step": 1064
+    },
+    {
+      "epoch": 2.3627287853577372,
+      "grad_norm": 1.9062873125076294,
+      "learning_rate": 1.1728395061728396e-05,
+      "loss": 0.1978,
+      "step": 1065
+    },
+    {
+      "epoch": 2.3649473100388243,
+      "grad_norm": 2.0245625972747803,
+      "learning_rate": 1.168724279835391e-05,
+      "loss": 0.2939,
+      "step": 1066
+    },
+    {
+      "epoch": 2.3671658347199114,
+      "grad_norm": 2.173107385635376,
+      "learning_rate": 1.1646090534979424e-05,
+      "loss": 0.3284,
+      "step": 1067
+    },
+    {
+      "epoch": 2.3693843594009985,
+      "grad_norm": 2.556859016418457,
+      "learning_rate": 1.1604938271604938e-05,
+      "loss": 0.3204,
+      "step": 1068
+    },
+    {
+      "epoch": 2.3716028840820855,
+      "grad_norm": 2.7206649780273438,
+      "learning_rate": 1.1563786008230454e-05,
+      "loss": 0.284,
+      "step": 1069
+    },
+    {
+      "epoch": 2.3738214087631726,
+      "grad_norm": 1.9049268960952759,
+      "learning_rate": 1.1522633744855968e-05,
+      "loss": 0.2459,
+      "step": 1070
+    },
+    {
+      "epoch": 2.3760399334442597,
+      "grad_norm": 2.269386053085327,
+      "learning_rate": 1.1481481481481482e-05,
+      "loss": 0.3702,
+      "step": 1071
+    },
+    {
+      "epoch": 2.3782584581253468,
+      "grad_norm": 2.250969409942627,
+      "learning_rate": 1.1440329218106997e-05,
+      "loss": 0.3365,
+      "step": 1072
+    },
+    {
+      "epoch": 2.380476982806434,
+      "grad_norm": 1.9424755573272705,
+      "learning_rate": 1.139917695473251e-05,
+      "loss": 0.3616,
+      "step": 1073
+    },
+    {
+      "epoch": 2.382695507487521,
+      "grad_norm": 2.3486974239349365,
+      "learning_rate": 1.1358024691358025e-05,
+      "loss": 0.3506,
+      "step": 1074
+    },
+    {
+      "epoch": 2.384914032168608,
+      "grad_norm": 2.0541155338287354,
+      "learning_rate": 1.1316872427983539e-05,
+      "loss": 0.2076,
+      "step": 1075
+    },
+    {
+      "epoch": 2.387132556849695,
+      "grad_norm": 1.9601181745529175,
+      "learning_rate": 1.1275720164609053e-05,
+      "loss": 0.1843,
+      "step": 1076
+    },
+    {
+      "epoch": 2.389351081530782,
+      "grad_norm": 2.027656078338623,
+      "learning_rate": 1.123456790123457e-05,
+      "loss": 0.3774,
+      "step": 1077
+    },
+    {
+      "epoch": 2.391569606211869,
+      "grad_norm": 2.1897811889648438,
+      "learning_rate": 1.1193415637860083e-05,
+      "loss": 0.3902,
+      "step": 1078
+    },
+    {
+      "epoch": 2.3937881308929563,
+      "grad_norm": 2.4011943340301514,
+      "learning_rate": 1.1152263374485597e-05,
+      "loss": 0.2586,
+      "step": 1079
+    },
+    {
+      "epoch": 2.3960066555740434,
+      "grad_norm": 1.8466497659683228,
+      "learning_rate": 1.1111111111111112e-05,
+      "loss": 0.2282,
+      "step": 1080
+    },
+    {
+      "epoch": 2.3982251802551304,
+      "grad_norm": 2.520879030227661,
+      "learning_rate": 1.1069958847736626e-05,
+      "loss": 0.298,
+      "step": 1081
+    },
+    {
+      "epoch": 2.4004437049362175,
+      "grad_norm": 2.1769232749938965,
+      "learning_rate": 1.102880658436214e-05,
+      "loss": 0.2707,
+      "step": 1082
+    },
+    {
+      "epoch": 2.4026622296173046,
+      "grad_norm": 1.8492012023925781,
+      "learning_rate": 1.0987654320987654e-05,
+      "loss": 0.3888,
+      "step": 1083
+    },
+    {
+      "epoch": 2.4048807542983917,
+      "grad_norm": 3.0312764644622803,
+      "learning_rate": 1.094650205761317e-05,
+      "loss": 0.2528,
+      "step": 1084
+    },
+    {
+      "epoch": 2.4070992789794787,
+      "grad_norm": 1.8016606569290161,
+      "learning_rate": 1.0905349794238684e-05,
+      "loss": 0.2223,
+      "step": 1085
+    },
+    {
+      "epoch": 2.409317803660566,
+      "grad_norm": 1.9718635082244873,
+      "learning_rate": 1.0864197530864198e-05,
+      "loss": 0.2485,
+      "step": 1086
+    },
+    {
+      "epoch": 2.411536328341653,
+      "grad_norm": 2.144784688949585,
+      "learning_rate": 1.0823045267489712e-05,
+      "loss": 0.2309,
+      "step": 1087
+    },
+    {
+      "epoch": 2.41375485302274,
+      "grad_norm": 2.1952500343322754,
+      "learning_rate": 1.0781893004115227e-05,
+      "loss": 0.2332,
+      "step": 1088
+    },
+    {
+      "epoch": 2.415973377703827,
+      "grad_norm": 2.1626763343811035,
+      "learning_rate": 1.074074074074074e-05,
+      "loss": 0.2588,
+      "step": 1089
+    },
+    {
+      "epoch": 2.418191902384914,
+      "grad_norm": 2.163473606109619,
+      "learning_rate": 1.0699588477366255e-05,
+      "loss": 0.3096,
+      "step": 1090
+    },
+    {
+      "epoch": 2.420410427066001,
+      "grad_norm": 2.2118074893951416,
+      "learning_rate": 1.0658436213991769e-05,
+      "loss": 0.3257,
+      "step": 1091
+    },
+    {
+      "epoch": 2.4226289517470883,
+      "grad_norm": 1.9706759452819824,
+      "learning_rate": 1.0617283950617285e-05,
+      "loss": 0.2495,
+      "step": 1092
+    },
+    {
+      "epoch": 2.4248474764281753,
+      "grad_norm": 1.8772380352020264,
+      "learning_rate": 1.0576131687242799e-05,
+      "loss": 0.2506,
+      "step": 1093
+    },
+    {
+      "epoch": 2.4270660011092624,
+      "grad_norm": 2.364847183227539,
+      "learning_rate": 1.0534979423868313e-05,
+      "loss": 0.3128,
+      "step": 1094
+    },
+    {
+      "epoch": 2.4292845257903495,
+      "grad_norm": 2.465916633605957,
+      "learning_rate": 1.0493827160493827e-05,
+      "loss": 0.2484,
+      "step": 1095
+    },
+    {
+      "epoch": 2.4315030504714366,
+      "grad_norm": 2.1951186656951904,
+      "learning_rate": 1.0452674897119343e-05,
+      "loss": 0.3583,
+      "step": 1096
+    },
+    {
+      "epoch": 2.4337215751525236,
+      "grad_norm": 1.8116748332977295,
+      "learning_rate": 1.0411522633744856e-05,
+      "loss": 0.3018,
+      "step": 1097
+    },
+    {
+      "epoch": 2.4359400998336107,
+      "grad_norm": 2.3656389713287354,
+      "learning_rate": 1.037037037037037e-05,
+      "loss": 0.2796,
+      "step": 1098
+    },
+    {
+      "epoch": 2.438158624514698,
+      "grad_norm": 1.848024845123291,
+      "learning_rate": 1.0329218106995884e-05,
+      "loss": 0.2426,
+      "step": 1099
+    },
+    {
+      "epoch": 2.440377149195785,
+      "grad_norm": 1.8439791202545166,
+      "learning_rate": 1.02880658436214e-05,
+      "loss": 0.261,
+      "step": 1100
+    },
+    {
+      "epoch": 2.442595673876872,
+      "grad_norm": 2.4032094478607178,
+      "learning_rate": 1.0246913580246914e-05,
+      "loss": 0.2133,
+      "step": 1101
+    },
+    {
+      "epoch": 2.444814198557959,
+      "grad_norm": 1.9726359844207764,
+      "learning_rate": 1.0205761316872428e-05,
+      "loss": 0.3042,
+      "step": 1102
+    },
+    {
+      "epoch": 2.447032723239046,
+      "grad_norm": 2.1109912395477295,
+      "learning_rate": 1.0164609053497944e-05,
+      "loss": 0.3005,
+      "step": 1103
+    },
+    {
+      "epoch": 2.449251247920133,
+      "grad_norm": 2.0022130012512207,
+      "learning_rate": 1.0123456790123458e-05,
+      "loss": 0.1828,
+      "step": 1104
+    },
+    {
+      "epoch": 2.4514697726012202,
+      "grad_norm": 2.2652170658111572,
+      "learning_rate": 1.008230452674897e-05,
+      "loss": 0.2775,
+      "step": 1105
+    },
+    {
+      "epoch": 2.4536882972823073,
+      "grad_norm": 2.101163387298584,
+      "learning_rate": 1.0041152263374485e-05,
+      "loss": 0.2531,
+      "step": 1106
+    },
+    {
+      "epoch": 2.4559068219633944,
+      "grad_norm": 2.183821201324463,
+      "learning_rate": 1e-05,
+      "loss": 0.25,
+      "step": 1107
+    },
+    {
+      "epoch": 2.4581253466444815,
+      "grad_norm": 2.1730072498321533,
+      "learning_rate": 9.958847736625515e-06,
+      "loss": 0.2257,
+      "step": 1108
+    },
+    {
+      "epoch": 2.4603438713255685,
+      "grad_norm": 1.8650232553482056,
+      "learning_rate": 9.917695473251029e-06,
+      "loss": 0.1525,
+      "step": 1109
+    },
+    {
+      "epoch": 2.4625623960066556,
+      "grad_norm": 2.4648489952087402,
+      "learning_rate": 9.876543209876543e-06,
+      "loss": 0.2327,
+      "step": 1110
+    },
+    {
+      "epoch": 2.4647809206877427,
+      "grad_norm": 2.3710527420043945,
+      "learning_rate": 9.83539094650206e-06,
+      "loss": 0.1934,
+      "step": 1111
+    },
+    {
+      "epoch": 2.4669994453688298,
+      "grad_norm": 2.130769968032837,
+      "learning_rate": 9.794238683127573e-06,
+      "loss": 0.2679,
+      "step": 1112
+    },
+    {
+      "epoch": 2.469217970049917,
+      "grad_norm": 1.8794821500778198,
+      "learning_rate": 9.753086419753086e-06,
+      "loss": 0.2548,
+      "step": 1113
+    },
+    {
+      "epoch": 2.471436494731004,
+      "grad_norm": 2.2746570110321045,
+      "learning_rate": 9.7119341563786e-06,
+      "loss": 0.2867,
+      "step": 1114
+    },
+    {
+      "epoch": 2.473655019412091,
+      "grad_norm": 2.5124692916870117,
+      "learning_rate": 9.670781893004116e-06,
+      "loss": 0.2256,
+      "step": 1115
+    },
+    {
+      "epoch": 2.475873544093178,
+      "grad_norm": 2.647364854812622,
+      "learning_rate": 9.62962962962963e-06,
+      "loss": 0.2484,
+      "step": 1116
+    },
+    {
+      "epoch": 2.478092068774265,
+      "grad_norm": 1.9132670164108276,
+      "learning_rate": 9.588477366255144e-06,
+      "loss": 0.3008,
+      "step": 1117
+    },
+    {
+      "epoch": 2.480310593455352,
+      "grad_norm": 1.867000699043274,
+      "learning_rate": 9.54732510288066e-06,
+      "loss": 0.2494,
+      "step": 1118
+    },
+    {
+      "epoch": 2.4825291181364393,
+      "grad_norm": 2.540458917617798,
+      "learning_rate": 9.506172839506174e-06,
+      "loss": 0.2958,
+      "step": 1119
+    },
+    {
+      "epoch": 2.4847476428175264,
+      "grad_norm": 2.183011531829834,
+      "learning_rate": 9.465020576131688e-06,
+      "loss": 0.3346,
+      "step": 1120
+    },
+    {
+      "epoch": 2.4869661674986134,
+      "grad_norm": 2.0224621295928955,
+      "learning_rate": 9.423868312757202e-06,
+      "loss": 0.1999,
+      "step": 1121
+    },
+    {
+      "epoch": 2.4891846921797005,
+      "grad_norm": 2.3942861557006836,
+      "learning_rate": 9.382716049382717e-06,
+      "loss": 0.2591,
+      "step": 1122
+    },
+    {
+      "epoch": 2.4914032168607876,
+      "grad_norm": 2.0364859104156494,
+      "learning_rate": 9.34156378600823e-06,
+      "loss": 0.2968,
+      "step": 1123
+    },
+    {
+      "epoch": 2.4936217415418747,
+      "grad_norm": 1.9936282634735107,
+      "learning_rate": 9.300411522633745e-06,
+      "loss": 0.3053,
+      "step": 1124
+    },
+    {
+      "epoch": 2.4958402662229617,
+      "grad_norm": 2.0221197605133057,
+      "learning_rate": 9.259259259259259e-06,
+      "loss": 0.4268,
+      "step": 1125
+    },
+    {
+      "epoch": 2.498058790904049,
+      "grad_norm": 1.8707597255706787,
+      "learning_rate": 9.218106995884775e-06,
+      "loss": 0.2906,
+      "step": 1126
+    },
+    {
+      "epoch": 2.500277315585136,
+      "grad_norm": 1.518580436706543,
+      "learning_rate": 9.17695473251029e-06,
+      "loss": 0.2196,
+      "step": 1127
+    },
+    {
+      "epoch": 2.502495840266223,
+      "grad_norm": 1.6165120601654053,
+      "learning_rate": 9.135802469135803e-06,
+      "loss": 0.3108,
+      "step": 1128
+    },
+    {
+      "epoch": 2.50471436494731,
+      "grad_norm": 2.0797204971313477,
+      "learning_rate": 9.094650205761317e-06,
+      "loss": 0.3116,
+      "step": 1129
+    },
+    {
+      "epoch": 2.506932889628397,
+      "grad_norm": 1.963379979133606,
+      "learning_rate": 9.053497942386832e-06,
+      "loss": 0.2461,
+      "step": 1130
+    },
+    {
+      "epoch": 2.509151414309484,
+      "grad_norm": 2.042454481124878,
+      "learning_rate": 9.012345679012346e-06,
+      "loss": 0.2313,
+      "step": 1131
+    },
+    {
+      "epoch": 2.5113699389905713,
+      "grad_norm": 2.6683237552642822,
+      "learning_rate": 8.97119341563786e-06,
+      "loss": 0.3333,
+      "step": 1132
+    },
+    {
+      "epoch": 2.5135884636716583,
+      "grad_norm": 2.5000858306884766,
+      "learning_rate": 8.930041152263374e-06,
+      "loss": 0.2434,
+      "step": 1133
+    },
+    {
+      "epoch": 2.5158069883527454,
+      "grad_norm": 2.219658374786377,
+      "learning_rate": 8.88888888888889e-06,
+      "loss": 0.3688,
+      "step": 1134
+    },
+    {
+      "epoch": 2.5180255130338325,
+      "grad_norm": 1.9160106182098389,
+      "learning_rate": 8.847736625514404e-06,
+      "loss": 0.1949,
+      "step": 1135
+    },
+    {
+      "epoch": 2.5202440377149196,
+      "grad_norm": 2.5271098613739014,
+      "learning_rate": 8.806584362139918e-06,
+      "loss": 0.2998,
+      "step": 1136
+    },
+    {
+      "epoch": 2.5224625623960066,
+      "grad_norm": 2.20500111579895,
+      "learning_rate": 8.765432098765432e-06,
+      "loss": 0.2655,
+      "step": 1137
+    },
+    {
+      "epoch": 2.5246810870770937,
+      "grad_norm": 2.1261844635009766,
+      "learning_rate": 8.724279835390947e-06,
+      "loss": 0.2767,
+      "step": 1138
+    },
+    {
+      "epoch": 2.526899611758181,
+      "grad_norm": 2.715569257736206,
+      "learning_rate": 8.68312757201646e-06,
+      "loss": 0.3259,
+      "step": 1139
+    },
+    {
+      "epoch": 2.529118136439268,
+      "grad_norm": 2.2820847034454346,
+      "learning_rate": 8.641975308641975e-06,
+      "loss": 0.2806,
+      "step": 1140
+    },
+    {
+      "epoch": 2.531336661120355,
+      "grad_norm": 2.1151108741760254,
+      "learning_rate": 8.60082304526749e-06,
+      "loss": 0.2184,
+      "step": 1141
+    },
+    {
+      "epoch": 2.533555185801442,
+      "grad_norm": 2.131166934967041,
+      "learning_rate": 8.559670781893005e-06,
+      "loss": 0.2362,
+      "step": 1142
+    },
+    {
+      "epoch": 2.535773710482529,
+      "grad_norm": 2.415280818939209,
+      "learning_rate": 8.518518518518519e-06,
+      "loss": 0.3311,
+      "step": 1143
+    },
+    {
+      "epoch": 2.537992235163616,
+      "grad_norm": 2.4645557403564453,
+      "learning_rate": 8.477366255144033e-06,
+      "loss": 0.5671,
+      "step": 1144
+    },
+    {
+      "epoch": 2.5402107598447032,
+      "grad_norm": 1.9307013750076294,
+      "learning_rate": 8.436213991769547e-06,
+      "loss": 0.2279,
+      "step": 1145
+    },
+    {
+      "epoch": 2.5424292845257903,
+      "grad_norm": 2.2439448833465576,
+      "learning_rate": 8.395061728395062e-06,
+      "loss": 0.2979,
+      "step": 1146
+    },
+    {
+      "epoch": 2.5446478092068774,
+      "grad_norm": 2.7459795475006104,
+      "learning_rate": 8.353909465020576e-06,
+      "loss": 0.5218,
+      "step": 1147
+    },
+    {
+      "epoch": 2.5468663338879645,
+      "grad_norm": 2.3554394245147705,
+      "learning_rate": 8.31275720164609e-06,
+      "loss": 0.3276,
+      "step": 1148
+    },
+    {
+      "epoch": 2.5490848585690515,
+      "grad_norm": 2.0213518142700195,
+      "learning_rate": 8.271604938271606e-06,
+      "loss": 0.2486,
+      "step": 1149
+    },
+    {
+      "epoch": 2.5513033832501386,
+      "grad_norm": 2.139070749282837,
+      "learning_rate": 8.23045267489712e-06,
+      "loss": 0.2356,
+      "step": 1150
+    },
+    {
+      "epoch": 2.5535219079312257,
+      "grad_norm": 1.8092575073242188,
+      "learning_rate": 8.189300411522634e-06,
+      "loss": 0.3092,
+      "step": 1151
+    },
+    {
+      "epoch": 2.5557404326123128,
+      "grad_norm": 2.756450653076172,
+      "learning_rate": 8.14814814814815e-06,
+      "loss": 0.2545,
+      "step": 1152
+    },
+    {
+      "epoch": 2.5579589572934,
+      "grad_norm": 1.9591535329818726,
+      "learning_rate": 8.106995884773662e-06,
+      "loss": 0.2841,
+      "step": 1153
+    },
+    {
+      "epoch": 2.560177481974487,
+      "grad_norm": 1.720252513885498,
+      "learning_rate": 8.065843621399177e-06,
+      "loss": 0.2489,
+      "step": 1154
+    },
+    {
+      "epoch": 2.562396006655574,
+      "grad_norm": 2.693087339401245,
+      "learning_rate": 8.02469135802469e-06,
+      "loss": 0.2604,
+      "step": 1155
+    },
+    {
+      "epoch": 2.564614531336661,
+      "grad_norm": 1.9718818664550781,
+      "learning_rate": 7.983539094650207e-06,
+      "loss": 0.2159,
+      "step": 1156
+    },
+    {
+      "epoch": 2.566833056017748,
+      "grad_norm": 1.7725666761398315,
+      "learning_rate": 7.94238683127572e-06,
+      "loss": 0.1737,
+      "step": 1157
+    },
+    {
+      "epoch": 2.569051580698835,
+      "grad_norm": 1.9044920206069946,
+      "learning_rate": 7.901234567901235e-06,
+      "loss": 0.3333,
+      "step": 1158
+    },
+    {
+      "epoch": 2.5712701053799223,
+      "grad_norm": 2.145756721496582,
+      "learning_rate": 7.860082304526749e-06,
+      "loss": 0.2462,
+      "step": 1159
+    },
+    {
+      "epoch": 2.5734886300610094,
+      "grad_norm": 2.2089123725891113,
+      "learning_rate": 7.818930041152265e-06,
+      "loss": 0.2576,
+      "step": 1160
+    },
+    {
+      "epoch": 2.5757071547420964,
+      "grad_norm": 1.9764944314956665,
+      "learning_rate": 7.777777777777777e-06,
+      "loss": 0.3201,
+      "step": 1161
+    },
+    {
+      "epoch": 2.5779256794231835,
+      "grad_norm": 1.9338030815124512,
+      "learning_rate": 7.736625514403292e-06,
+      "loss": 0.2795,
+      "step": 1162
+    },
+    {
+      "epoch": 2.5801442041042706,
+      "grad_norm": 2.56862473487854,
+      "learning_rate": 7.695473251028806e-06,
+      "loss": 0.3127,
+      "step": 1163
+    },
+    {
+      "epoch": 2.5823627287853577,
+      "grad_norm": 2.0762412548065186,
+      "learning_rate": 7.654320987654322e-06,
+      "loss": 0.2496,
+      "step": 1164
+    },
+    {
+      "epoch": 2.5845812534664447,
+      "grad_norm": 2.10241961479187,
+      "learning_rate": 7.613168724279836e-06,
+      "loss": 0.2663,
+      "step": 1165
+    },
+    {
+      "epoch": 2.586799778147532,
+      "grad_norm": 2.4089910984039307,
+      "learning_rate": 7.57201646090535e-06,
+      "loss": 0.2628,
+      "step": 1166
+    },
+    {
+      "epoch": 2.589018302828619,
+      "grad_norm": 2.2896392345428467,
+      "learning_rate": 7.530864197530865e-06,
+      "loss": 0.3563,
+      "step": 1167
+    },
+    {
+      "epoch": 2.591236827509706,
+      "grad_norm": 1.9563854932785034,
+      "learning_rate": 7.489711934156379e-06,
+      "loss": 0.2391,
+      "step": 1168
+    },
+    {
+      "epoch": 2.593455352190793,
+      "grad_norm": 1.8316631317138672,
+      "learning_rate": 7.448559670781893e-06,
+      "loss": 0.2115,
+      "step": 1169
+    },
+    {
+      "epoch": 2.59567387687188,
+      "grad_norm": 1.8860446214675903,
+      "learning_rate": 7.4074074074074075e-06,
+      "loss": 0.3157,
+      "step": 1170
+    },
+    {
+      "epoch": 2.597892401552967,
+      "grad_norm": 2.391240358352661,
+      "learning_rate": 7.3662551440329225e-06,
+      "loss": 0.2485,
+      "step": 1171
+    },
+    {
+      "epoch": 2.6001109262340543,
+      "grad_norm": 2.0663042068481445,
+      "learning_rate": 7.325102880658437e-06,
+      "loss": 0.3215,
+      "step": 1172
+    },
+    {
+      "epoch": 2.6023294509151413,
+      "grad_norm": 1.8824645280838013,
+      "learning_rate": 7.283950617283951e-06,
+      "loss": 0.2452,
+      "step": 1173
+    },
+    {
+      "epoch": 2.6045479755962284,
+      "grad_norm": 1.8970526456832886,
+      "learning_rate": 7.242798353909465e-06,
+      "loss": 0.2445,
+      "step": 1174
+    },
+    {
+      "epoch": 2.6067665002773155,
+      "grad_norm": 3.369706869125366,
+      "learning_rate": 7.20164609053498e-06,
+      "loss": 0.2721,
+      "step": 1175
+    },
+    {
+      "epoch": 2.6089850249584026,
+      "grad_norm": 1.8989901542663574,
+      "learning_rate": 7.160493827160494e-06,
+      "loss": 0.3023,
+      "step": 1176
+    },
+    {
+      "epoch": 2.6112035496394896,
+      "grad_norm": 1.937748670578003,
+      "learning_rate": 7.119341563786008e-06,
+      "loss": 0.2399,
+      "step": 1177
+    },
+    {
+      "epoch": 2.6134220743205767,
+      "grad_norm": 2.3710336685180664,
+      "learning_rate": 7.0781893004115225e-06,
+      "loss": 0.2696,
+      "step": 1178
+    },
+    {
+      "epoch": 2.615640599001664,
+      "grad_norm": 2.3407680988311768,
+      "learning_rate": 7.0370370370370375e-06,
+      "loss": 0.3026,
+      "step": 1179
+    },
+    {
+      "epoch": 2.617859123682751,
+      "grad_norm": 2.110283613204956,
+      "learning_rate": 6.995884773662552e-06,
+      "loss": 0.3731,
+      "step": 1180
+    },
+    {
+      "epoch": 2.620077648363838,
+      "grad_norm": 1.7156367301940918,
+      "learning_rate": 6.954732510288066e-06,
+      "loss": 0.2705,
+      "step": 1181
+    },
+    {
+      "epoch": 2.622296173044925,
+      "grad_norm": 1.8122705221176147,
+      "learning_rate": 6.91358024691358e-06,
+      "loss": 0.2638,
+      "step": 1182
+    },
+    {
+      "epoch": 2.624514697726012,
+      "grad_norm": 1.5438884496688843,
+      "learning_rate": 6.872427983539095e-06,
+      "loss": 0.1384,
+      "step": 1183
+    },
+    {
+      "epoch": 2.626733222407099,
+      "grad_norm": 2.266481637954712,
+      "learning_rate": 6.831275720164609e-06,
+      "loss": 0.2904,
+      "step": 1184
+    },
+    {
+      "epoch": 2.6289517470881862,
+      "grad_norm": 1.9336191415786743,
+      "learning_rate": 6.790123456790123e-06,
+      "loss": 0.2888,
+      "step": 1185
+    },
+    {
+      "epoch": 2.6311702717692733,
+      "grad_norm": 1.7089993953704834,
+      "learning_rate": 6.748971193415639e-06,
+      "loss": 0.208,
+      "step": 1186
+    },
+    {
+      "epoch": 2.6333887964503604,
+      "grad_norm": 1.8265358209609985,
+      "learning_rate": 6.7078189300411525e-06,
+      "loss": 0.2981,
+      "step": 1187
+    },
+    {
+      "epoch": 2.6356073211314475,
+      "grad_norm": 2.208658456802368,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.1948,
+      "step": 1188
+    },
+    {
+      "epoch": 2.6378258458125345,
+      "grad_norm": 2.623060703277588,
+      "learning_rate": 6.625514403292181e-06,
+      "loss": 0.2961,
+      "step": 1189
+    },
+    {
+      "epoch": 2.6400443704936216,
+      "grad_norm": 1.832409381866455,
+      "learning_rate": 6.584362139917697e-06,
+      "loss": 0.2717,
+      "step": 1190
+    },
+    {
+      "epoch": 2.6422628951747087,
+      "grad_norm": 2.274874687194824,
+      "learning_rate": 6.54320987654321e-06,
+      "loss": 0.2338,
+      "step": 1191
+    },
+    {
+      "epoch": 2.6444814198557958,
+      "grad_norm": 2.073835611343384,
+      "learning_rate": 6.502057613168724e-06,
+      "loss": 0.3403,
+      "step": 1192
+    },
+    {
+      "epoch": 2.646699944536883,
+      "grad_norm": 2.124485492706299,
+      "learning_rate": 6.460905349794238e-06,
+      "loss": 0.3952,
+      "step": 1193
+    },
+    {
+      "epoch": 2.64891846921797,
+      "grad_norm": 2.1886112689971924,
+      "learning_rate": 6.419753086419754e-06,
+      "loss": 0.176,
+      "step": 1194
+    },
+    {
+      "epoch": 2.651136993899057,
+      "grad_norm": 1.9813517332077026,
+      "learning_rate": 6.3786008230452675e-06,
+      "loss": 0.2529,
+      "step": 1195
+    },
+    {
+      "epoch": 2.653355518580144,
+      "grad_norm": 2.1491875648498535,
+      "learning_rate": 6.337448559670782e-06,
+      "loss": 0.2699,
+      "step": 1196
+    },
+    {
+      "epoch": 2.655574043261231,
+      "grad_norm": 1.6620157957077026,
+      "learning_rate": 6.296296296296296e-06,
+      "loss": 0.2021,
+      "step": 1197
+    },
+    {
+      "epoch": 2.657792567942318,
+      "grad_norm": 1.7726198434829712,
+      "learning_rate": 6.255144032921812e-06,
+      "loss": 0.2213,
+      "step": 1198
+    },
+    {
+      "epoch": 2.6600110926234053,
+      "grad_norm": 1.9511240720748901,
+      "learning_rate": 6.213991769547326e-06,
+      "loss": 0.2662,
+      "step": 1199
+    },
+    {
+      "epoch": 2.6622296173044924,
+      "grad_norm": 2.2734010219573975,
+      "learning_rate": 6.172839506172839e-06,
+      "loss": 0.1958,
+      "step": 1200
+    },
+    {
+      "epoch": 2.6644481419855794,
+      "grad_norm": 2.194063186645508,
+      "learning_rate": 6.131687242798354e-06,
+      "loss": 0.2592,
+      "step": 1201
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 1.8995997905731201,
+      "learning_rate": 6.090534979423869e-06,
+      "loss": 0.3351,
+      "step": 1202
+    },
+    {
+      "epoch": 2.6688851913477536,
+      "grad_norm": 2.201388120651245,
+      "learning_rate": 6.049382716049383e-06,
+      "loss": 0.3515,
+      "step": 1203
+    },
+    {
+      "epoch": 2.6711037160288407,
+      "grad_norm": 2.87540340423584,
+      "learning_rate": 6.0082304526748975e-06,
+      "loss": 0.3559,
+      "step": 1204
+    },
+    {
+      "epoch": 2.6733222407099277,
+      "grad_norm": 2.180742025375366,
+      "learning_rate": 5.967078189300412e-06,
+      "loss": 0.2616,
+      "step": 1205
+    },
+    {
+      "epoch": 2.675540765391015,
+      "grad_norm": 2.0573318004608154,
+      "learning_rate": 5.925925925925927e-06,
+      "loss": 0.2843,
+      "step": 1206
+    },
+    {
+      "epoch": 2.677759290072102,
+      "grad_norm": 1.818904995918274,
+      "learning_rate": 5.884773662551441e-06,
+      "loss": 0.3119,
+      "step": 1207
+    },
+    {
+      "epoch": 2.679977814753189,
+      "grad_norm": 2.3869171142578125,
+      "learning_rate": 5.843621399176955e-06,
+      "loss": 0.2925,
+      "step": 1208
+    },
+    {
+      "epoch": 2.682196339434276,
+      "grad_norm": 2.0359601974487305,
+      "learning_rate": 5.802469135802469e-06,
+      "loss": 0.2605,
+      "step": 1209
+    },
+    {
+      "epoch": 2.684414864115363,
+      "grad_norm": 1.7741838693618774,
+      "learning_rate": 5.761316872427984e-06,
+      "loss": 0.2325,
+      "step": 1210
+    },
+    {
+      "epoch": 2.68663338879645,
+      "grad_norm": 2.1508171558380127,
+      "learning_rate": 5.720164609053498e-06,
+      "loss": 0.2805,
+      "step": 1211
+    },
+    {
+      "epoch": 2.6888519134775377,
+      "grad_norm": 2.265312671661377,
+      "learning_rate": 5.6790123456790125e-06,
+      "loss": 0.3615,
+      "step": 1212
+    },
+    {
+      "epoch": 2.691070438158625,
+      "grad_norm": 1.9257813692092896,
+      "learning_rate": 5.637860082304527e-06,
+      "loss": 0.3181,
+      "step": 1213
+    },
+    {
+      "epoch": 2.693288962839712,
+      "grad_norm": 2.41951060295105,
+      "learning_rate": 5.596707818930042e-06,
+      "loss": 0.2199,
+      "step": 1214
+    },
+    {
+      "epoch": 2.695507487520799,
+      "grad_norm": 1.9586740732192993,
+      "learning_rate": 5.555555555555556e-06,
+      "loss": 0.2239,
+      "step": 1215
+    },
+    {
+      "epoch": 2.697726012201886,
+      "grad_norm": 2.365335464477539,
+      "learning_rate": 5.51440329218107e-06,
+      "loss": 0.309,
+      "step": 1216
+    },
+    {
+      "epoch": 2.699944536882973,
+      "grad_norm": 2.933830499649048,
+      "learning_rate": 5.473251028806585e-06,
+      "loss": 0.2715,
+      "step": 1217
+    },
+    {
+      "epoch": 2.70216306156406,
+      "grad_norm": 2.2714312076568604,
+      "learning_rate": 5.432098765432099e-06,
+      "loss": 0.1797,
+      "step": 1218
+    },
+    {
+      "epoch": 2.7043815862451472,
+      "grad_norm": 2.8794028759002686,
+      "learning_rate": 5.390946502057613e-06,
+      "loss": 0.2826,
+      "step": 1219
+    },
+    {
+      "epoch": 2.7066001109262343,
+      "grad_norm": 2.6753485202789307,
+      "learning_rate": 5.3497942386831275e-06,
+      "loss": 0.2913,
+      "step": 1220
+    },
+    {
+      "epoch": 2.7088186356073214,
+      "grad_norm": 2.585235595703125,
+      "learning_rate": 5.3086419753086425e-06,
+      "loss": 0.3193,
+      "step": 1221
+    },
+    {
+      "epoch": 2.7110371602884085,
+      "grad_norm": 2.2004103660583496,
+      "learning_rate": 5.267489711934157e-06,
+      "loss": 0.4991,
+      "step": 1222
+    },
+    {
+      "epoch": 2.7132556849694955,
+      "grad_norm": 1.9427120685577393,
+      "learning_rate": 5.226337448559672e-06,
+      "loss": 0.3428,
+      "step": 1223
+    },
+    {
+      "epoch": 2.7154742096505826,
+      "grad_norm": 2.127936840057373,
+      "learning_rate": 5.185185185185185e-06,
+      "loss": 0.4401,
+      "step": 1224
+    },
+    {
+      "epoch": 2.7176927343316697,
+      "grad_norm": 3.177671194076538,
+      "learning_rate": 5.1440329218107e-06,
+      "loss": 0.2559,
+      "step": 1225
+    },
+    {
+      "epoch": 2.7199112590127568,
+      "grad_norm": 1.9325515031814575,
+      "learning_rate": 5.102880658436214e-06,
+      "loss": 0.2143,
+      "step": 1226
+    },
+    {
+      "epoch": 2.722129783693844,
+      "grad_norm": 2.2911150455474854,
+      "learning_rate": 5.061728395061729e-06,
+      "loss": 0.2833,
+      "step": 1227
+    },
+    {
+      "epoch": 2.724348308374931,
+      "grad_norm": 1.973244309425354,
+      "learning_rate": 5.0205761316872425e-06,
+      "loss": 0.3489,
+      "step": 1228
+    },
+    {
+      "epoch": 2.726566833056018,
+      "grad_norm": 3.597142457962036,
+      "learning_rate": 4.9794238683127575e-06,
+      "loss": 0.2527,
+      "step": 1229
+    },
+    {
+      "epoch": 2.728785357737105,
+      "grad_norm": 2.1094846725463867,
+      "learning_rate": 4.938271604938272e-06,
+      "loss": 0.2222,
+      "step": 1230
+    },
+    {
+      "epoch": 2.731003882418192,
+      "grad_norm": 1.816705346107483,
+      "learning_rate": 4.897119341563787e-06,
+      "loss": 0.2923,
+      "step": 1231
+    },
+    {
+      "epoch": 2.733222407099279,
+      "grad_norm": 2.2864887714385986,
+      "learning_rate": 4.8559670781893e-06,
+      "loss": 0.1585,
+      "step": 1232
+    },
+    {
+      "epoch": 2.7354409317803663,
+      "grad_norm": 1.693095088005066,
+      "learning_rate": 4.814814814814815e-06,
+      "loss": 0.2133,
+      "step": 1233
+    },
+    {
+      "epoch": 2.7376594564614534,
+      "grad_norm": 1.8832907676696777,
+      "learning_rate": 4.77366255144033e-06,
+      "loss": 0.2262,
+      "step": 1234
+    },
+    {
+      "epoch": 2.7398779811425404,
+      "grad_norm": 2.3603062629699707,
+      "learning_rate": 4.732510288065844e-06,
+      "loss": 0.3244,
+      "step": 1235
+    },
+    {
+      "epoch": 2.7420965058236275,
+      "grad_norm": 1.904842495918274,
+      "learning_rate": 4.691358024691358e-06,
+      "loss": 0.26,
+      "step": 1236
+    },
+    {
+      "epoch": 2.7443150305047146,
+      "grad_norm": 2.0658421516418457,
+      "learning_rate": 4.6502057613168725e-06,
+      "loss": 0.3774,
+      "step": 1237
+    },
+    {
+      "epoch": 2.7465335551858017,
+      "grad_norm": 1.8961118459701538,
+      "learning_rate": 4.6090534979423875e-06,
+      "loss": 0.2988,
+      "step": 1238
+    },
+    {
+      "epoch": 2.7487520798668887,
+      "grad_norm": 2.080554485321045,
+      "learning_rate": 4.567901234567902e-06,
+      "loss": 0.2757,
+      "step": 1239
+    },
+    {
+      "epoch": 2.750970604547976,
+      "grad_norm": 2.751340866088867,
+      "learning_rate": 4.526748971193416e-06,
+      "loss": 0.3095,
+      "step": 1240
+    },
+    {
+      "epoch": 2.753189129229063,
+      "grad_norm": 2.883486270904541,
+      "learning_rate": 4.48559670781893e-06,
+      "loss": 0.3613,
+      "step": 1241
+    },
+    {
+      "epoch": 2.75540765391015,
+      "grad_norm": 1.829049825668335,
+      "learning_rate": 4.444444444444445e-06,
+      "loss": 0.2674,
+      "step": 1242
+    },
+    {
+      "epoch": 2.757626178591237,
+      "grad_norm": 1.8414299488067627,
+      "learning_rate": 4.403292181069959e-06,
+      "loss": 0.291,
+      "step": 1243
+    },
+    {
+      "epoch": 2.759844703272324,
+      "grad_norm": 2.1617231369018555,
+      "learning_rate": 4.362139917695473e-06,
+      "loss": 0.2981,
+      "step": 1244
+    },
+    {
+      "epoch": 2.762063227953411,
+      "grad_norm": 2.1167025566101074,
+      "learning_rate": 4.3209876543209875e-06,
+      "loss": 0.4385,
+      "step": 1245
+    },
+    {
+      "epoch": 2.7642817526344983,
+      "grad_norm": 2.0715112686157227,
+      "learning_rate": 4.2798353909465025e-06,
+      "loss": 0.2928,
+      "step": 1246
+    },
+    {
+      "epoch": 2.7665002773155853,
+      "grad_norm": 2.157557487487793,
+      "learning_rate": 4.238683127572017e-06,
+      "loss": 0.3556,
+      "step": 1247
+    },
+    {
+      "epoch": 2.7687188019966724,
+      "grad_norm": 2.044938564300537,
+      "learning_rate": 4.197530864197531e-06,
+      "loss": 0.2202,
+      "step": 1248
+    },
+    {
+      "epoch": 2.7709373266777595,
+      "grad_norm": 2.103027820587158,
+      "learning_rate": 4.156378600823045e-06,
+      "loss": 0.402,
+      "step": 1249
+    },
+    {
+      "epoch": 2.7731558513588466,
+      "grad_norm": 1.923935890197754,
+      "learning_rate": 4.11522633744856e-06,
+      "loss": 0.2624,
+      "step": 1250
+    },
+    {
+      "epoch": 2.7753743760399336,
+      "grad_norm": 2.259981632232666,
+      "learning_rate": 4.074074074074075e-06,
+      "loss": 0.2978,
+      "step": 1251
+    },
+    {
+      "epoch": 2.7775929007210207,
+      "grad_norm": 1.9912458658218384,
+      "learning_rate": 4.032921810699588e-06,
+      "loss": 0.255,
+      "step": 1252
+    },
+    {
+      "epoch": 2.779811425402108,
+      "grad_norm": 2.565723180770874,
+      "learning_rate": 3.991769547325103e-06,
+      "loss": 0.3798,
+      "step": 1253
+    },
+    {
+      "epoch": 2.782029950083195,
+      "grad_norm": 1.8826565742492676,
+      "learning_rate": 3.9506172839506175e-06,
+      "loss": 0.136,
+      "step": 1254
+    },
+    {
+      "epoch": 2.784248474764282,
+      "grad_norm": 1.8294789791107178,
+      "learning_rate": 3.9094650205761325e-06,
+      "loss": 0.3177,
+      "step": 1255
+    },
+    {
+      "epoch": 2.786466999445369,
+      "grad_norm": 2.041980504989624,
+      "learning_rate": 3.868312757201646e-06,
+      "loss": 0.251,
+      "step": 1256
+    },
+    {
+      "epoch": 2.788685524126456,
+      "grad_norm": 2.7729990482330322,
+      "learning_rate": 3.827160493827161e-06,
+      "loss": 0.34,
+      "step": 1257
+    },
+    {
+      "epoch": 2.790904048807543,
+      "grad_norm": 2.1244142055511475,
+      "learning_rate": 3.786008230452675e-06,
+      "loss": 0.243,
+      "step": 1258
+    },
+    {
+      "epoch": 2.7931225734886302,
+      "grad_norm": 2.200659990310669,
+      "learning_rate": 3.7448559670781896e-06,
+      "loss": 0.3849,
+      "step": 1259
+    },
+    {
+      "epoch": 2.7953410981697173,
+      "grad_norm": 1.8824336528778076,
+      "learning_rate": 3.7037037037037037e-06,
+      "loss": 0.2984,
+      "step": 1260
+    },
+    {
+      "epoch": 2.7975596228508044,
+      "grad_norm": 1.6326273679733276,
+      "learning_rate": 3.6625514403292183e-06,
+      "loss": 0.2096,
+      "step": 1261
+    },
+    {
+      "epoch": 2.7997781475318915,
+      "grad_norm": 1.7609343528747559,
+      "learning_rate": 3.6213991769547325e-06,
+      "loss": 0.2654,
+      "step": 1262
+    },
+    {
+      "epoch": 2.8019966722129785,
+      "grad_norm": 1.6756187677383423,
+      "learning_rate": 3.580246913580247e-06,
+      "loss": 0.2377,
+      "step": 1263
+    },
+    {
+      "epoch": 2.8042151968940656,
+      "grad_norm": 2.0622122287750244,
+      "learning_rate": 3.5390946502057612e-06,
+      "loss": 0.2774,
+      "step": 1264
+    },
+    {
+      "epoch": 2.8064337215751527,
+      "grad_norm": 2.5383427143096924,
+      "learning_rate": 3.497942386831276e-06,
+      "loss": 0.3156,
+      "step": 1265
+    },
+    {
+      "epoch": 2.8086522462562398,
+      "grad_norm": 1.9302273988723755,
+      "learning_rate": 3.45679012345679e-06,
+      "loss": 0.1794,
+      "step": 1266
+    },
+    {
+      "epoch": 2.810870770937327,
+      "grad_norm": 2.0592775344848633,
+      "learning_rate": 3.4156378600823046e-06,
+      "loss": 0.3902,
+      "step": 1267
+    },
+    {
+      "epoch": 2.813089295618414,
+      "grad_norm": 2.1761741638183594,
+      "learning_rate": 3.3744855967078196e-06,
+      "loss": 0.2759,
+      "step": 1268
+    },
+    {
+      "epoch": 2.815307820299501,
+      "grad_norm": 2.390305757522583,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 0.2819,
+      "step": 1269
+    },
+    {
+      "epoch": 2.817526344980588,
+      "grad_norm": 1.826488733291626,
+      "learning_rate": 3.2921810699588483e-06,
+      "loss": 0.2132,
+      "step": 1270
+    },
+    {
+      "epoch": 2.819744869661675,
+      "grad_norm": 2.015404462814331,
+      "learning_rate": 3.251028806584362e-06,
+      "loss": 0.2583,
+      "step": 1271
+    },
+    {
+      "epoch": 2.821963394342762,
+      "grad_norm": 2.203927516937256,
+      "learning_rate": 3.209876543209877e-06,
+      "loss": 0.2698,
+      "step": 1272
+    },
+    {
+      "epoch": 2.8241819190238493,
+      "grad_norm": 3.9384591579437256,
+      "learning_rate": 3.168724279835391e-06,
+      "loss": 0.3698,
+      "step": 1273
+    },
+    {
+      "epoch": 2.8264004437049364,
+      "grad_norm": 2.8700945377349854,
+      "learning_rate": 3.127572016460906e-06,
+      "loss": 0.278,
+      "step": 1274
+    },
+    {
+      "epoch": 2.8286189683860234,
+      "grad_norm": 2.4028589725494385,
+      "learning_rate": 3.0864197530864196e-06,
+      "loss": 0.2708,
+      "step": 1275
+    },
+    {
+      "epoch": 2.8308374930671105,
+      "grad_norm": 2.0339558124542236,
+      "learning_rate": 3.0452674897119346e-06,
+      "loss": 0.2074,
+      "step": 1276
+    },
+    {
+      "epoch": 2.8330560177481976,
+      "grad_norm": 2.316519260406494,
+      "learning_rate": 3.0041152263374487e-06,
+      "loss": 0.3231,
+      "step": 1277
+    },
+    {
+      "epoch": 2.8352745424292847,
+      "grad_norm": 2.2555036544799805,
+      "learning_rate": 2.9629629629629633e-06,
+      "loss": 0.1941,
+      "step": 1278
+    },
+    {
+      "epoch": 2.8374930671103717,
+      "grad_norm": 1.9826371669769287,
+      "learning_rate": 2.9218106995884775e-06,
+      "loss": 0.2307,
+      "step": 1279
+    },
+    {
+      "epoch": 2.839711591791459,
+      "grad_norm": 2.127824306488037,
+      "learning_rate": 2.880658436213992e-06,
+      "loss": 0.269,
+      "step": 1280
+    },
+    {
+      "epoch": 2.841930116472546,
+      "grad_norm": 1.919284462928772,
+      "learning_rate": 2.8395061728395062e-06,
+      "loss": 0.298,
+      "step": 1281
+    },
+    {
+      "epoch": 2.844148641153633,
+      "grad_norm": 2.0485053062438965,
+      "learning_rate": 2.798353909465021e-06,
+      "loss": 0.3455,
+      "step": 1282
+    },
+    {
+      "epoch": 2.84636716583472,
+      "grad_norm": 2.4598727226257324,
+      "learning_rate": 2.757201646090535e-06,
+      "loss": 0.2928,
+      "step": 1283
+    },
+    {
+      "epoch": 2.848585690515807,
+      "grad_norm": 1.5110982656478882,
+      "learning_rate": 2.7160493827160496e-06,
+      "loss": 0.232,
+      "step": 1284
+    },
+    {
+      "epoch": 2.850804215196894,
+      "grad_norm": 2.149141311645508,
+      "learning_rate": 2.6748971193415637e-06,
+      "loss": 0.2659,
+      "step": 1285
+    },
+    {
+      "epoch": 2.8530227398779813,
+      "grad_norm": 1.604087471961975,
+      "learning_rate": 2.6337448559670783e-06,
+      "loss": 0.2381,
+      "step": 1286
+    },
+    {
+      "epoch": 2.8552412645590683,
+      "grad_norm": 1.8919169902801514,
+      "learning_rate": 2.5925925925925925e-06,
+      "loss": 0.1438,
+      "step": 1287
+    },
+    {
+      "epoch": 2.8574597892401554,
+      "grad_norm": 2.6281063556671143,
+      "learning_rate": 2.551440329218107e-06,
+      "loss": 0.3273,
+      "step": 1288
+    },
+    {
+      "epoch": 2.8596783139212425,
+      "grad_norm": 2.1523869037628174,
+      "learning_rate": 2.5102880658436212e-06,
+      "loss": 0.2251,
+      "step": 1289
+    },
+    {
+      "epoch": 2.8618968386023296,
+      "grad_norm": 1.8037960529327393,
+      "learning_rate": 2.469135802469136e-06,
+      "loss": 0.2312,
+      "step": 1290
+    },
+    {
+      "epoch": 2.8641153632834166,
+      "grad_norm": 2.3673555850982666,
+      "learning_rate": 2.42798353909465e-06,
+      "loss": 0.3077,
+      "step": 1291
+    },
+    {
+      "epoch": 2.8663338879645037,
+      "grad_norm": 3.1339378356933594,
+      "learning_rate": 2.386831275720165e-06,
+      "loss": 0.1955,
+      "step": 1292
+    },
+    {
+      "epoch": 2.868552412645591,
+      "grad_norm": 1.8979777097702026,
+      "learning_rate": 2.345679012345679e-06,
+      "loss": 0.4415,
+      "step": 1293
+    },
+    {
+      "epoch": 2.870770937326678,
+      "grad_norm": 2.0577971935272217,
+      "learning_rate": 2.3045267489711937e-06,
+      "loss": 0.3583,
+      "step": 1294
+    },
+    {
+      "epoch": 2.872989462007765,
+      "grad_norm": 2.2515969276428223,
+      "learning_rate": 2.263374485596708e-06,
+      "loss": 0.3085,
+      "step": 1295
+    },
+    {
+      "epoch": 2.875207986688852,
+      "grad_norm": 1.8030778169631958,
+      "learning_rate": 2.2222222222222225e-06,
+      "loss": 0.336,
+      "step": 1296
+    },
+    {
+      "epoch": 2.877426511369939,
+      "grad_norm": 1.8946157693862915,
+      "learning_rate": 2.1810699588477367e-06,
+      "loss": 0.2516,
+      "step": 1297
+    },
+    {
+      "epoch": 2.879645036051026,
+      "grad_norm": 2.145420789718628,
+      "learning_rate": 2.1399176954732512e-06,
+      "loss": 0.3029,
+      "step": 1298
+    },
+    {
+      "epoch": 2.8818635607321132,
+      "grad_norm": 1.9406195878982544,
+      "learning_rate": 2.0987654320987654e-06,
+      "loss": 0.3157,
+      "step": 1299
+    },
+    {
+      "epoch": 2.8840820854132003,
+      "grad_norm": 1.7278413772583008,
+      "learning_rate": 2.05761316872428e-06,
+      "loss": 0.2637,
+      "step": 1300
+    },
+    {
+      "epoch": 2.8863006100942874,
+      "grad_norm": 1.8568264245986938,
+      "learning_rate": 2.016460905349794e-06,
+      "loss": 0.2399,
+      "step": 1301
+    },
+    {
+      "epoch": 2.8885191347753745,
+      "grad_norm": 1.8403472900390625,
+      "learning_rate": 1.9753086419753087e-06,
+      "loss": 0.1775,
+      "step": 1302
+    },
+    {
+      "epoch": 2.8907376594564616,
+      "grad_norm": 2.399531126022339,
+      "learning_rate": 1.934156378600823e-06,
+      "loss": 0.3012,
+      "step": 1303
+    },
+    {
+      "epoch": 2.8929561841375486,
+      "grad_norm": 2.408196449279785,
+      "learning_rate": 1.8930041152263375e-06,
+      "loss": 0.2736,
+      "step": 1304
+    },
+    {
+      "epoch": 2.8951747088186357,
+      "grad_norm": 2.902273416519165,
+      "learning_rate": 1.8518518518518519e-06,
+      "loss": 0.1985,
+      "step": 1305
+    },
+    {
+      "epoch": 2.8973932334997228,
+      "grad_norm": 1.6674243211746216,
+      "learning_rate": 1.8106995884773662e-06,
+      "loss": 0.2331,
+      "step": 1306
+    },
+    {
+      "epoch": 2.89961175818081,
+      "grad_norm": 1.6928460597991943,
+      "learning_rate": 1.7695473251028806e-06,
+      "loss": 0.2262,
+      "step": 1307
+    },
+    {
+      "epoch": 2.901830282861897,
+      "grad_norm": 2.151395559310913,
+      "learning_rate": 1.728395061728395e-06,
+      "loss": 0.3129,
+      "step": 1308
+    },
+    {
+      "epoch": 2.904048807542984,
+      "grad_norm": 2.017479419708252,
+      "learning_rate": 1.6872427983539098e-06,
+      "loss": 0.282,
+      "step": 1309
+    },
+    {
+      "epoch": 2.906267332224071,
+      "grad_norm": 1.7803949117660522,
+      "learning_rate": 1.6460905349794242e-06,
+      "loss": 0.1847,
+      "step": 1310
+    },
+    {
+      "epoch": 2.908485856905158,
+      "grad_norm": 2.072348117828369,
+      "learning_rate": 1.6049382716049385e-06,
+      "loss": 0.3533,
+      "step": 1311
+    },
+    {
+      "epoch": 2.9107043815862452,
+      "grad_norm": 1.8940985202789307,
+      "learning_rate": 1.563786008230453e-06,
+      "loss": 0.2884,
+      "step": 1312
+    },
+    {
+      "epoch": 2.9129229062673323,
+      "grad_norm": 2.2202045917510986,
+      "learning_rate": 1.5226337448559673e-06,
+      "loss": 0.1632,
+      "step": 1313
+    },
+    {
+      "epoch": 2.9151414309484194,
+      "grad_norm": 1.9068801403045654,
+      "learning_rate": 1.4814814814814817e-06,
+      "loss": 0.3166,
+      "step": 1314
+    },
+    {
+      "epoch": 2.9173599556295065,
+      "grad_norm": 2.3244311809539795,
+      "learning_rate": 1.440329218106996e-06,
+      "loss": 0.3098,
+      "step": 1315
+    },
+    {
+      "epoch": 2.9195784803105935,
+      "grad_norm": 2.1253225803375244,
+      "learning_rate": 1.3991769547325104e-06,
+      "loss": 0.3864,
+      "step": 1316
+    },
+    {
+      "epoch": 2.9217970049916806,
+      "grad_norm": 2.0462803840637207,
+      "learning_rate": 1.3580246913580248e-06,
+      "loss": 0.2668,
+      "step": 1317
+    },
+    {
+      "epoch": 2.9240155296727677,
+      "grad_norm": 2.3019838333129883,
+      "learning_rate": 1.3168724279835392e-06,
+      "loss": 0.2902,
+      "step": 1318
+    },
+    {
+      "epoch": 2.9262340543538548,
+      "grad_norm": 1.7224133014678955,
+      "learning_rate": 1.2757201646090535e-06,
+      "loss": 0.2791,
+      "step": 1319
+    },
+    {
+      "epoch": 2.928452579034942,
+      "grad_norm": 1.9416980743408203,
+      "learning_rate": 1.234567901234568e-06,
+      "loss": 0.2651,
+      "step": 1320
+    },
+    {
+      "epoch": 2.930671103716029,
+      "grad_norm": 2.03052020072937,
+      "learning_rate": 1.1934156378600825e-06,
+      "loss": 0.3808,
+      "step": 1321
+    },
+    {
+      "epoch": 2.932889628397116,
+      "grad_norm": 2.0344176292419434,
+      "learning_rate": 1.1522633744855969e-06,
+      "loss": 0.2624,
+      "step": 1322
+    },
+    {
+      "epoch": 2.935108153078203,
+      "grad_norm": 3.6349754333496094,
+      "learning_rate": 1.1111111111111112e-06,
+      "loss": 0.2465,
+      "step": 1323
+    },
+    {
+      "epoch": 2.93732667775929,
+      "grad_norm": 1.8408985137939453,
+      "learning_rate": 1.0699588477366256e-06,
+      "loss": 0.2466,
+      "step": 1324
+    },
+    {
+      "epoch": 2.939545202440377,
+      "grad_norm": 2.6119086742401123,
+      "learning_rate": 1.02880658436214e-06,
+      "loss": 0.3089,
+      "step": 1325
+    },
+    {
+      "epoch": 2.9417637271214643,
+      "grad_norm": 2.0840513706207275,
+      "learning_rate": 9.876543209876544e-07,
+      "loss": 0.23,
+      "step": 1326
+    },
+    {
+      "epoch": 2.9439822518025514,
+      "grad_norm": 2.022264242172241,
+      "learning_rate": 9.465020576131687e-07,
+      "loss": 0.2613,
+      "step": 1327
+    },
+    {
+      "epoch": 2.9462007764836384,
+      "grad_norm": 2.26277756690979,
+      "learning_rate": 9.053497942386831e-07,
+      "loss": 0.3212,
+      "step": 1328
+    },
+    {
+      "epoch": 2.9484193011647255,
+      "grad_norm": 1.7865893840789795,
+      "learning_rate": 8.641975308641975e-07,
+      "loss": 0.1608,
+      "step": 1329
+    },
+    {
+      "epoch": 2.9506378258458126,
+      "grad_norm": 1.847663402557373,
+      "learning_rate": 8.230452674897121e-07,
+      "loss": 0.2707,
+      "step": 1330
+    },
+    {
+      "epoch": 2.9528563505268997,
+      "grad_norm": 2.044006109237671,
+      "learning_rate": 7.818930041152265e-07,
+      "loss": 0.2727,
+      "step": 1331
+    },
+    {
+      "epoch": 2.9550748752079867,
+      "grad_norm": 2.0943212509155273,
+      "learning_rate": 7.407407407407408e-07,
+      "loss": 0.3326,
+      "step": 1332
+    },
+    {
+      "epoch": 2.957293399889074,
+      "grad_norm": 2.242365837097168,
+      "learning_rate": 6.995884773662552e-07,
+      "loss": 0.2244,
+      "step": 1333
+    },
+    {
+      "epoch": 2.959511924570161,
+      "grad_norm": 2.062919855117798,
+      "learning_rate": 6.584362139917696e-07,
+      "loss": 0.2566,
+      "step": 1334
+    },
+    {
+      "epoch": 2.961730449251248,
+      "grad_norm": 2.168520212173462,
+      "learning_rate": 6.17283950617284e-07,
+      "loss": 0.306,
+      "step": 1335
+    },
+    {
+      "epoch": 2.963948973932335,
+      "grad_norm": 2.6685831546783447,
+      "learning_rate": 5.761316872427984e-07,
+      "loss": 0.3108,
+      "step": 1336
+    },
+    {
+      "epoch": 2.966167498613422,
+      "grad_norm": 2.169834852218628,
+      "learning_rate": 5.349794238683128e-07,
+      "loss": 0.2576,
+      "step": 1337
+    },
+    {
+      "epoch": 2.968386023294509,
+      "grad_norm": 1.7842198610305786,
+      "learning_rate": 4.938271604938272e-07,
+      "loss": 0.2293,
+      "step": 1338
+    },
+    {
+      "epoch": 2.9706045479755963,
+      "grad_norm": 1.6931025981903076,
+      "learning_rate": 4.5267489711934156e-07,
+      "loss": 0.2432,
+      "step": 1339
+    },
+    {
+      "epoch": 2.9728230726566833,
+      "grad_norm": 2.2900238037109375,
+      "learning_rate": 4.1152263374485604e-07,
+      "loss": 0.355,
+      "step": 1340
+    },
+    {
+      "epoch": 2.9750415973377704,
+      "grad_norm": 2.149487257003784,
+      "learning_rate": 3.703703703703704e-07,
+      "loss": 0.3087,
+      "step": 1341
+    },
+    {
+      "epoch": 2.9772601220188575,
+      "grad_norm": 2.8405861854553223,
+      "learning_rate": 3.292181069958848e-07,
+      "loss": 0.2741,
+      "step": 1342
+    },
+    {
+      "epoch": 2.9794786466999446,
+      "grad_norm": 2.11864972114563,
+      "learning_rate": 2.880658436213992e-07,
+      "loss": 0.3088,
+      "step": 1343
+    },
+    {
+      "epoch": 2.9816971713810316,
+      "grad_norm": 3.0862112045288086,
+      "learning_rate": 2.469135802469136e-07,
+      "loss": 0.2184,
+      "step": 1344
+    },
+    {
+      "epoch": 2.9839156960621187,
+      "grad_norm": 2.1256487369537354,
+      "learning_rate": 2.0576131687242802e-07,
+      "loss": 0.5191,
+      "step": 1345
+    },
+    {
+      "epoch": 2.986134220743206,
+      "grad_norm": 2.5985140800476074,
+      "learning_rate": 1.646090534979424e-07,
+      "loss": 0.2977,
+      "step": 1346
+    },
+    {
+      "epoch": 2.988352745424293,
+      "grad_norm": 1.9144142866134644,
+      "learning_rate": 1.234567901234568e-07,
+      "loss": 0.2851,
+      "step": 1347
+    },
+    {
+      "epoch": 2.99057127010538,
+      "grad_norm": 2.070836067199707,
+      "learning_rate": 8.23045267489712e-08,
+      "loss": 0.2416,
+      "step": 1348
+    },
+    {
+      "epoch": 2.992789794786467,
+      "grad_norm": 2.599292516708374,
+      "learning_rate": 4.11522633744856e-08,
+      "loss": 0.278,
+      "step": 1349
+    },
+    {
+      "epoch": 2.995008319467554,
+      "grad_norm": 2.4956448078155518,
+      "learning_rate": 0.0,
+      "loss": 0.2371,
+      "step": 1350
+    },
+    {
+      "epoch": 2.995008319467554,
+      "step": 1350,
+      "total_flos": 2.324441285315789e+16,
+      "train_loss": 0.5607477950608288,
+      "train_runtime": 2259.369,
+      "train_samples_per_second": 9.575,
+      "train_steps_per_second": 0.598
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1350,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.324441285315789e+16,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}