diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9169 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.8352,
+  "eval_steps": 500,
+  "global_step": 1305,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00064,
+      "grad_norm": 2.353485086238037,
+      "learning_rate": 0.0,
+      "loss": 0.5764,
+      "step": 1
+    },
+    {
+      "epoch": 0.00128,
+      "grad_norm": 2.5676819953400525,
+      "learning_rate": 6.329113924050633e-07,
+      "loss": 0.6154,
+      "step": 2
+    },
+    {
+      "epoch": 0.00192,
+      "grad_norm": 2.4406327433603328,
+      "learning_rate": 1.2658227848101265e-06,
+      "loss": 0.5765,
+      "step": 3
+    },
+    {
+      "epoch": 0.00256,
+      "grad_norm": 2.503604928634966,
+      "learning_rate": 1.8987341772151901e-06,
+      "loss": 0.5896,
+      "step": 4
+    },
+    {
+      "epoch": 0.0032,
+      "grad_norm": 2.3561613670472124,
+      "learning_rate": 2.531645569620253e-06,
+      "loss": 0.5848,
+      "step": 5
+    },
+    {
+      "epoch": 0.00384,
+      "grad_norm": 2.487557281842397,
+      "learning_rate": 3.1645569620253167e-06,
+      "loss": 0.6005,
+      "step": 6
+    },
+    {
+      "epoch": 0.00448,
+      "grad_norm": 1.8603603695583997,
+      "learning_rate": 3.7974683544303802e-06,
+      "loss": 0.558,
+      "step": 7
+    },
+    {
+      "epoch": 0.00512,
+      "grad_norm": 2.0016494810134997,
+      "learning_rate": 4.430379746835443e-06,
+      "loss": 0.586,
+      "step": 8
+    },
+    {
+      "epoch": 0.00576,
+      "grad_norm": 1.3866160295372776,
+      "learning_rate": 5.063291139240506e-06,
+      "loss": 0.5485,
+      "step": 9
+    },
+    {
+      "epoch": 0.0064,
+      "grad_norm": 1.2536265188961684,
+      "learning_rate": 5.69620253164557e-06,
+      "loss": 0.5304,
+      "step": 10
+    },
+    {
+      "epoch": 0.00704,
+      "grad_norm": 1.1405181059205818,
+      "learning_rate": 6.329113924050633e-06,
+      "loss": 0.5573,
+      "step": 11
+    },
+    {
+      "epoch": 0.00768,
+      "grad_norm": 1.4739138986648344,
+      "learning_rate": 6.9620253164556965e-06,
+      "loss": 0.5111,
+      "step": 12
+    },
+    {
+      "epoch": 0.00832,
+      "grad_norm": 1.5987665650407654,
+      "learning_rate": 7.5949367088607605e-06,
+      "loss": 0.4672,
+      "step": 13
+    },
+    {
+      "epoch": 0.00896,
+      "grad_norm": 1.5801662763772257,
+      "learning_rate": 8.227848101265822e-06,
+      "loss": 0.4803,
+      "step": 14
+    },
+    {
+      "epoch": 0.0096,
+      "grad_norm": 1.1595227910276567,
+      "learning_rate": 8.860759493670886e-06,
+      "loss": 0.4413,
+      "step": 15
+    },
+    {
+      "epoch": 0.01024,
+      "grad_norm": 1.0232799236259744,
+      "learning_rate": 9.49367088607595e-06,
+      "loss": 0.426,
+      "step": 16
+    },
+    {
+      "epoch": 0.01088,
+      "grad_norm": 1.4200883806078057,
+      "learning_rate": 1.0126582278481012e-05,
+      "loss": 0.4544,
+      "step": 17
+    },
+    {
+      "epoch": 0.01152,
+      "grad_norm": 1.2500413307829548,
+      "learning_rate": 1.0759493670886076e-05,
+      "loss": 0.453,
+      "step": 18
+    },
+    {
+      "epoch": 0.01216,
+      "grad_norm": 0.9645673541147013,
+      "learning_rate": 1.139240506329114e-05,
+      "loss": 0.4149,
+      "step": 19
+    },
+    {
+      "epoch": 0.0128,
+      "grad_norm": 0.7619491956637449,
+      "learning_rate": 1.2025316455696203e-05,
+      "loss": 0.4268,
+      "step": 20
+    },
+    {
+      "epoch": 0.01344,
+      "grad_norm": 0.7950927664975571,
+      "learning_rate": 1.2658227848101267e-05,
+      "loss": 0.4315,
+      "step": 21
+    },
+    {
+      "epoch": 0.01408,
+      "grad_norm": 0.9130569306644377,
+      "learning_rate": 1.3291139240506329e-05,
+      "loss": 0.367,
+      "step": 22
+    },
+    {
+      "epoch": 0.01472,
+      "grad_norm": 0.6877425501125356,
+      "learning_rate": 1.3924050632911393e-05,
+      "loss": 0.3838,
+      "step": 23
+    },
+    {
+      "epoch": 0.01536,
+      "grad_norm": 0.678423749873734,
+      "learning_rate": 1.4556962025316457e-05,
+      "loss": 0.3542,
+      "step": 24
+    },
+    {
+      "epoch": 0.016,
+      "grad_norm": 0.8086215595785854,
+      "learning_rate": 1.5189873417721521e-05,
+      "loss": 0.3947,
+      "step": 25
+    },
+    {
+      "epoch": 0.01664,
+      "grad_norm": 0.7365716120703668,
+      "learning_rate": 1.5822784810126583e-05,
+      "loss": 0.3761,
+      "step": 26
+    },
+    {
+      "epoch": 0.01728,
+      "grad_norm": 0.57410441755097,
+      "learning_rate": 1.6455696202531644e-05,
+      "loss": 0.3722,
+      "step": 27
+    },
+    {
+      "epoch": 0.01792,
+      "grad_norm": 0.5067303990784363,
+      "learning_rate": 1.7088607594936708e-05,
+      "loss": 0.371,
+      "step": 28
+    },
+    {
+      "epoch": 0.01856,
+      "grad_norm": 0.5387436191633763,
+      "learning_rate": 1.7721518987341772e-05,
+      "loss": 0.3577,
+      "step": 29
+    },
+    {
+      "epoch": 0.0192,
+      "grad_norm": 0.6059447121681072,
+      "learning_rate": 1.8354430379746836e-05,
+      "loss": 0.3836,
+      "step": 30
+    },
+    {
+      "epoch": 0.01984,
+      "grad_norm": 0.5055998458711023,
+      "learning_rate": 1.89873417721519e-05,
+      "loss": 0.3783,
+      "step": 31
+    },
+    {
+      "epoch": 0.02048,
+      "grad_norm": 0.5708720902953378,
+      "learning_rate": 1.962025316455696e-05,
+      "loss": 0.3846,
+      "step": 32
+    },
+    {
+      "epoch": 0.02112,
+      "grad_norm": 0.557929165518282,
+      "learning_rate": 2.0253164556962025e-05,
+      "loss": 0.3484,
+      "step": 33
+    },
+    {
+      "epoch": 0.02176,
+      "grad_norm": 0.4519934652780253,
+      "learning_rate": 2.088607594936709e-05,
+      "loss": 0.3615,
+      "step": 34
+    },
+    {
+      "epoch": 0.0224,
+      "grad_norm": 0.43727165616564423,
+      "learning_rate": 2.1518987341772153e-05,
+      "loss": 0.3599,
+      "step": 35
+    },
+    {
+      "epoch": 0.02304,
+      "grad_norm": 0.44623181877706214,
+      "learning_rate": 2.2151898734177217e-05,
+      "loss": 0.3722,
+      "step": 36
+    },
+    {
+      "epoch": 0.02368,
+      "grad_norm": 0.4635506547428751,
+      "learning_rate": 2.278481012658228e-05,
+      "loss": 0.3628,
+      "step": 37
+    },
+    {
+      "epoch": 0.02432,
+      "grad_norm": 0.4583180814360715,
+      "learning_rate": 2.341772151898734e-05,
+      "loss": 0.3584,
+      "step": 38
+    },
+    {
+      "epoch": 0.02496,
+      "grad_norm": 0.40794109495076275,
+      "learning_rate": 2.4050632911392405e-05,
+      "loss": 0.3528,
+      "step": 39
+    },
+    {
+      "epoch": 0.0256,
+      "grad_norm": 0.3709932525452395,
+      "learning_rate": 2.468354430379747e-05,
+      "loss": 0.345,
+      "step": 40
+    },
+    {
+      "epoch": 0.02624,
+      "grad_norm": 0.4210238800775235,
+      "learning_rate": 2.5316455696202533e-05,
+      "loss": 0.3056,
+      "step": 41
+    },
+    {
+      "epoch": 0.02688,
+      "grad_norm": 0.4786731092715819,
+      "learning_rate": 2.5949367088607597e-05,
+      "loss": 0.3329,
+      "step": 42
+    },
+    {
+      "epoch": 0.02752,
+      "grad_norm": 0.38803905635556746,
+      "learning_rate": 2.6582278481012658e-05,
+      "loss": 0.3227,
+      "step": 43
+    },
+    {
+      "epoch": 0.02816,
+      "grad_norm": 0.43575361580226557,
+      "learning_rate": 2.7215189873417722e-05,
+      "loss": 0.3398,
+      "step": 44
+    },
+    {
+      "epoch": 0.0288,
+      "grad_norm": 0.452207614606097,
+      "learning_rate": 2.7848101265822786e-05,
+      "loss": 0.3512,
+      "step": 45
+    },
+    {
+      "epoch": 0.02944,
+      "grad_norm": 0.4698411017252047,
+      "learning_rate": 2.848101265822785e-05,
+      "loss": 0.3435,
+      "step": 46
+    },
+    {
+      "epoch": 0.03008,
+      "grad_norm": 0.40870853690109027,
+      "learning_rate": 2.9113924050632914e-05,
+      "loss": 0.3367,
+      "step": 47
+    },
+    {
+      "epoch": 0.03072,
+      "grad_norm": 0.4405164959813624,
+      "learning_rate": 2.9746835443037974e-05,
+      "loss": 0.3476,
+      "step": 48
+    },
+    {
+      "epoch": 0.03136,
+      "grad_norm": 0.4259862667341005,
+      "learning_rate": 3.0379746835443042e-05,
+      "loss": 0.3305,
+      "step": 49
+    },
+    {
+      "epoch": 0.032,
+      "grad_norm": 0.39771125914492406,
+      "learning_rate": 3.10126582278481e-05,
+      "loss": 0.3395,
+      "step": 50
+    },
+    {
+      "epoch": 0.03264,
+      "grad_norm": 0.4111747463996418,
+      "learning_rate": 3.1645569620253167e-05,
+      "loss": 0.3274,
+      "step": 51
+    },
+    {
+      "epoch": 0.03328,
+      "grad_norm": 0.42586319667255845,
+      "learning_rate": 3.227848101265823e-05,
+      "loss": 0.3258,
+      "step": 52
+    },
+    {
+      "epoch": 0.03392,
+      "grad_norm": 0.40788830009535015,
+      "learning_rate": 3.291139240506329e-05,
+      "loss": 0.2907,
+      "step": 53
+    },
+    {
+      "epoch": 0.03456,
+      "grad_norm": 0.43296600863158063,
+      "learning_rate": 3.354430379746836e-05,
+      "loss": 0.3345,
+      "step": 54
+    },
+    {
+      "epoch": 0.0352,
+      "grad_norm": 0.42372120204803543,
+      "learning_rate": 3.4177215189873416e-05,
+      "loss": 0.3293,
+      "step": 55
+    },
+    {
+      "epoch": 0.03584,
+      "grad_norm": 0.4469884725915102,
+      "learning_rate": 3.4810126582278487e-05,
+      "loss": 0.3287,
+      "step": 56
+    },
+    {
+      "epoch": 0.03648,
+      "grad_norm": 0.41612674822365947,
+      "learning_rate": 3.5443037974683544e-05,
+      "loss": 0.3279,
+      "step": 57
+    },
+    {
+      "epoch": 0.03712,
+      "grad_norm": 0.45327242403181206,
+      "learning_rate": 3.607594936708861e-05,
+      "loss": 0.3468,
+      "step": 58
+    },
+    {
+      "epoch": 0.03776,
+      "grad_norm": 0.44353765655450733,
+      "learning_rate": 3.670886075949367e-05,
+      "loss": 0.3341,
+      "step": 59
+    },
+    {
+      "epoch": 0.0384,
+      "grad_norm": 0.44486424984007067,
+      "learning_rate": 3.7341772151898736e-05,
+      "loss": 0.3275,
+      "step": 60
+    },
+    {
+      "epoch": 0.03904,
+      "grad_norm": 0.4383809063442672,
+      "learning_rate": 3.79746835443038e-05,
+      "loss": 0.3342,
+      "step": 61
+    },
+    {
+      "epoch": 0.03968,
+      "grad_norm": 0.42608132493729306,
+      "learning_rate": 3.8607594936708864e-05,
+      "loss": 0.3155,
+      "step": 62
+    },
+    {
+      "epoch": 0.04032,
+      "grad_norm": 0.4555942982493556,
+      "learning_rate": 3.924050632911392e-05,
+      "loss": 0.3205,
+      "step": 63
+    },
+    {
+      "epoch": 0.04096,
+      "grad_norm": 0.494264400213869,
+      "learning_rate": 3.987341772151899e-05,
+      "loss": 0.3542,
+      "step": 64
+    },
+    {
+      "epoch": 0.0416,
+      "grad_norm": 0.3924518429856125,
+      "learning_rate": 4.050632911392405e-05,
+      "loss": 0.3116,
+      "step": 65
+    },
+    {
+      "epoch": 0.04224,
+      "grad_norm": 0.4658349233271844,
+      "learning_rate": 4.113924050632912e-05,
+      "loss": 0.3461,
+      "step": 66
+    },
+    {
+      "epoch": 0.04288,
+      "grad_norm": 0.4151003836242027,
+      "learning_rate": 4.177215189873418e-05,
+      "loss": 0.3152,
+      "step": 67
+    },
+    {
+      "epoch": 0.04352,
+      "grad_norm": 0.4328542434853278,
+      "learning_rate": 4.240506329113924e-05,
+      "loss": 0.3267,
+      "step": 68
+    },
+    {
+      "epoch": 0.04416,
+      "grad_norm": 0.4255328771601494,
+      "learning_rate": 4.3037974683544305e-05,
+      "loss": 0.2943,
+      "step": 69
+    },
+    {
+      "epoch": 0.0448,
+      "grad_norm": 0.3960781879337196,
+      "learning_rate": 4.367088607594937e-05,
+      "loss": 0.3295,
+      "step": 70
+    },
+    {
+      "epoch": 0.04544,
+      "grad_norm": 0.42237083664040964,
+      "learning_rate": 4.430379746835443e-05,
+      "loss": 0.3156,
+      "step": 71
+    },
+    {
+      "epoch": 0.04608,
+      "grad_norm": 0.44252752029376236,
+      "learning_rate": 4.49367088607595e-05,
+      "loss": 0.3238,
+      "step": 72
+    },
+    {
+      "epoch": 0.04672,
+      "grad_norm": 0.4254192779309176,
+      "learning_rate": 4.556962025316456e-05,
+      "loss": 0.3137,
+      "step": 73
+    },
+    {
+      "epoch": 0.04736,
+      "grad_norm": 0.4068734829882739,
+      "learning_rate": 4.6202531645569625e-05,
+      "loss": 0.3082,
+      "step": 74
+    },
+    {
+      "epoch": 0.048,
+      "grad_norm": 0.38540339351470876,
+      "learning_rate": 4.683544303797468e-05,
+      "loss": 0.298,
+      "step": 75
+    },
+    {
+      "epoch": 0.04864,
+      "grad_norm": 0.43228604399379483,
+      "learning_rate": 4.7468354430379746e-05,
+      "loss": 0.3202,
+      "step": 76
+    },
+    {
+      "epoch": 0.04928,
+      "grad_norm": 0.4662811772881026,
+      "learning_rate": 4.810126582278481e-05,
+      "loss": 0.3267,
+      "step": 77
+    },
+    {
+      "epoch": 0.04992,
+      "grad_norm": 0.4101396625445811,
+      "learning_rate": 4.8734177215189874e-05,
+      "loss": 0.2908,
+      "step": 78
+    },
+    {
+      "epoch": 0.05056,
+      "grad_norm": 0.4407899025941757,
+      "learning_rate": 4.936708860759494e-05,
+      "loss": 0.3234,
+      "step": 79
+    },
+    {
+      "epoch": 0.0512,
+      "grad_norm": 0.42766797963707764,
+      "learning_rate": 5e-05,
+      "loss": 0.315,
+      "step": 80
+    },
+    {
+      "epoch": 0.05184,
+      "grad_norm": 0.39551732993824784,
+      "learning_rate": 4.999994390459423e-05,
+      "loss": 0.3241,
+      "step": 81
+    },
+    {
+      "epoch": 0.05248,
+      "grad_norm": 0.37307344185693914,
+      "learning_rate": 4.999977561862863e-05,
+      "loss": 0.3171,
+      "step": 82
+    },
+    {
+      "epoch": 0.05312,
+      "grad_norm": 0.47799834282771486,
+      "learning_rate": 4.999949514285842e-05,
+      "loss": 0.3174,
+      "step": 83
+    },
+    {
+      "epoch": 0.05376,
+      "grad_norm": 0.45391420249819664,
+      "learning_rate": 4.999910247854227e-05,
+      "loss": 0.3301,
+      "step": 84
+    },
+    {
+      "epoch": 0.0544,
+      "grad_norm": 0.49271462048539344,
+      "learning_rate": 4.99985976274423e-05,
+      "loss": 0.3271,
+      "step": 85
+    },
+    {
+      "epoch": 0.05504,
+      "grad_norm": 0.505091627450717,
+      "learning_rate": 4.9997980591824126e-05,
+      "loss": 0.308,
+      "step": 86
+    },
+    {
+      "epoch": 0.05568,
+      "grad_norm": 0.3949565660235225,
+      "learning_rate": 4.999725137445674e-05,
+      "loss": 0.3185,
+      "step": 87
+    },
+    {
+      "epoch": 0.05632,
+      "grad_norm": 0.5337665218445341,
+      "learning_rate": 4.999640997861264e-05,
+      "loss": 0.3211,
+      "step": 88
+    },
+    {
+      "epoch": 0.05696,
+      "grad_norm": 0.45754751119526205,
+      "learning_rate": 4.999545640806766e-05,
+      "loss": 0.3228,
+      "step": 89
+    },
+    {
+      "epoch": 0.0576,
+      "grad_norm": 0.45385353033639203,
+      "learning_rate": 4.999439066710111e-05,
+      "loss": 0.3415,
+      "step": 90
+    },
+    {
+      "epoch": 0.05824,
+      "grad_norm": 0.49073962043136393,
+      "learning_rate": 4.999321276049563e-05,
+      "loss": 0.3179,
+      "step": 91
+    },
+    {
+      "epoch": 0.05888,
+      "grad_norm": 0.42999497815595844,
+      "learning_rate": 4.9991922693537226e-05,
+      "loss": 0.3067,
+      "step": 92
+    },
+    {
+      "epoch": 0.05952,
+      "grad_norm": 0.4625502305860712,
+      "learning_rate": 4.9990520472015246e-05,
+      "loss": 0.2949,
+      "step": 93
+    },
+    {
+      "epoch": 0.06016,
+      "grad_norm": 0.40226854319567656,
+      "learning_rate": 4.998900610222236e-05,
+      "loss": 0.3051,
+      "step": 94
+    },
+    {
+      "epoch": 0.0608,
+      "grad_norm": 0.39982821508119915,
+      "learning_rate": 4.998737959095449e-05,
+      "loss": 0.2778,
+      "step": 95
+    },
+    {
+      "epoch": 0.06144,
+      "grad_norm": 1.0464838609726888,
+      "learning_rate": 4.998564094551081e-05,
+      "loss": 0.3161,
+      "step": 96
+    },
+    {
+      "epoch": 0.06208,
+      "grad_norm": 0.4514203583299103,
+      "learning_rate": 4.998379017369375e-05,
+      "loss": 0.3119,
+      "step": 97
+    },
+    {
+      "epoch": 0.06272,
+      "grad_norm": 0.48265032091647747,
+      "learning_rate": 4.9981827283808855e-05,
+      "loss": 0.3083,
+      "step": 98
+    },
+    {
+      "epoch": 0.06336,
+      "grad_norm": 0.42207276430371016,
+      "learning_rate": 4.997975228466489e-05,
+      "loss": 0.335,
+      "step": 99
+    },
+    {
+      "epoch": 0.064,
+      "grad_norm": 0.4109136525662046,
+      "learning_rate": 4.997756518557368e-05,
+      "loss": 0.2801,
+      "step": 100
+    },
+    {
+      "epoch": 0.06464,
+      "grad_norm": 0.40459874238442717,
+      "learning_rate": 4.997526599635011e-05,
+      "loss": 0.2852,
+      "step": 101
+    },
+    {
+      "epoch": 0.06528,
+      "grad_norm": 0.48908720050476456,
+      "learning_rate": 4.997285472731211e-05,
+      "loss": 0.335,
+      "step": 102
+    },
+    {
+      "epoch": 0.06592,
+      "grad_norm": 0.42083268237680455,
+      "learning_rate": 4.997033138928056e-05,
+      "loss": 0.3143,
+      "step": 103
+    },
+    {
+      "epoch": 0.06656,
+      "grad_norm": 0.45257051558741634,
+      "learning_rate": 4.996769599357927e-05,
+      "loss": 0.309,
+      "step": 104
+    },
+    {
+      "epoch": 0.0672,
+      "grad_norm": 0.43169779281959547,
+      "learning_rate": 4.996494855203493e-05,
+      "loss": 0.3239,
+      "step": 105
+    },
+    {
+      "epoch": 0.06784,
+      "grad_norm": 0.5600681218655813,
+      "learning_rate": 4.996208907697706e-05,
+      "loss": 0.3218,
+      "step": 106
+    },
+    {
+      "epoch": 0.06848,
+      "grad_norm": 0.5148554799775189,
+      "learning_rate": 4.995911758123791e-05,
+      "loss": 0.2851,
+      "step": 107
+    },
+    {
+      "epoch": 0.06912,
+      "grad_norm": 1.6520154087150252,
+      "learning_rate": 4.995603407815249e-05,
+      "loss": 0.2973,
+      "step": 108
+    },
+    {
+      "epoch": 0.06976,
+      "grad_norm": 0.560865604631309,
+      "learning_rate": 4.9952838581558406e-05,
+      "loss": 0.3056,
+      "step": 109
+    },
+    {
+      "epoch": 0.0704,
+      "grad_norm": 0.45796131093813686,
+      "learning_rate": 4.994953110579588e-05,
+      "loss": 0.3151,
+      "step": 110
+    },
+    {
+      "epoch": 0.07104,
+      "grad_norm": 0.9637610478065832,
+      "learning_rate": 4.9946111665707645e-05,
+      "loss": 0.3279,
+      "step": 111
+    },
+    {
+      "epoch": 0.07168,
+      "grad_norm": 0.364055859969717,
+      "learning_rate": 4.99425802766389e-05,
+      "loss": 0.2899,
+      "step": 112
+    },
+    {
+      "epoch": 0.07232,
+      "grad_norm": 0.4976556238289915,
+      "learning_rate": 4.993893695443721e-05,
+      "loss": 0.3031,
+      "step": 113
+    },
+    {
+      "epoch": 0.07296,
+      "grad_norm": 0.43623766283866644,
+      "learning_rate": 4.993518171545247e-05,
+      "loss": 0.2915,
+      "step": 114
+    },
+    {
+      "epoch": 0.0736,
+      "grad_norm": 0.4725640071790753,
+      "learning_rate": 4.993131457653682e-05,
+      "loss": 0.3027,
+      "step": 115
+    },
+    {
+      "epoch": 0.07424,
+      "grad_norm": 0.4338862090037202,
+      "learning_rate": 4.992733555504454e-05,
+      "loss": 0.3179,
+      "step": 116
+    },
+    {
+      "epoch": 0.07488,
+      "grad_norm": 0.46363028669830225,
+      "learning_rate": 4.9923244668832025e-05,
+      "loss": 0.3075,
+      "step": 117
+    },
+    {
+      "epoch": 0.07552,
+      "grad_norm": 0.4285975451064772,
+      "learning_rate": 4.991904193625768e-05,
+      "loss": 0.3117,
+      "step": 118
+    },
+    {
+      "epoch": 0.07616,
+      "grad_norm": 0.37832077577217255,
+      "learning_rate": 4.991472737618181e-05,
+      "loss": 0.2741,
+      "step": 119
+    },
+    {
+      "epoch": 0.0768,
+      "grad_norm": 0.41682487577691063,
+      "learning_rate": 4.991030100796659e-05,
+      "loss": 0.3019,
+      "step": 120
+    },
+    {
+      "epoch": 0.07744,
+      "grad_norm": 0.3716575375425735,
+      "learning_rate": 4.9905762851475915e-05,
+      "loss": 0.3154,
+      "step": 121
+    },
+    {
+      "epoch": 0.07808,
+      "grad_norm": 0.39573210518813595,
+      "learning_rate": 4.990111292707537e-05,
+      "loss": 0.3141,
+      "step": 122
+    },
+    {
+      "epoch": 0.07872,
+      "grad_norm": 0.4590220102268012,
+      "learning_rate": 4.989635125563211e-05,
+      "loss": 0.2911,
+      "step": 123
+    },
+    {
+      "epoch": 0.07936,
+      "grad_norm": 0.37422052213335605,
+      "learning_rate": 4.989147785851477e-05,
+      "loss": 0.3101,
+      "step": 124
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.4283999831025601,
+      "learning_rate": 4.988649275759335e-05,
+      "loss": 0.32,
+      "step": 125
+    },
+    {
+      "epoch": 0.08064,
+      "grad_norm": 0.4133286791179184,
+      "learning_rate": 4.988139597523916e-05,
+      "loss": 0.2921,
+      "step": 126
+    },
+    {
+      "epoch": 0.08128,
+      "grad_norm": 0.4338853261587351,
+      "learning_rate": 4.98761875343247e-05,
+      "loss": 0.2846,
+      "step": 127
+    },
+    {
+      "epoch": 0.08192,
+      "grad_norm": 0.40905061401860165,
+      "learning_rate": 4.987086745822352e-05,
+      "loss": 0.2859,
+      "step": 128
+    },
+    {
+      "epoch": 0.08256,
+      "grad_norm": 0.39198335440260496,
+      "learning_rate": 4.986543577081017e-05,
+      "loss": 0.3214,
+      "step": 129
+    },
+    {
+      "epoch": 0.0832,
+      "grad_norm": 0.4072147961679465,
+      "learning_rate": 4.985989249646007e-05,
+      "loss": 0.303,
+      "step": 130
+    },
+    {
+      "epoch": 0.08384,
+      "grad_norm": 0.3960141589346821,
+      "learning_rate": 4.985423766004939e-05,
+      "loss": 0.3113,
+      "step": 131
+    },
+    {
+      "epoch": 0.08448,
+      "grad_norm": 0.4901660833242505,
+      "learning_rate": 4.9848471286954965e-05,
+      "loss": 0.3275,
+      "step": 132
+    },
+    {
+      "epoch": 0.08512,
+      "grad_norm": 0.3589215784625634,
+      "learning_rate": 4.984259340305416e-05,
+      "loss": 0.2894,
+      "step": 133
+    },
+    {
+      "epoch": 0.08576,
+      "grad_norm": 0.3998417320795738,
+      "learning_rate": 4.9836604034724753e-05,
+      "loss": 0.2926,
+      "step": 134
+    },
+    {
+      "epoch": 0.0864,
+      "grad_norm": 0.4296580917834302,
+      "learning_rate": 4.9830503208844835e-05,
+      "loss": 0.2946,
+      "step": 135
+    },
+    {
+      "epoch": 0.08704,
+      "grad_norm": 0.3544925944429582,
+      "learning_rate": 4.982429095279266e-05,
+      "loss": 0.2945,
+      "step": 136
+    },
+    {
+      "epoch": 0.08768,
+      "grad_norm": 0.4295052142302722,
+      "learning_rate": 4.981796729444655e-05,
+      "loss": 0.3251,
+      "step": 137
+    },
+    {
+      "epoch": 0.08832,
+      "grad_norm": 0.3810698917249912,
+      "learning_rate": 4.981153226218477e-05,
+      "loss": 0.3075,
+      "step": 138
+    },
+    {
+      "epoch": 0.08896,
+      "grad_norm": 0.3415977052001699,
+      "learning_rate": 4.9804985884885366e-05,
+      "loss": 0.2879,
+      "step": 139
+    },
+    {
+      "epoch": 0.0896,
+      "grad_norm": 0.3776850662597046,
+      "learning_rate": 4.979832819192608e-05,
+      "loss": 0.3129,
+      "step": 140
+    },
+    {
+      "epoch": 0.09024,
+      "grad_norm": 0.36338720848583483,
+      "learning_rate": 4.97915592131842e-05,
+      "loss": 0.2894,
+      "step": 141
+    },
+    {
+      "epoch": 0.09088,
+      "grad_norm": 0.39018083120154434,
+      "learning_rate": 4.97846789790364e-05,
+      "loss": 0.3073,
+      "step": 142
+    },
+    {
+      "epoch": 0.09152,
+      "grad_norm": 0.35656151957057863,
+      "learning_rate": 4.977768752035865e-05,
+      "loss": 0.3119,
+      "step": 143
+    },
+    {
+      "epoch": 0.09216,
+      "grad_norm": 0.3952926600732054,
+      "learning_rate": 4.9770584868526046e-05,
+      "loss": 0.312,
+      "step": 144
+    },
+    {
+      "epoch": 0.0928,
+      "grad_norm": 0.4288986528346642,
+      "learning_rate": 4.9763371055412685e-05,
+      "loss": 0.3084,
+      "step": 145
+    },
+    {
+      "epoch": 0.09344,
+      "grad_norm": 0.44696008744325466,
+      "learning_rate": 4.9756046113391496e-05,
+      "loss": 0.3292,
+      "step": 146
+    },
+    {
+      "epoch": 0.09408,
+      "grad_norm": 0.35163986650006296,
+      "learning_rate": 4.974861007533413e-05,
+      "loss": 0.2859,
+      "step": 147
+    },
+    {
+      "epoch": 0.09472,
+      "grad_norm": 0.42759675331208585,
+      "learning_rate": 4.97410629746108e-05,
+      "loss": 0.2919,
+      "step": 148
+    },
+    {
+      "epoch": 0.09536,
+      "grad_norm": 0.43529839553806815,
+      "learning_rate": 4.9733404845090116e-05,
+      "loss": 0.2842,
+      "step": 149
+    },
+    {
+      "epoch": 0.096,
+      "grad_norm": 0.41401104367991204,
+      "learning_rate": 4.972563572113894e-05,
+      "loss": 0.2904,
+      "step": 150
+    },
+    {
+      "epoch": 0.09664,
+      "grad_norm": 0.41056372455002155,
+      "learning_rate": 4.971775563762226e-05,
+      "loss": 0.291,
+      "step": 151
+    },
+    {
+      "epoch": 0.09728,
+      "grad_norm": 0.42450110637249844,
+      "learning_rate": 4.970976462990299e-05,
+      "loss": 0.3107,
+      "step": 152
+    },
+    {
+      "epoch": 0.09792,
+      "grad_norm": 0.36252875803102586,
+      "learning_rate": 4.970166273384183e-05,
+      "loss": 0.2867,
+      "step": 153
+    },
+    {
+      "epoch": 0.09856,
+      "grad_norm": 0.34724268049235696,
+      "learning_rate": 4.969344998579712e-05,
+      "loss": 0.2663,
+      "step": 154
+    },
+    {
+      "epoch": 0.0992,
+      "grad_norm": 0.4062042219778404,
+      "learning_rate": 4.9685126422624644e-05,
+      "loss": 0.3156,
+      "step": 155
+    },
+    {
+      "epoch": 0.09984,
+      "grad_norm": 0.38599214382639446,
+      "learning_rate": 4.96766920816775e-05,
+      "loss": 0.3147,
+      "step": 156
+    },
+    {
+      "epoch": 0.10048,
+      "grad_norm": 0.3662678782254152,
+      "learning_rate": 4.966814700080592e-05,
+      "loss": 0.3184,
+      "step": 157
+    },
+    {
+      "epoch": 0.10112,
+      "grad_norm": 0.37758974795094674,
+      "learning_rate": 4.965949121835707e-05,
+      "loss": 0.3341,
+      "step": 158
+    },
+    {
+      "epoch": 0.10176,
+      "grad_norm": 0.42335782183402376,
+      "learning_rate": 4.965072477317493e-05,
+      "loss": 0.3105,
+      "step": 159
+    },
+    {
+      "epoch": 0.1024,
+      "grad_norm": 0.3445716615812455,
+      "learning_rate": 4.964184770460007e-05,
+      "loss": 0.2935,
+      "step": 160
+    },
+    {
+      "epoch": 0.10304,
+      "grad_norm": 0.40006390954972004,
+      "learning_rate": 4.963286005246953e-05,
+      "loss": 0.2834,
+      "step": 161
+    },
+    {
+      "epoch": 0.10368,
+      "grad_norm": 0.3664434351809571,
+      "learning_rate": 4.9623761857116594e-05,
+      "loss": 0.3191,
+      "step": 162
+    },
+    {
+      "epoch": 0.10432,
+      "grad_norm": 0.3746346471394553,
+      "learning_rate": 4.961455315937059e-05,
+      "loss": 0.282,
+      "step": 163
+    },
+    {
+      "epoch": 0.10496,
+      "grad_norm": 0.3543125314610225,
+      "learning_rate": 4.960523400055679e-05,
+      "loss": 0.2939,
+      "step": 164
+    },
+    {
+      "epoch": 0.1056,
+      "grad_norm": 0.34883234374623584,
+      "learning_rate": 4.959580442249615e-05,
+      "loss": 0.3011,
+      "step": 165
+    },
+    {
+      "epoch": 0.10624,
+      "grad_norm": 0.33879752208803143,
+      "learning_rate": 4.958626446750515e-05,
+      "loss": 0.3211,
+      "step": 166
+    },
+    {
+      "epoch": 0.10688,
+      "grad_norm": 0.37300311895352173,
+      "learning_rate": 4.9576614178395595e-05,
+      "loss": 0.301,
+      "step": 167
+    },
+    {
+      "epoch": 0.10752,
+      "grad_norm": 0.3273073442908486,
+      "learning_rate": 4.956685359847445e-05,
+      "loss": 0.2969,
+      "step": 168
+    },
+    {
+      "epoch": 0.10816,
+      "grad_norm": 0.3116171444679886,
+      "learning_rate": 4.95569827715436e-05,
+      "loss": 0.2817,
+      "step": 169
+    },
+    {
+      "epoch": 0.1088,
+      "grad_norm": 0.3361243539743496,
+      "learning_rate": 4.9547001741899693e-05,
+      "loss": 0.2977,
+      "step": 170
+    },
+    {
+      "epoch": 0.10944,
+      "grad_norm": 0.3418189113063408,
+      "learning_rate": 4.953691055433391e-05,
+      "loss": 0.2916,
+      "step": 171
+    },
+    {
+      "epoch": 0.11008,
+      "grad_norm": 0.3625913676240481,
+      "learning_rate": 4.952670925413182e-05,
+      "loss": 0.3038,
+      "step": 172
+    },
+    {
+      "epoch": 0.11072,
+      "grad_norm": 0.34636880625513666,
+      "learning_rate": 4.951639788707307e-05,
+      "loss": 0.2854,
+      "step": 173
+    },
+    {
+      "epoch": 0.11136,
+      "grad_norm": 0.35977242712508983,
+      "learning_rate": 4.950597649943132e-05,
+      "loss": 0.2786,
+      "step": 174
+    },
+    {
+      "epoch": 0.112,
+      "grad_norm": 0.33680285828380363,
+      "learning_rate": 4.9495445137973905e-05,
+      "loss": 0.2934,
+      "step": 175
+    },
+    {
+      "epoch": 0.11264,
+      "grad_norm": 0.3270580360967154,
+      "learning_rate": 4.948480384996172e-05,
+      "loss": 0.2878,
+      "step": 176
+    },
+    {
+      "epoch": 0.11328,
+      "grad_norm": 0.3363028989990994,
+      "learning_rate": 4.947405268314894e-05,
+      "loss": 0.3014,
+      "step": 177
+    },
+    {
+      "epoch": 0.11392,
+      "grad_norm": 0.39424588001597727,
+      "learning_rate": 4.9463191685782865e-05,
+      "loss": 0.3076,
+      "step": 178
+    },
+    {
+      "epoch": 0.11456,
+      "grad_norm": 0.31183590038965797,
+      "learning_rate": 4.945222090660364e-05,
+      "loss": 0.2859,
+      "step": 179
+    },
+    {
+      "epoch": 0.1152,
+      "grad_norm": 0.3487969513905697,
+      "learning_rate": 4.944114039484411e-05,
+      "loss": 0.3057,
+      "step": 180
+    },
+    {
+      "epoch": 0.11584,
+      "grad_norm": 0.3502275557092516,
+      "learning_rate": 4.942995020022954e-05,
+      "loss": 0.3255,
+      "step": 181
+    },
+    {
+      "epoch": 0.11648,
+      "grad_norm": 0.3051981568928157,
+      "learning_rate": 4.9418650372977384e-05,
+      "loss": 0.2881,
+      "step": 182
+    },
+    {
+      "epoch": 0.11712,
+      "grad_norm": 0.3213236160412315,
+      "learning_rate": 4.940724096379714e-05,
+      "loss": 0.2832,
+      "step": 183
+    },
+    {
+      "epoch": 0.11776,
+      "grad_norm": 0.2804638187646734,
+      "learning_rate": 4.9395722023890035e-05,
+      "loss": 0.2807,
+      "step": 184
+    },
+    {
+      "epoch": 0.1184,
+      "grad_norm": 0.32770405872721187,
+      "learning_rate": 4.938409360494883e-05,
+      "loss": 0.2843,
+      "step": 185
+    },
+    {
+      "epoch": 0.11904,
+      "grad_norm": 0.3196607887549686,
+      "learning_rate": 4.937235575915761e-05,
+      "loss": 0.2928,
+      "step": 186
+    },
+    {
+      "epoch": 0.11968,
+      "grad_norm": 0.31014707115252577,
+      "learning_rate": 4.93605085391915e-05,
+      "loss": 0.2926,
+      "step": 187
+    },
+    {
+      "epoch": 0.12032,
+      "grad_norm": 0.3667018178707909,
+      "learning_rate": 4.934855199821647e-05,
+      "loss": 0.2837,
+      "step": 188
+    },
+    {
+      "epoch": 0.12096,
+      "grad_norm": 0.3370594761880159,
+      "learning_rate": 4.93364861898891e-05,
+      "loss": 0.2776,
+      "step": 189
+    },
+    {
+      "epoch": 0.1216,
+      "grad_norm": 0.3889677526233924,
+      "learning_rate": 4.932431116835628e-05,
+      "loss": 0.2976,
+      "step": 190
+    },
+    {
+      "epoch": 0.12224,
+      "grad_norm": 0.32895998318993375,
+      "learning_rate": 4.931202698825503e-05,
+      "loss": 0.2823,
+      "step": 191
+    },
+    {
+      "epoch": 0.12288,
+      "grad_norm": 0.30268716856662087,
+      "learning_rate": 4.929963370471226e-05,
+      "loss": 0.2889,
+      "step": 192
+    },
+    {
+      "epoch": 0.12352,
+      "grad_norm": 0.2784816493323616,
+      "learning_rate": 4.928713137334445e-05,
+      "loss": 0.2819,
+      "step": 193
+    },
+    {
+      "epoch": 0.12416,
+      "grad_norm": 0.31652215898467684,
+      "learning_rate": 4.927452005025748e-05,
+      "loss": 0.3023,
+      "step": 194
+    },
+    {
+      "epoch": 0.1248,
+      "grad_norm": 0.30644827432002086,
+      "learning_rate": 4.926179979204633e-05,
+      "loss": 0.303,
+      "step": 195
+    },
+    {
+      "epoch": 0.12544,
+      "grad_norm": 0.298870893997602,
+      "learning_rate": 4.9248970655794836e-05,
+      "loss": 0.2924,
+      "step": 196
+    },
+    {
+      "epoch": 0.12608,
+      "grad_norm": 0.3019490103946191,
+      "learning_rate": 4.923603269907545e-05,
+      "loss": 0.2749,
+      "step": 197
+    },
+    {
+      "epoch": 0.12672,
+      "grad_norm": 0.34179704436331393,
+      "learning_rate": 4.922298597994897e-05,
+      "loss": 0.2837,
+      "step": 198
+    },
+    {
+      "epoch": 0.12736,
+      "grad_norm": 0.34714244369014874,
+      "learning_rate": 4.920983055696428e-05,
+      "loss": 0.2941,
+      "step": 199
+    },
+    {
+      "epoch": 0.128,
+      "grad_norm": 0.34514991924274774,
+      "learning_rate": 4.919656648915807e-05,
+      "loss": 0.2852,
+      "step": 200
+    },
+    {
+      "epoch": 0.12864,
+      "grad_norm": 0.2845599679176182,
+      "learning_rate": 4.918319383605462e-05,
+      "loss": 0.2583,
+      "step": 201
+    },
+    {
+      "epoch": 0.12928,
+      "grad_norm": 0.3683838687495882,
+      "learning_rate": 4.916971265766547e-05,
+      "loss": 0.2954,
+      "step": 202
+    },
+    {
+      "epoch": 0.12992,
+      "grad_norm": 0.35457325941860496,
+      "learning_rate": 4.915612301448919e-05,
+      "loss": 0.2957,
+      "step": 203
+    },
+    {
+      "epoch": 0.13056,
+      "grad_norm": 0.4102065495806397,
+      "learning_rate": 4.914242496751111e-05,
+      "loss": 0.289,
+      "step": 204
+    },
+    {
+      "epoch": 0.1312,
+      "grad_norm": 0.391931082619993,
+      "learning_rate": 4.912861857820303e-05,
+      "loss": 0.303,
+      "step": 205
+    },
+    {
+      "epoch": 0.13184,
+      "grad_norm": 0.37855393422147743,
+      "learning_rate": 4.911470390852295e-05,
+      "loss": 0.3013,
+      "step": 206
+    },
+    {
+      "epoch": 0.13248,
+      "grad_norm": 0.31160560780141033,
+      "learning_rate": 4.91006810209148e-05,
+      "loss": 0.2888,
+      "step": 207
+    },
+    {
+      "epoch": 0.13312,
+      "grad_norm": 0.38185424339716356,
+      "learning_rate": 4.9086549978308125e-05,
+      "loss": 0.2999,
+      "step": 208
+    },
+    {
+      "epoch": 0.13376,
+      "grad_norm": 0.30532616858387496,
+      "learning_rate": 4.9072310844117875e-05,
+      "loss": 0.2732,
+      "step": 209
+    },
+    {
+      "epoch": 0.1344,
+      "grad_norm": 0.36508067170966735,
+      "learning_rate": 4.9057963682244045e-05,
+      "loss": 0.2909,
+      "step": 210
+    },
+    {
+      "epoch": 0.13504,
+      "grad_norm": 0.3148112484256156,
+      "learning_rate": 4.904350855707142e-05,
+      "loss": 0.2874,
+      "step": 211
+    },
+    {
+      "epoch": 0.13568,
+      "grad_norm": 0.3274325146184032,
+      "learning_rate": 4.9028945533469276e-05,
+      "loss": 0.2661,
+      "step": 212
+    },
+    {
+      "epoch": 0.13632,
+      "grad_norm": 0.3685996593265165,
+      "learning_rate": 4.9014274676791134e-05,
+      "loss": 0.3037,
+      "step": 213
+    },
+    {
+      "epoch": 0.13696,
+      "grad_norm": 0.34087816560391343,
+      "learning_rate": 4.899949605287439e-05,
+      "loss": 0.2797,
+      "step": 214
+    },
+    {
+      "epoch": 0.1376,
+      "grad_norm": 0.3613310322508522,
+      "learning_rate": 4.898460972804009e-05,
+      "loss": 0.3096,
+      "step": 215
+    },
+    {
+      "epoch": 0.13824,
+      "grad_norm": 0.375457846044839,
+      "learning_rate": 4.896961576909258e-05,
+      "loss": 0.2822,
+      "step": 216
+    },
+    {
+      "epoch": 0.13888,
+      "grad_norm": 0.3437437239754465,
+      "learning_rate": 4.895451424331924e-05,
+      "loss": 0.2725,
+      "step": 217
+    },
+    {
+      "epoch": 0.13952,
+      "grad_norm": 0.35840101574935934,
+      "learning_rate": 4.893930521849016e-05,
+      "loss": 0.268,
+      "step": 218
+    },
+    {
+      "epoch": 0.14016,
+      "grad_norm": 0.3767989948671444,
+      "learning_rate": 4.892398876285785e-05,
+      "loss": 0.2901,
+      "step": 219
+    },
+    {
+      "epoch": 0.1408,
+      "grad_norm": 0.4065322507872538,
+      "learning_rate": 4.890856494515696e-05,
+      "loss": 0.2892,
+      "step": 220
+    },
+    {
+      "epoch": 0.14144,
+      "grad_norm": 0.3526522553070582,
+      "learning_rate": 4.889303383460389e-05,
+      "loss": 0.2818,
+      "step": 221
+    },
+    {
+      "epoch": 0.14208,
+      "grad_norm": 0.36687661566425206,
+      "learning_rate": 4.887739550089656e-05,
+      "loss": 0.2788,
+      "step": 222
+    },
+    {
+      "epoch": 0.14272,
+      "grad_norm": 0.36376035976937565,
+      "learning_rate": 4.886165001421408e-05,
+      "loss": 0.3102,
+      "step": 223
+    },
+    {
+      "epoch": 0.14336,
+      "grad_norm": 0.32605502970392203,
+      "learning_rate": 4.884579744521638e-05,
+      "loss": 0.2858,
+      "step": 224
+    },
+    {
+      "epoch": 0.144,
+      "grad_norm": 0.3473388829365213,
+      "learning_rate": 4.8829837865043984e-05,
+      "loss": 0.2927,
+      "step": 225
+    },
+    {
+      "epoch": 0.14464,
+      "grad_norm": 0.3161945789520979,
+      "learning_rate": 4.881377134531762e-05,
+      "loss": 0.2684,
+      "step": 226
+    },
+    {
+      "epoch": 0.14528,
+      "grad_norm": 0.3233058980574442,
+      "learning_rate": 4.8797597958137914e-05,
+      "loss": 0.2809,
+      "step": 227
+    },
+    {
+      "epoch": 0.14592,
+      "grad_norm": 0.35628813657982605,
+      "learning_rate": 4.878131777608509e-05,
+      "loss": 0.2857,
+      "step": 228
+    },
+    {
+      "epoch": 0.14656,
+      "grad_norm": 0.34474317130830845,
+      "learning_rate": 4.876493087221862e-05,
+      "loss": 0.2981,
+      "step": 229
+    },
+    {
+      "epoch": 0.1472,
+      "grad_norm": 0.3332723948946639,
+      "learning_rate": 4.87484373200769e-05,
+      "loss": 0.2904,
+      "step": 230
+    },
+    {
+      "epoch": 0.14784,
+      "grad_norm": 0.36212328347879097,
+      "learning_rate": 4.8731837193676944e-05,
+      "loss": 0.2803,
+      "step": 231
+    },
+    {
+      "epoch": 0.14848,
+      "grad_norm": 0.32285084742216597,
+      "learning_rate": 4.871513056751401e-05,
+      "loss": 0.2991,
+      "step": 232
+    },
+    {
+      "epoch": 0.14912,
+      "grad_norm": 0.3757428283729143,
+      "learning_rate": 4.8698317516561295e-05,
+      "loss": 0.3007,
+      "step": 233
+    },
+    {
+      "epoch": 0.14976,
+      "grad_norm": 0.352925038021312,
+      "learning_rate": 4.8681398116269596e-05,
+      "loss": 0.2865,
+      "step": 234
+    },
+    {
+      "epoch": 0.1504,
+      "grad_norm": 0.33841490625531084,
+      "learning_rate": 4.8664372442566954e-05,
+      "loss": 0.294,
+      "step": 235
+    },
+    {
+      "epoch": 0.15104,
+      "grad_norm": 0.34183505785167,
+      "learning_rate": 4.864724057185834e-05,
+      "loss": 0.2691,
+      "step": 236
+    },
+    {
+      "epoch": 0.15168,
+      "grad_norm": 0.339769370877407,
+      "learning_rate": 4.86300025810253e-05,
+      "loss": 0.2818,
+      "step": 237
+    },
+    {
+      "epoch": 0.15232,
+      "grad_norm": 0.30700076377351593,
+      "learning_rate": 4.861265854742559e-05,
+      "loss": 0.2748,
+      "step": 238
+    },
+    {
+      "epoch": 0.15296,
+      "grad_norm": 0.3489951897607549,
+      "learning_rate": 4.859520854889287e-05,
+      "loss": 0.2937,
+      "step": 239
+    },
+    {
+      "epoch": 0.1536,
+      "grad_norm": 0.29156819501858466,
+      "learning_rate": 4.857765266373631e-05,
+      "loss": 0.2958,
+      "step": 240
+    },
+    {
+      "epoch": 0.15424,
+      "grad_norm": 0.3384643439513567,
+      "learning_rate": 4.855999097074027e-05,
+      "loss": 0.2939,
+      "step": 241
+    },
+    {
+      "epoch": 0.15488,
+      "grad_norm": 0.3077298938535667,
+      "learning_rate": 4.8542223549163946e-05,
+      "loss": 0.2724,
+      "step": 242
+    },
+    {
+      "epoch": 0.15552,
+      "grad_norm": 0.31947348723786806,
+      "learning_rate": 4.8524350478740985e-05,
+      "loss": 0.2666,
+      "step": 243
+    },
+    {
+      "epoch": 0.15616,
+      "grad_norm": 0.3774412893720739,
+      "learning_rate": 4.850637183967917e-05,
+      "loss": 0.2852,
+      "step": 244
+    },
+    {
+      "epoch": 0.1568,
+      "grad_norm": 0.3066835259607928,
+      "learning_rate": 4.848828771266002e-05,
+      "loss": 0.2883,
+      "step": 245
+    },
+    {
+      "epoch": 0.15744,
+      "grad_norm": 0.39290641044610064,
+      "learning_rate": 4.847009817883844e-05,
+      "loss": 0.284,
+      "step": 246
+    },
+    {
+      "epoch": 0.15808,
+      "grad_norm": 0.3087733867875181,
+      "learning_rate": 4.845180331984239e-05,
+      "loss": 0.2726,
+      "step": 247
+    },
+    {
+      "epoch": 0.15872,
+      "grad_norm": 0.3823066055668157,
+      "learning_rate": 4.843340321777246e-05,
+      "loss": 0.2656,
+      "step": 248
+    },
+    {
+      "epoch": 0.15936,
+      "grad_norm": 0.3211297194539539,
+      "learning_rate": 4.8414897955201556e-05,
+      "loss": 0.2628,
+      "step": 249
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.3120564620889703,
+      "learning_rate": 4.8396287615174484e-05,
+      "loss": 0.2587,
+      "step": 250
+    },
+    {
+      "epoch": 0.16064,
+      "grad_norm": 0.3289936954141222,
+      "learning_rate": 4.837757228120762e-05,
+      "loss": 0.2896,
+      "step": 251
+    },
+    {
+      "epoch": 0.16128,
+      "grad_norm": 0.3188145095365673,
+      "learning_rate": 4.835875203728849e-05,
+      "loss": 0.2555,
+      "step": 252
+    },
+    {
+      "epoch": 0.16192,
+      "grad_norm": 0.3889796632457812,
+      "learning_rate": 4.8339826967875443e-05,
+      "loss": 0.2872,
+      "step": 253
+    },
+    {
+      "epoch": 0.16256,
+      "grad_norm": 0.381368538527566,
+      "learning_rate": 4.832079715789724e-05,
+      "loss": 0.2988,
+      "step": 254
+    },
+    {
+      "epoch": 0.1632,
+      "grad_norm": 0.36630965780444474,
+      "learning_rate": 4.830166269275267e-05,
+      "loss": 0.2906,
+      "step": 255
+    },
+    {
+      "epoch": 0.16384,
+      "grad_norm": 0.3552050245169036,
+      "learning_rate": 4.828242365831016e-05,
+      "loss": 0.2761,
+      "step": 256
+    },
+    {
+      "epoch": 0.16448,
+      "grad_norm": 0.37404437653877487,
+      "learning_rate": 4.8263080140907455e-05,
+      "loss": 0.2908,
+      "step": 257
+    },
+    {
+      "epoch": 0.16512,
+      "grad_norm": 0.3507329450496866,
+      "learning_rate": 4.824363222735114e-05,
+      "loss": 0.275,
+      "step": 258
+    },
+    {
+      "epoch": 0.16576,
+      "grad_norm": 0.37299320025474214,
+      "learning_rate": 4.82240800049163e-05,
+      "loss": 0.2801,
+      "step": 259
+    },
+    {
+      "epoch": 0.1664,
+      "grad_norm": 0.38754434557373796,
+      "learning_rate": 4.820442356134612e-05,
+      "loss": 0.2709,
+      "step": 260
+    },
+    {
+      "epoch": 0.16704,
+      "grad_norm": 0.33798911863787995,
+      "learning_rate": 4.8184662984851505e-05,
+      "loss": 0.3009,
+      "step": 261
+    },
+    {
+      "epoch": 0.16768,
+      "grad_norm": 0.3559262440354945,
+      "learning_rate": 4.816479836411065e-05,
+      "loss": 0.2835,
+      "step": 262
+    },
+    {
+      "epoch": 0.16832,
+      "grad_norm": 0.4158079436380025,
+      "learning_rate": 4.8144829788268674e-05,
+      "loss": 0.2954,
+      "step": 263
+    },
+    {
+      "epoch": 0.16896,
+      "grad_norm": 0.3603676966517829,
+      "learning_rate": 4.8124757346937214e-05,
+      "loss": 0.2637,
+      "step": 264
+    },
+    {
+      "epoch": 0.1696,
+      "grad_norm": 0.3288136132861365,
+      "learning_rate": 4.810458113019401e-05,
+      "loss": 0.2747,
+      "step": 265
+    },
+    {
+      "epoch": 0.17024,
+      "grad_norm": 0.3402933784236876,
+      "learning_rate": 4.8084301228582485e-05,
+      "loss": 0.2577,
+      "step": 266
+    },
+    {
+      "epoch": 0.17088,
+      "grad_norm": 0.36930227507381896,
+      "learning_rate": 4.8063917733111406e-05,
+      "loss": 0.3098,
+      "step": 267
+    },
+    {
+      "epoch": 0.17152,
+      "grad_norm": 0.309709429764299,
+      "learning_rate": 4.8043430735254404e-05,
+      "loss": 0.2808,
+      "step": 268
+    },
+    {
+      "epoch": 0.17216,
+      "grad_norm": 0.35813482945344854,
+      "learning_rate": 4.8022840326949595e-05,
+      "loss": 0.2812,
+      "step": 269
+    },
+    {
+      "epoch": 0.1728,
+      "grad_norm": 0.3631234651320861,
+      "learning_rate": 4.800214660059916e-05,
+      "loss": 0.2797,
+      "step": 270
+    },
+    {
+      "epoch": 0.17344,
+      "grad_norm": 0.3309005171733229,
+      "learning_rate": 4.798134964906894e-05,
+      "loss": 0.2867,
+      "step": 271
+    },
+    {
+      "epoch": 0.17408,
+      "grad_norm": 0.35715429838212853,
+      "learning_rate": 4.7960449565688e-05,
+      "loss": 0.2638,
+      "step": 272
+    },
+    {
+      "epoch": 0.17472,
+      "grad_norm": 0.3279629544159387,
+      "learning_rate": 4.793944644424826e-05,
+      "loss": 0.2855,
+      "step": 273
+    },
+    {
+      "epoch": 0.17536,
+      "grad_norm": 0.37196752238968245,
+      "learning_rate": 4.7918340379003985e-05,
+      "loss": 0.2875,
+      "step": 274
+    },
+    {
+      "epoch": 0.176,
+      "grad_norm": 0.31834632572626353,
+      "learning_rate": 4.7897131464671444e-05,
+      "loss": 0.3006,
+      "step": 275
+    },
+    {
+      "epoch": 0.17664,
+      "grad_norm": 0.34089694844336105,
+      "learning_rate": 4.787581979642845e-05,
+      "loss": 0.2557,
+      "step": 276
+    },
+    {
+      "epoch": 0.17728,
+      "grad_norm": 0.31522922339322745,
+      "learning_rate": 4.7854405469913944e-05,
+      "loss": 0.2741,
+      "step": 277
+    },
+    {
+      "epoch": 0.17792,
+      "grad_norm": 0.31669044816633024,
+      "learning_rate": 4.7832888581227545e-05,
+      "loss": 0.2766,
+      "step": 278
+    },
+    {
+      "epoch": 0.17856,
+      "grad_norm": 0.3460333243052226,
+      "learning_rate": 4.7811269226929135e-05,
+      "loss": 0.2802,
+      "step": 279
+    },
+    {
+      "epoch": 0.1792,
+      "grad_norm": 0.33307222568871414,
+      "learning_rate": 4.7789547504038443e-05,
+      "loss": 0.2696,
+      "step": 280
+    },
+    {
+      "epoch": 0.17984,
+      "grad_norm": 0.3326671192080615,
+      "learning_rate": 4.776772351003458e-05,
+      "loss": 0.2811,
+      "step": 281
+    },
+    {
+      "epoch": 0.18048,
+      "grad_norm": 0.32991380728873526,
+      "learning_rate": 4.774579734285559e-05,
+      "loss": 0.2878,
+      "step": 282
+    },
+    {
+      "epoch": 0.18112,
+      "grad_norm": 0.3247648689257416,
+      "learning_rate": 4.772376910089808e-05,
+      "loss": 0.2786,
+      "step": 283
+    },
+    {
+      "epoch": 0.18176,
+      "grad_norm": 0.3282388341197448,
+      "learning_rate": 4.770163888301668e-05,
+      "loss": 0.2708,
+      "step": 284
+    },
+    {
+      "epoch": 0.1824,
+      "grad_norm": 0.3314880902653025,
+      "learning_rate": 4.7679406788523685e-05,
+      "loss": 0.278,
+      "step": 285
+    },
+    {
+      "epoch": 0.18304,
+      "grad_norm": 0.3223575932382192,
+      "learning_rate": 4.765707291718856e-05,
+      "loss": 0.2752,
+      "step": 286
+    },
+    {
+      "epoch": 0.18368,
+      "grad_norm": 0.33183064717108757,
+      "learning_rate": 4.7634637369237526e-05,
+      "loss": 0.2772,
+      "step": 287
+    },
+    {
+      "epoch": 0.18432,
+      "grad_norm": 0.3408606960521383,
+      "learning_rate": 4.7612100245353045e-05,
+      "loss": 0.288,
+      "step": 288
+    },
+    {
+      "epoch": 0.18496,
+      "grad_norm": 0.35025830647742257,
+      "learning_rate": 4.758946164667347e-05,
+      "loss": 0.2908,
+      "step": 289
+    },
+    {
+      "epoch": 0.1856,
+      "grad_norm": 0.385301045406101,
+      "learning_rate": 4.7566721674792515e-05,
+      "loss": 0.279,
+      "step": 290
+    },
+    {
+      "epoch": 0.18624,
+      "grad_norm": 0.3365158470714505,
+      "learning_rate": 4.7543880431758805e-05,
+      "loss": 0.2785,
+      "step": 291
+    },
+    {
+      "epoch": 0.18688,
+      "grad_norm": 0.3520029833034493,
+      "learning_rate": 4.752093802007543e-05,
+      "loss": 0.3015,
+      "step": 292
+    },
+    {
+      "epoch": 0.18752,
+      "grad_norm": 0.30981353591353944,
+      "learning_rate": 4.7497894542699534e-05,
+      "loss": 0.2621,
+      "step": 293
+    },
+    {
+      "epoch": 0.18816,
+      "grad_norm": 0.33678488047130234,
+      "learning_rate": 4.747475010304175e-05,
+      "loss": 0.2809,
+      "step": 294
+    },
+    {
+      "epoch": 0.1888,
+      "grad_norm": 0.30138038658401767,
+      "learning_rate": 4.745150480496583e-05,
+      "loss": 0.2968,
+      "step": 295
+    },
+    {
+      "epoch": 0.18944,
+      "grad_norm": 0.32368771492417925,
+      "learning_rate": 4.742815875278811e-05,
+      "loss": 0.2947,
+      "step": 296
+    },
+    {
+      "epoch": 0.19008,
+      "grad_norm": 0.3523596181548406,
+      "learning_rate": 4.7404712051277115e-05,
+      "loss": 0.2563,
+      "step": 297
+    },
+    {
+      "epoch": 0.19072,
+      "grad_norm": 0.3476373280918836,
+      "learning_rate": 4.738116480565301e-05,
+      "loss": 0.2954,
+      "step": 298
+    },
+    {
+      "epoch": 0.19136,
+      "grad_norm": 0.3305958149949148,
+      "learning_rate": 4.7357517121587183e-05,
+      "loss": 0.2792,
+      "step": 299
+    },
+    {
+      "epoch": 0.192,
+      "grad_norm": 0.3260613594105097,
+      "learning_rate": 4.7333769105201735e-05,
+      "loss": 0.2435,
+      "step": 300
+    },
+    {
+      "epoch": 0.19264,
+      "grad_norm": 0.35007749460908505,
+      "learning_rate": 4.730992086306906e-05,
+      "loss": 0.2839,
+      "step": 301
+    },
+    {
+      "epoch": 0.19328,
+      "grad_norm": 0.3124453815680245,
+      "learning_rate": 4.728597250221129e-05,
+      "loss": 0.2835,
+      "step": 302
+    },
+    {
+      "epoch": 0.19392,
+      "grad_norm": 0.32999884470742785,
+      "learning_rate": 4.726192413009986e-05,
+      "loss": 0.2962,
+      "step": 303
+    },
+    {
+      "epoch": 0.19456,
+      "grad_norm": 0.3238134773271861,
+      "learning_rate": 4.723777585465503e-05,
+      "loss": 0.2938,
+      "step": 304
+    },
+    {
+      "epoch": 0.1952,
+      "grad_norm": 0.3561001066562547,
+      "learning_rate": 4.7213527784245395e-05,
+      "loss": 0.2772,
+      "step": 305
+    },
+    {
+      "epoch": 0.19584,
+      "grad_norm": 0.3207708243138028,
+      "learning_rate": 4.718918002768738e-05,
+      "loss": 0.2769,
+      "step": 306
+    },
+    {
+      "epoch": 0.19648,
+      "grad_norm": 0.2923323638292908,
+      "learning_rate": 4.716473269424476e-05,
+      "loss": 0.2842,
+      "step": 307
+    },
+    {
+      "epoch": 0.19712,
+      "grad_norm": 0.312740800108901,
+      "learning_rate": 4.714018589362818e-05,
+      "loss": 0.3113,
+      "step": 308
+    },
+    {
+      "epoch": 0.19776,
+      "grad_norm": 0.2764921423909185,
+      "learning_rate": 4.7115539735994666e-05,
+      "loss": 0.2735,
+      "step": 309
+    },
+    {
+      "epoch": 0.1984,
+      "grad_norm": 0.33245051095638717,
+      "learning_rate": 4.709079433194712e-05,
+      "loss": 0.2783,
+      "step": 310
+    },
+    {
+      "epoch": 0.19904,
+      "grad_norm": 0.2883468898345436,
+      "learning_rate": 4.7065949792533804e-05,
+      "loss": 0.257,
+      "step": 311
+    },
+    {
+      "epoch": 0.19968,
+      "grad_norm": 0.3635737080055347,
+      "learning_rate": 4.70410062292479e-05,
+      "loss": 0.2933,
+      "step": 312
+    },
+    {
+      "epoch": 0.20032,
+      "grad_norm": 0.3254742482108127,
+      "learning_rate": 4.701596375402694e-05,
+      "loss": 0.2933,
+      "step": 313
+    },
+    {
+      "epoch": 0.20096,
+      "grad_norm": 0.30265278788416133,
+      "learning_rate": 4.699082247925234e-05,
+      "loss": 0.3004,
+      "step": 314
+    },
+    {
+      "epoch": 0.2016,
+      "grad_norm": 0.3229708540790062,
+      "learning_rate": 4.6965582517748923e-05,
+      "loss": 0.2935,
+      "step": 315
+    },
+    {
+      "epoch": 0.20224,
+      "grad_norm": 0.2955440929603968,
+      "learning_rate": 4.694024398278434e-05,
+      "loss": 0.2889,
+      "step": 316
+    },
+    {
+      "epoch": 0.20288,
+      "grad_norm": 0.32188591609611733,
+      "learning_rate": 4.691480698806864e-05,
+      "loss": 0.3002,
+      "step": 317
+    },
+    {
+      "epoch": 0.20352,
+      "grad_norm": 0.27836417438079064,
+      "learning_rate": 4.6889271647753694e-05,
+      "loss": 0.2641,
+      "step": 318
+    },
+    {
+      "epoch": 0.20416,
+      "grad_norm": 0.30503117060048834,
+      "learning_rate": 4.686363807643272e-05,
+      "loss": 0.2557,
+      "step": 319
+    },
+    {
+      "epoch": 0.2048,
+      "grad_norm": 0.31391016617770173,
+      "learning_rate": 4.683790638913979e-05,
+      "loss": 0.3008,
+      "step": 320
+    },
+    {
+      "epoch": 0.20544,
+      "grad_norm": 0.30297458481431416,
+      "learning_rate": 4.681207670134923e-05,
+      "loss": 0.2855,
+      "step": 321
+    },
+    {
+      "epoch": 0.20608,
+      "grad_norm": 0.31472600754542884,
+      "learning_rate": 4.6786149128975196e-05,
+      "loss": 0.2775,
+      "step": 322
+    },
+    {
+      "epoch": 0.20672,
+      "grad_norm": 0.2973831240453251,
+      "learning_rate": 4.6760123788371114e-05,
+      "loss": 0.2822,
+      "step": 323
+    },
+    {
+      "epoch": 0.20736,
+      "grad_norm": 0.3325176485448801,
+      "learning_rate": 4.6734000796329125e-05,
+      "loss": 0.2844,
+      "step": 324
+    },
+    {
+      "epoch": 0.208,
+      "grad_norm": 0.2781447146189205,
+      "learning_rate": 4.670778027007964e-05,
+      "loss": 0.2972,
+      "step": 325
+    },
+    {
+      "epoch": 0.20864,
+      "grad_norm": 0.35102418370987387,
+      "learning_rate": 4.6681462327290725e-05,
+      "loss": 0.2674,
+      "step": 326
+    },
+    {
+      "epoch": 0.20928,
+      "grad_norm": 0.3031132755291398,
+      "learning_rate": 4.6655047086067636e-05,
+      "loss": 0.2783,
+      "step": 327
+    },
+    {
+      "epoch": 0.20992,
+      "grad_norm": 0.2914541802538774,
+      "learning_rate": 4.662853466495229e-05,
+      "loss": 0.2718,
+      "step": 328
+    },
+    {
+      "epoch": 0.21056,
+      "grad_norm": 0.2845108849079057,
+      "learning_rate": 4.660192518292266e-05,
+      "loss": 0.2701,
+      "step": 329
+    },
+    {
+      "epoch": 0.2112,
+      "grad_norm": 0.30612752816767386,
+      "learning_rate": 4.6575218759392336e-05,
+      "loss": 0.2706,
+      "step": 330
+    },
+    {
+      "epoch": 0.21184,
+      "grad_norm": 0.3343001012996569,
+      "learning_rate": 4.654841551420993e-05,
+      "loss": 0.2592,
+      "step": 331
+    },
+    {
+      "epoch": 0.21248,
+      "grad_norm": 0.28779385954712927,
+      "learning_rate": 4.6521515567658557e-05,
+      "loss": 0.2673,
+      "step": 332
+    },
+    {
+      "epoch": 0.21312,
+      "grad_norm": 0.3130712333478661,
+      "learning_rate": 4.6494519040455275e-05,
+      "loss": 0.2779,
+      "step": 333
+    },
+    {
+      "epoch": 0.21376,
+      "grad_norm": 0.29656018238823806,
+      "learning_rate": 4.6467426053750596e-05,
+      "loss": 0.2634,
+      "step": 334
+    },
+    {
+      "epoch": 0.2144,
+      "grad_norm": 0.29429809703015725,
+      "learning_rate": 4.644023672912788e-05,
+      "loss": 0.2771,
+      "step": 335
+    },
+    {
+      "epoch": 0.21504,
+      "grad_norm": 0.3092350735511281,
+      "learning_rate": 4.641295118860281e-05,
+      "loss": 0.2872,
+      "step": 336
+    },
+    {
+      "epoch": 0.21568,
+      "grad_norm": 0.25928563534355026,
+      "learning_rate": 4.638556955462289e-05,
+      "loss": 0.2602,
+      "step": 337
+    },
+    {
+      "epoch": 0.21632,
+      "grad_norm": 0.32248922953258674,
+      "learning_rate": 4.635809195006681e-05,
+      "loss": 0.2909,
+      "step": 338
+    },
+    {
+      "epoch": 0.21696,
+      "grad_norm": 0.3066112074635602,
+      "learning_rate": 4.633051849824396e-05,
+      "loss": 0.2579,
+      "step": 339
+    },
+    {
+      "epoch": 0.2176,
+      "grad_norm": 0.34512498246286594,
+      "learning_rate": 4.630284932289386e-05,
+      "loss": 0.2853,
+      "step": 340
+    },
+    {
+      "epoch": 0.21824,
+      "grad_norm": 0.32772544741225995,
+      "learning_rate": 4.6275084548185605e-05,
+      "loss": 0.2936,
+      "step": 341
+    },
+    {
+      "epoch": 0.21888,
+      "grad_norm": 0.30638642844295744,
+      "learning_rate": 4.6247224298717294e-05,
+      "loss": 0.255,
+      "step": 342
+    },
+    {
+      "epoch": 0.21952,
+      "grad_norm": 0.2948939777755622,
+      "learning_rate": 4.621926869951549e-05,
+      "loss": 0.2461,
+      "step": 343
+    },
+    {
+      "epoch": 0.22016,
+      "grad_norm": 0.3571814132315835,
+      "learning_rate": 4.619121787603464e-05,
+      "loss": 0.2749,
+      "step": 344
+    },
+    {
+      "epoch": 0.2208,
+      "grad_norm": 0.3019198830803027,
+      "learning_rate": 4.6163071954156546e-05,
+      "loss": 0.2764,
+      "step": 345
+    },
+    {
+      "epoch": 0.22144,
+      "grad_norm": 0.2947039259040435,
+      "learning_rate": 4.6134831060189743e-05,
+      "loss": 0.2824,
+      "step": 346
+    },
+    {
+      "epoch": 0.22208,
+      "grad_norm": 0.311517126488146,
+      "learning_rate": 4.610649532086899e-05,
+      "loss": 0.273,
+      "step": 347
+    },
+    {
+      "epoch": 0.22272,
+      "grad_norm": 0.3202495311575204,
+      "learning_rate": 4.607806486335468e-05,
+      "loss": 0.2892,
+      "step": 348
+    },
+    {
+      "epoch": 0.22336,
+      "grad_norm": 0.29593870703206343,
+      "learning_rate": 4.604953981523225e-05,
+      "loss": 0.2874,
+      "step": 349
+    },
+    {
+      "epoch": 0.224,
+      "grad_norm": 0.3046217268454308,
+      "learning_rate": 4.602092030451162e-05,
+      "loss": 0.2697,
+      "step": 350
+    },
+    {
+      "epoch": 0.22464,
+      "grad_norm": 0.32393845693805695,
+      "learning_rate": 4.599220645962665e-05,
+      "loss": 0.2857,
+      "step": 351
+    },
+    {
+      "epoch": 0.22528,
+      "grad_norm": 0.3054582480479615,
+      "learning_rate": 4.596339840943452e-05,
+      "loss": 0.2681,
+      "step": 352
+    },
+    {
+      "epoch": 0.22592,
+      "grad_norm": 0.28840335456316024,
+      "learning_rate": 4.5934496283215164e-05,
+      "loss": 0.2669,
+      "step": 353
+    },
+    {
+      "epoch": 0.22656,
+      "grad_norm": 0.2773362865048385,
+      "learning_rate": 4.590550021067072e-05,
+      "loss": 0.2579,
+      "step": 354
+    },
+    {
+      "epoch": 0.2272,
+      "grad_norm": 0.29887466103776417,
+      "learning_rate": 4.587641032192488e-05,
+      "loss": 0.2668,
+      "step": 355
+    },
+    {
+      "epoch": 0.22784,
+      "grad_norm": 0.3307188902703665,
+      "learning_rate": 4.58472267475224e-05,
+      "loss": 0.2819,
+      "step": 356
+    },
+    {
+      "epoch": 0.22848,
+      "grad_norm": 0.3390971905379334,
+      "learning_rate": 4.5817949618428405e-05,
+      "loss": 0.2678,
+      "step": 357
+    },
+    {
+      "epoch": 0.22912,
+      "grad_norm": 0.3013575293157085,
+      "learning_rate": 4.578857906602791e-05,
+      "loss": 0.2682,
+      "step": 358
+    },
+    {
+      "epoch": 0.22976,
+      "grad_norm": 0.3013139799493687,
+      "learning_rate": 4.575911522212516e-05,
+      "loss": 0.2854,
+      "step": 359
+    },
+    {
+      "epoch": 0.2304,
+      "grad_norm": 0.27853806891578586,
+      "learning_rate": 4.572955821894305e-05,
+      "loss": 0.2708,
+      "step": 360
+    },
+    {
+      "epoch": 0.23104,
+      "grad_norm": 0.27288786971452156,
+      "learning_rate": 4.5699908189122556e-05,
+      "loss": 0.2365,
+      "step": 361
+    },
+    {
+      "epoch": 0.23168,
+      "grad_norm": 0.33213885368239576,
+      "learning_rate": 4.5670165265722096e-05,
+      "loss": 0.2571,
+      "step": 362
+    },
+    {
+      "epoch": 0.23232,
+      "grad_norm": 0.3316342870811453,
+      "learning_rate": 4.5640329582217004e-05,
+      "loss": 0.2859,
+      "step": 363
+    },
+    {
+      "epoch": 0.23296,
+      "grad_norm": 0.31718513233843937,
+      "learning_rate": 4.5610401272498836e-05,
+      "loss": 0.255,
+      "step": 364
+    },
+    {
+      "epoch": 0.2336,
+      "grad_norm": 0.3003964869956768,
+      "learning_rate": 4.558038047087486e-05,
+      "loss": 0.2638,
+      "step": 365
+    },
+    {
+      "epoch": 0.23424,
+      "grad_norm": 0.3113573197668127,
+      "learning_rate": 4.555026731206741e-05,
+      "loss": 0.2492,
+      "step": 366
+    },
+    {
+      "epoch": 0.23488,
+      "grad_norm": 0.2870220952890737,
+      "learning_rate": 4.5520061931213255e-05,
+      "loss": 0.2789,
+      "step": 367
+    },
+    {
+      "epoch": 0.23552,
+      "grad_norm": 0.27167839521619286,
+      "learning_rate": 4.548976446386305e-05,
+      "loss": 0.2735,
+      "step": 368
+    },
+    {
+      "epoch": 0.23616,
+      "grad_norm": 0.285331504835825,
+      "learning_rate": 4.545937504598069e-05,
+      "loss": 0.2723,
+      "step": 369
+    },
+    {
+      "epoch": 0.2368,
+      "grad_norm": 0.27918688536225067,
+      "learning_rate": 4.5428893813942727e-05,
+      "loss": 0.2685,
+      "step": 370
+    },
+    {
+      "epoch": 0.23744,
+      "grad_norm": 0.2668837009377814,
+      "learning_rate": 4.539832090453771e-05,
+      "loss": 0.2655,
+      "step": 371
+    },
+    {
+      "epoch": 0.23808,
+      "grad_norm": 0.3044254239966202,
+      "learning_rate": 4.536765645496564e-05,
+      "loss": 0.2617,
+      "step": 372
+    },
+    {
+      "epoch": 0.23872,
+      "grad_norm": 0.284750804617016,
+      "learning_rate": 4.533690060283727e-05,
+      "loss": 0.2663,
+      "step": 373
+    },
+    {
+      "epoch": 0.23936,
+      "grad_norm": 0.28718270627176934,
+      "learning_rate": 4.530605348617357e-05,
+      "loss": 0.2752,
+      "step": 374
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.29634840394464135,
+      "learning_rate": 4.527511524340508e-05,
+      "loss": 0.2645,
+      "step": 375
+    },
+    {
+      "epoch": 0.24064,
+      "grad_norm": 0.3012232953009457,
+      "learning_rate": 4.5244086013371246e-05,
+      "loss": 0.2641,
+      "step": 376
+    },
+    {
+      "epoch": 0.24128,
+      "grad_norm": 0.3023357075993031,
+      "learning_rate": 4.521296593531985e-05,
+      "loss": 0.2618,
+      "step": 377
+    },
+    {
+      "epoch": 0.24192,
+      "grad_norm": 0.27441105527916904,
+      "learning_rate": 4.518175514890636e-05,
+      "loss": 0.2663,
+      "step": 378
+    },
+    {
+      "epoch": 0.24256,
+      "grad_norm": 0.3325569029148721,
+      "learning_rate": 4.515045379419332e-05,
+      "loss": 0.2795,
+      "step": 379
+    },
+    {
+      "epoch": 0.2432,
+      "grad_norm": 0.30652258454129583,
+      "learning_rate": 4.5119062011649714e-05,
+      "loss": 0.2756,
+      "step": 380
+    },
+    {
+      "epoch": 0.24384,
+      "grad_norm": 0.33712951361660654,
+      "learning_rate": 4.508757994215031e-05,
+      "loss": 0.2806,
+      "step": 381
+    },
+    {
+      "epoch": 0.24448,
+      "grad_norm": 0.3426531131298772,
+      "learning_rate": 4.505600772697507e-05,
+      "loss": 0.2782,
+      "step": 382
+    },
+    {
+      "epoch": 0.24512,
+      "grad_norm": 0.33303701942677855,
+      "learning_rate": 4.5024345507808495e-05,
+      "loss": 0.2662,
+      "step": 383
+    },
+    {
+      "epoch": 0.24576,
+      "grad_norm": 0.31791526383279123,
+      "learning_rate": 4.499259342673898e-05,
+      "loss": 0.2768,
+      "step": 384
+    },
+    {
+      "epoch": 0.2464,
+      "grad_norm": 0.30503735620868655,
+      "learning_rate": 4.4960751626258205e-05,
+      "loss": 0.2642,
+      "step": 385
+    },
+    {
+      "epoch": 0.24704,
+      "grad_norm": 0.3312474346993656,
+      "learning_rate": 4.492882024926046e-05,
+      "loss": 0.2732,
+      "step": 386
+    },
+    {
+      "epoch": 0.24768,
+      "grad_norm": 0.3133944170389108,
+      "learning_rate": 4.489679943904203e-05,
+      "loss": 0.2933,
+      "step": 387
+    },
+    {
+      "epoch": 0.24832,
+      "grad_norm": 0.2852787428119327,
+      "learning_rate": 4.486468933930055e-05,
+      "loss": 0.2806,
+      "step": 388
+    },
+    {
+      "epoch": 0.24896,
+      "grad_norm": 0.3092836612089024,
+      "learning_rate": 4.483249009413433e-05,
+      "loss": 0.2588,
+      "step": 389
+    },
+    {
+      "epoch": 0.2496,
+      "grad_norm": 0.3112336325754041,
+      "learning_rate": 4.4800201848041765e-05,
+      "loss": 0.2889,
+      "step": 390
+    },
+    {
+      "epoch": 0.25024,
+      "grad_norm": 0.3125491047833594,
+      "learning_rate": 4.476782474592062e-05,
+      "loss": 0.2628,
+      "step": 391
+    },
+    {
+      "epoch": 0.25088,
+      "grad_norm": 0.2912676631270001,
+      "learning_rate": 4.473535893306744e-05,
+      "loss": 0.2696,
+      "step": 392
+    },
+    {
+      "epoch": 0.25152,
+      "grad_norm": 0.28628694727112053,
+      "learning_rate": 4.470280455517687e-05,
+      "loss": 0.2619,
+      "step": 393
+    },
+    {
+      "epoch": 0.25216,
+      "grad_norm": 0.3293686387916455,
+      "learning_rate": 4.4670161758340966e-05,
+      "loss": 0.2739,
+      "step": 394
+    },
+    {
+      "epoch": 0.2528,
+      "grad_norm": 0.31348173029121257,
+      "learning_rate": 4.4637430689048634e-05,
+      "loss": 0.2599,
+      "step": 395
+    },
+    {
+      "epoch": 0.25344,
+      "grad_norm": 0.32951196314360054,
+      "learning_rate": 4.460461149418486e-05,
+      "loss": 0.2567,
+      "step": 396
+    },
+    {
+      "epoch": 0.25408,
+      "grad_norm": 0.34019629325427525,
+      "learning_rate": 4.4571704321030125e-05,
+      "loss": 0.2913,
+      "step": 397
+    },
+    {
+      "epoch": 0.25472,
+      "grad_norm": 0.3301120643416949,
+      "learning_rate": 4.4538709317259754e-05,
+      "loss": 0.2472,
+      "step": 398
+    },
+    {
+      "epoch": 0.25536,
+      "grad_norm": 0.29742911532500216,
+      "learning_rate": 4.4505626630943174e-05,
+      "loss": 0.2617,
+      "step": 399
+    },
+    {
+      "epoch": 0.256,
+      "grad_norm": 0.3417634987932657,
+      "learning_rate": 4.447245641054334e-05,
+      "loss": 0.2742,
+      "step": 400
+    },
+    {
+      "epoch": 0.25664,
+      "grad_norm": 0.2826147129731043,
+      "learning_rate": 4.443919880491598e-05,
+      "loss": 0.2535,
+      "step": 401
+    },
+    {
+      "epoch": 0.25728,
+      "grad_norm": 0.29947144671717485,
+      "learning_rate": 4.440585396330904e-05,
+      "loss": 0.2688,
+      "step": 402
+    },
+    {
+      "epoch": 0.25792,
+      "grad_norm": 0.31610077864520053,
+      "learning_rate": 4.437242203536189e-05,
+      "loss": 0.2724,
+      "step": 403
+    },
+    {
+      "epoch": 0.25856,
+      "grad_norm": 0.3137992831884045,
+      "learning_rate": 4.433890317110475e-05,
+      "loss": 0.2848,
+      "step": 404
+    },
+    {
+      "epoch": 0.2592,
+      "grad_norm": 0.2796187619572129,
+      "learning_rate": 4.430529752095795e-05,
+      "loss": 0.2714,
+      "step": 405
+    },
+    {
+      "epoch": 0.25984,
+      "grad_norm": 0.2912248245493093,
+      "learning_rate": 4.427160523573131e-05,
+      "loss": 0.2795,
+      "step": 406
+    },
+    {
+      "epoch": 0.26048,
+      "grad_norm": 0.2980162603474383,
+      "learning_rate": 4.4237826466623406e-05,
+      "loss": 0.2867,
+      "step": 407
+    },
+    {
+      "epoch": 0.26112,
+      "grad_norm": 0.29652170821748175,
+      "learning_rate": 4.420396136522095e-05,
+      "loss": 0.2727,
+      "step": 408
+    },
+    {
+      "epoch": 0.26176,
+      "grad_norm": 0.2915563003617483,
+      "learning_rate": 4.417001008349807e-05,
+      "loss": 0.2583,
+      "step": 409
+    },
+    {
+      "epoch": 0.2624,
+      "grad_norm": 0.3225628807617719,
+      "learning_rate": 4.413597277381564e-05,
+      "loss": 0.271,
+      "step": 410
+    },
+    {
+      "epoch": 0.26304,
+      "grad_norm": 0.28870166282131143,
+      "learning_rate": 4.410184958892059e-05,
+      "loss": 0.2663,
+      "step": 411
+    },
+    {
+      "epoch": 0.26368,
+      "grad_norm": 0.29553450210347276,
+      "learning_rate": 4.4067640681945244e-05,
+      "loss": 0.2639,
+      "step": 412
+    },
+    {
+      "epoch": 0.26432,
+      "grad_norm": 0.3332862277579268,
+      "learning_rate": 4.403334620640659e-05,
+      "loss": 0.2609,
+      "step": 413
+    },
+    {
+      "epoch": 0.26496,
+      "grad_norm": 0.30107711275005455,
+      "learning_rate": 4.3998966316205636e-05,
+      "loss": 0.2541,
+      "step": 414
+    },
+    {
+      "epoch": 0.2656,
+      "grad_norm": 0.2591023620947492,
+      "learning_rate": 4.396450116562669e-05,
+      "loss": 0.2451,
+      "step": 415
+    },
+    {
+      "epoch": 0.26624,
+      "grad_norm": 0.27765150058129984,
+      "learning_rate": 4.392995090933669e-05,
+      "loss": 0.252,
+      "step": 416
+    },
+    {
+      "epoch": 0.26688,
+      "grad_norm": 0.27831462093368287,
+      "learning_rate": 4.389531570238448e-05,
+      "loss": 0.241,
+      "step": 417
+    },
+    {
+      "epoch": 0.26752,
+      "grad_norm": 0.3393673672608069,
+      "learning_rate": 4.386059570020015e-05,
+      "loss": 0.2781,
+      "step": 418
+    },
+    {
+      "epoch": 0.26816,
+      "grad_norm": 0.2709067486737786,
+      "learning_rate": 4.382579105859428e-05,
+      "loss": 0.2746,
+      "step": 419
+    },
+    {
+      "epoch": 0.2688,
+      "grad_norm": 0.31024807728799664,
+      "learning_rate": 4.379090193375735e-05,
+      "loss": 0.269,
+      "step": 420
+    },
+    {
+      "epoch": 0.26944,
+      "grad_norm": 0.29437773626601327,
+      "learning_rate": 4.375592848225889e-05,
+      "loss": 0.2497,
+      "step": 421
+    },
+    {
+      "epoch": 0.27008,
+      "grad_norm": 0.27249340717742593,
+      "learning_rate": 4.372087086104692e-05,
+      "loss": 0.2769,
+      "step": 422
+    },
+    {
+      "epoch": 0.27072,
+      "grad_norm": 0.3182175494652926,
+      "learning_rate": 4.368572922744715e-05,
+      "loss": 0.2709,
+      "step": 423
+    },
+    {
+      "epoch": 0.27136,
+      "grad_norm": 0.2906720843900486,
+      "learning_rate": 4.365050373916233e-05,
+      "loss": 0.275,
+      "step": 424
+    },
+    {
+      "epoch": 0.272,
+      "grad_norm": 0.3098393205417728,
+      "learning_rate": 4.3615194554271485e-05,
+      "loss": 0.234,
+      "step": 425
+    },
+    {
+      "epoch": 0.27264,
+      "grad_norm": 0.30226227188570165,
+      "learning_rate": 4.357980183122927e-05,
+      "loss": 0.2728,
+      "step": 426
+    },
+    {
+      "epoch": 0.27328,
+      "grad_norm": 0.3101806085164065,
+      "learning_rate": 4.354432572886522e-05,
+      "loss": 0.272,
+      "step": 427
+    },
+    {
+      "epoch": 0.27392,
+      "grad_norm": 0.28245936335261307,
+      "learning_rate": 4.3508766406383036e-05,
+      "loss": 0.278,
+      "step": 428
+    },
+    {
+      "epoch": 0.27456,
+      "grad_norm": 0.27797054709506586,
+      "learning_rate": 4.347312402335989e-05,
+      "loss": 0.2684,
+      "step": 429
+    },
+    {
+      "epoch": 0.2752,
+      "grad_norm": 0.2990500240447701,
+      "learning_rate": 4.34373987397457e-05,
+      "loss": 0.278,
+      "step": 430
+    },
+    {
+      "epoch": 0.27584,
+      "grad_norm": 0.2727197517887389,
+      "learning_rate": 4.340159071586241e-05,
+      "loss": 0.2454,
+      "step": 431
+    },
+    {
+      "epoch": 0.27648,
+      "grad_norm": 0.30243924092224234,
+      "learning_rate": 4.336570011240326e-05,
+      "loss": 0.2455,
+      "step": 432
+    },
+    {
+      "epoch": 0.27712,
+      "grad_norm": 0.2803609027017541,
+      "learning_rate": 4.33297270904321e-05,
+      "loss": 0.2681,
+      "step": 433
+    },
+    {
+      "epoch": 0.27776,
+      "grad_norm": 0.2920451450586205,
+      "learning_rate": 4.329367181138262e-05,
+      "loss": 0.2762,
+      "step": 434
+    },
+    {
+      "epoch": 0.2784,
+      "grad_norm": 0.30901997193481695,
+      "learning_rate": 4.3257534437057674e-05,
+      "loss": 0.2575,
+      "step": 435
+    },
+    {
+      "epoch": 0.27904,
+      "grad_norm": 0.28493311811209304,
+      "learning_rate": 4.3221315129628506e-05,
+      "loss": 0.2493,
+      "step": 436
+    },
+    {
+      "epoch": 0.27968,
+      "grad_norm": 0.3039818748430415,
+      "learning_rate": 4.318501405163406e-05,
+      "loss": 0.2624,
+      "step": 437
+    },
+    {
+      "epoch": 0.28032,
+      "grad_norm": 0.26937915062885653,
+      "learning_rate": 4.314863136598022e-05,
+      "loss": 0.247,
+      "step": 438
+    },
+    {
+      "epoch": 0.28096,
+      "grad_norm": 0.33555541975858594,
+      "learning_rate": 4.311216723593913e-05,
+      "loss": 0.2867,
+      "step": 439
+    },
+    {
+      "epoch": 0.2816,
+      "grad_norm": 0.31925378826398826,
+      "learning_rate": 4.307562182514838e-05,
+      "loss": 0.2617,
+      "step": 440
+    },
+    {
+      "epoch": 0.28224,
+      "grad_norm": 0.32623010034841626,
+      "learning_rate": 4.3038995297610364e-05,
+      "loss": 0.249,
+      "step": 441
+    },
+    {
+      "epoch": 0.28288,
+      "grad_norm": 0.30358330160420666,
+      "learning_rate": 4.3002287817691466e-05,
+      "loss": 0.2865,
+      "step": 442
+    },
+    {
+      "epoch": 0.28352,
+      "grad_norm": 0.33915628852280005,
+      "learning_rate": 4.2965499550121354e-05,
+      "loss": 0.2685,
+      "step": 443
+    },
+    {
+      "epoch": 0.28416,
+      "grad_norm": 0.3547465820682451,
+      "learning_rate": 4.2928630659992267e-05,
+      "loss": 0.2711,
+      "step": 444
+    },
+    {
+      "epoch": 0.2848,
+      "grad_norm": 0.3336725360323176,
+      "learning_rate": 4.2891681312758225e-05,
+      "loss": 0.261,
+      "step": 445
+    },
+    {
+      "epoch": 0.28544,
+      "grad_norm": 0.30303539171283983,
+      "learning_rate": 4.285465167423432e-05,
+      "loss": 0.2719,
+      "step": 446
+    },
+    {
+      "epoch": 0.28608,
+      "grad_norm": 0.311033741890672,
+      "learning_rate": 4.281754191059596e-05,
+      "loss": 0.2616,
+      "step": 447
+    },
+    {
+      "epoch": 0.28672,
+      "grad_norm": 0.3125370811674668,
+      "learning_rate": 4.278035218837813e-05,
+      "loss": 0.2889,
+      "step": 448
+    },
+    {
+      "epoch": 0.28736,
+      "grad_norm": 0.2756629291524462,
+      "learning_rate": 4.274308267447463e-05,
+      "loss": 0.2518,
+      "step": 449
+    },
+    {
+      "epoch": 0.288,
+      "grad_norm": 0.2984734533438556,
+      "learning_rate": 4.270573353613734e-05,
+      "loss": 0.2712,
+      "step": 450
+    },
+    {
+      "epoch": 0.28864,
+      "grad_norm": 0.29595878739145964,
+      "learning_rate": 4.266830494097546e-05,
+      "loss": 0.255,
+      "step": 451
+    },
+    {
+      "epoch": 0.28928,
+      "grad_norm": 0.3205021385414694,
+      "learning_rate": 4.263079705695478e-05,
+      "loss": 0.2636,
+      "step": 452
+    },
+    {
+      "epoch": 0.28992,
+      "grad_norm": 0.3600626165622021,
+      "learning_rate": 4.2593210052396894e-05,
+      "loss": 0.2541,
+      "step": 453
+    },
+    {
+      "epoch": 0.29056,
+      "grad_norm": 0.25969620151233935,
+      "learning_rate": 4.255554409597846e-05,
+      "loss": 0.2401,
+      "step": 454
+    },
+    {
+      "epoch": 0.2912,
+      "grad_norm": 0.33149623415612184,
+      "learning_rate": 4.2517799356730445e-05,
+      "loss": 0.2687,
+      "step": 455
+    },
+    {
+      "epoch": 0.29184,
+      "grad_norm": 0.30226474356531935,
+      "learning_rate": 4.247997600403737e-05,
+      "loss": 0.2594,
+      "step": 456
+    },
+    {
+      "epoch": 0.29248,
+      "grad_norm": 0.30007952734237353,
+      "learning_rate": 4.244207420763655e-05,
+      "loss": 0.2498,
+      "step": 457
+    },
+    {
+      "epoch": 0.29312,
+      "grad_norm": 0.28047639127931384,
+      "learning_rate": 4.240409413761731e-05,
+      "loss": 0.2556,
+      "step": 458
+    },
+    {
+      "epoch": 0.29376,
+      "grad_norm": 0.29832917410059595,
+      "learning_rate": 4.236603596442023e-05,
+      "loss": 0.2632,
+      "step": 459
+    },
+    {
+      "epoch": 0.2944,
+      "grad_norm": 0.29802598060016766,
+      "learning_rate": 4.232789985883642e-05,
+      "loss": 0.2693,
+      "step": 460
+    },
+    {
+      "epoch": 0.29504,
+      "grad_norm": 0.2911033726222457,
+      "learning_rate": 4.2289685992006704e-05,
+      "loss": 0.2772,
+      "step": 461
+    },
+    {
+      "epoch": 0.29568,
+      "grad_norm": 0.2903001309362164,
+      "learning_rate": 4.225139453542087e-05,
+      "loss": 0.2607,
+      "step": 462
+    },
+    {
+      "epoch": 0.29632,
+      "grad_norm": 0.3078582069653545,
+      "learning_rate": 4.2213025660916904e-05,
+      "loss": 0.2667,
+      "step": 463
+    },
+    {
+      "epoch": 0.29696,
+      "grad_norm": 0.3030493045200669,
+      "learning_rate": 4.21745795406802e-05,
+      "loss": 0.2467,
+      "step": 464
+    },
+    {
+      "epoch": 0.2976,
+      "grad_norm": 0.26660015999204884,
+      "learning_rate": 4.2136056347242835e-05,
+      "loss": 0.25,
+      "step": 465
+    },
+    {
+      "epoch": 0.29824,
+      "grad_norm": 0.30144794273216646,
+      "learning_rate": 4.209745625348273e-05,
+      "loss": 0.284,
+      "step": 466
+    },
+    {
+      "epoch": 0.29888,
+      "grad_norm": 0.3284286693049362,
+      "learning_rate": 4.205877943262291e-05,
+      "loss": 0.2464,
+      "step": 467
+    },
+    {
+      "epoch": 0.29952,
+      "grad_norm": 0.25954840338063495,
+      "learning_rate": 4.2020026058230756e-05,
+      "loss": 0.2684,
+      "step": 468
+    },
+    {
+      "epoch": 0.30016,
+      "grad_norm": 0.2945590110420638,
+      "learning_rate": 4.1981196304217154e-05,
+      "loss": 0.264,
+      "step": 469
+    },
+    {
+      "epoch": 0.3008,
+      "grad_norm": 0.2927523824074909,
+      "learning_rate": 4.1942290344835765e-05,
+      "loss": 0.2645,
+      "step": 470
+    },
+    {
+      "epoch": 0.30144,
+      "grad_norm": 0.30771010182551334,
+      "learning_rate": 4.190330835468224e-05,
+      "loss": 0.2671,
+      "step": 471
+    },
+    {
+      "epoch": 0.30208,
+      "grad_norm": 0.29384348361000195,
+      "learning_rate": 4.1864250508693424e-05,
+      "loss": 0.2794,
+      "step": 472
+    },
+    {
+      "epoch": 0.30272,
+      "grad_norm": 0.2896660319426316,
+      "learning_rate": 4.182511698214657e-05,
+      "loss": 0.2706,
+      "step": 473
+    },
+    {
+      "epoch": 0.30336,
+      "grad_norm": 0.28080256128136577,
+      "learning_rate": 4.1785907950658573e-05,
+      "loss": 0.259,
+      "step": 474
+    },
+    {
+      "epoch": 0.304,
+      "grad_norm": 0.30478773930609737,
+      "learning_rate": 4.1746623590185155e-05,
+      "loss": 0.2528,
+      "step": 475
+    },
+    {
+      "epoch": 0.30464,
+      "grad_norm": 0.2961976249198792,
+      "learning_rate": 4.170726407702008e-05,
+      "loss": 0.2466,
+      "step": 476
+    },
+    {
+      "epoch": 0.30528,
+      "grad_norm": 0.31686233103136463,
+      "learning_rate": 4.1667829587794374e-05,
+      "loss": 0.283,
+      "step": 477
+    },
+    {
+      "epoch": 0.30592,
+      "grad_norm": 0.3159707315375394,
+      "learning_rate": 4.162832029947553e-05,
+      "loss": 0.2735,
+      "step": 478
+    },
+    {
+      "epoch": 0.30656,
+      "grad_norm": 0.27811512588290743,
+      "learning_rate": 4.158873638936672e-05,
+      "loss": 0.248,
+      "step": 479
+    },
+    {
+      "epoch": 0.3072,
+      "grad_norm": 0.45896640549255663,
+      "learning_rate": 4.154907803510599e-05,
+      "loss": 0.2525,
+      "step": 480
+    },
+    {
+      "epoch": 0.30784,
+      "grad_norm": 0.28996697916337744,
+      "learning_rate": 4.150934541466545e-05,
+      "loss": 0.2627,
+      "step": 481
+    },
+    {
+      "epoch": 0.30848,
+      "grad_norm": 0.3052139609989655,
+      "learning_rate": 4.146953870635049e-05,
+      "loss": 0.273,
+      "step": 482
+    },
+    {
+      "epoch": 0.30912,
+      "grad_norm": 0.32057984599943073,
+      "learning_rate": 4.1429658088798996e-05,
+      "loss": 0.2636,
+      "step": 483
+    },
+    {
+      "epoch": 0.30976,
+      "grad_norm": 0.29200883447827164,
+      "learning_rate": 4.138970374098052e-05,
+      "loss": 0.2498,
+      "step": 484
+    },
+    {
+      "epoch": 0.3104,
+      "grad_norm": 0.3211437061784034,
+      "learning_rate": 4.1349675842195494e-05,
+      "loss": 0.2688,
+      "step": 485
+    },
+    {
+      "epoch": 0.31104,
+      "grad_norm": 0.2692839546004917,
+      "learning_rate": 4.130957457207441e-05,
+      "loss": 0.2513,
+      "step": 486
+    },
+    {
+      "epoch": 0.31168,
+      "grad_norm": 0.29738376417175055,
+      "learning_rate": 4.126940011057703e-05,
+      "loss": 0.242,
+      "step": 487
+    },
+    {
+      "epoch": 0.31232,
+      "grad_norm": 0.2925125860829048,
+      "learning_rate": 4.1229152637991576e-05,
+      "loss": 0.269,
+      "step": 488
+    },
+    {
+      "epoch": 0.31296,
+      "grad_norm": 0.31841073786254037,
+      "learning_rate": 4.1188832334933906e-05,
+      "loss": 0.2491,
+      "step": 489
+    },
+    {
+      "epoch": 0.3136,
+      "grad_norm": 0.256917596278123,
+      "learning_rate": 4.114843938234673e-05,
+      "loss": 0.2759,
+      "step": 490
+    },
+    {
+      "epoch": 0.31424,
+      "grad_norm": 0.3063311474261881,
+      "learning_rate": 4.1107973961498763e-05,
+      "loss": 0.2598,
+      "step": 491
+    },
+    {
+      "epoch": 0.31488,
+      "grad_norm": 0.32855934285971683,
+      "learning_rate": 4.106743625398395e-05,
+      "loss": 0.2681,
+      "step": 492
+    },
+    {
+      "epoch": 0.31552,
+      "grad_norm": 0.36353433750679803,
+      "learning_rate": 4.102682644172061e-05,
+      "loss": 0.2873,
+      "step": 493
+    },
+    {
+      "epoch": 0.31616,
+      "grad_norm": 0.33622331775831543,
+      "learning_rate": 4.0986144706950677e-05,
+      "loss": 0.276,
+      "step": 494
+    },
+    {
+      "epoch": 0.3168,
+      "grad_norm": 0.31722692435482774,
+      "learning_rate": 4.094539123223881e-05,
+      "loss": 0.2444,
+      "step": 495
+    },
+    {
+      "epoch": 0.31744,
+      "grad_norm": 0.2780719831837547,
+      "learning_rate": 4.090456620047162e-05,
+      "loss": 0.2451,
+      "step": 496
+    },
+    {
+      "epoch": 0.31808,
+      "grad_norm": 0.35077422669493935,
+      "learning_rate": 4.086366979485685e-05,
+      "loss": 0.2816,
+      "step": 497
+    },
+    {
+      "epoch": 0.31872,
+      "grad_norm": 0.30473125946275736,
+      "learning_rate": 4.082270219892254e-05,
+      "loss": 0.2644,
+      "step": 498
+    },
+    {
+      "epoch": 0.31936,
+      "grad_norm": 0.27919846819675614,
+      "learning_rate": 4.0781663596516206e-05,
+      "loss": 0.244,
+      "step": 499
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.30869545887631733,
+      "learning_rate": 4.074055417180401e-05,
+      "loss": 0.2709,
+      "step": 500
+    },
+    {
+      "epoch": 0.32064,
+      "grad_norm": 0.32401310700356356,
+      "learning_rate": 4.069937410926994e-05,
+      "loss": 0.2555,
+      "step": 501
+    },
+    {
+      "epoch": 0.32128,
+      "grad_norm": 0.28989905607846644,
+      "learning_rate": 4.0658123593714984e-05,
+      "loss": 0.2648,
+      "step": 502
+    },
+    {
+      "epoch": 0.32192,
+      "grad_norm": 0.2970056265621276,
+      "learning_rate": 4.061680281025628e-05,
+      "loss": 0.2491,
+      "step": 503
+    },
+    {
+      "epoch": 0.32256,
+      "grad_norm": 0.30095176466887164,
+      "learning_rate": 4.057541194432634e-05,
+      "loss": 0.2428,
+      "step": 504
+    },
+    {
+      "epoch": 0.3232,
+      "grad_norm": 0.307577464320209,
+      "learning_rate": 4.053395118167214e-05,
+      "loss": 0.2591,
+      "step": 505
+    },
+    {
+      "epoch": 0.32384,
+      "grad_norm": 0.28875399318294626,
+      "learning_rate": 4.049242070835435e-05,
+      "loss": 0.2876,
+      "step": 506
+    },
+    {
+      "epoch": 0.32448,
+      "grad_norm": 0.3372288637186776,
+      "learning_rate": 4.0450820710746475e-05,
+      "loss": 0.2614,
+      "step": 507
+    },
+    {
+      "epoch": 0.32512,
+      "grad_norm": 0.2893542015604074,
+      "learning_rate": 4.0409151375534014e-05,
+      "loss": 0.2646,
+      "step": 508
+    },
+    {
+      "epoch": 0.32576,
+      "grad_norm": 0.30517166216359926,
+      "learning_rate": 4.036741288971362e-05,
+      "loss": 0.2563,
+      "step": 509
+    },
+    {
+      "epoch": 0.3264,
+      "grad_norm": 0.3132371089607073,
+      "learning_rate": 4.032560544059228e-05,
+      "loss": 0.2549,
+      "step": 510
+    },
+    {
+      "epoch": 0.32704,
+      "grad_norm": 0.2584564148069602,
+      "learning_rate": 4.028372921578647e-05,
+      "loss": 0.2642,
+      "step": 511
+    },
+    {
+      "epoch": 0.32768,
+      "grad_norm": 0.29056742266359853,
+      "learning_rate": 4.024178440322128e-05,
+      "loss": 0.2517,
+      "step": 512
+    },
+    {
+      "epoch": 0.32832,
+      "grad_norm": 0.2806424707934406,
+      "learning_rate": 4.019977119112962e-05,
+      "loss": 0.2612,
+      "step": 513
+    },
+    {
+      "epoch": 0.32896,
+      "grad_norm": 0.3213624904512003,
+      "learning_rate": 4.015768976805135e-05,
+      "loss": 0.3041,
+      "step": 514
+    },
+    {
+      "epoch": 0.3296,
+      "grad_norm": 0.3196533973528481,
+      "learning_rate": 4.011554032283242e-05,
+      "loss": 0.2351,
+      "step": 515
+    },
+    {
+      "epoch": 0.33024,
+      "grad_norm": 0.3329565293585606,
+      "learning_rate": 4.007332304462406e-05,
+      "loss": 0.251,
+      "step": 516
+    },
+    {
+      "epoch": 0.33088,
+      "grad_norm": 0.3109673121296247,
+      "learning_rate": 4.003103812288189e-05,
+      "loss": 0.2686,
+      "step": 517
+    },
+    {
+      "epoch": 0.33152,
+      "grad_norm": 0.2675208149986414,
+      "learning_rate": 3.99886857473651e-05,
+      "loss": 0.2525,
+      "step": 518
+    },
+    {
+      "epoch": 0.33216,
+      "grad_norm": 0.34229548303640794,
+      "learning_rate": 3.9946266108135575e-05,
+      "loss": 0.2772,
+      "step": 519
+    },
+    {
+      "epoch": 0.3328,
+      "grad_norm": 0.32006866438809456,
+      "learning_rate": 3.9903779395557087e-05,
+      "loss": 0.2622,
+      "step": 520
+    },
+    {
+      "epoch": 0.33344,
+      "grad_norm": 0.26715648446115114,
+      "learning_rate": 3.9861225800294366e-05,
+      "loss": 0.2308,
+      "step": 521
+    },
+    {
+      "epoch": 0.33408,
+      "grad_norm": 0.3480585053707105,
+      "learning_rate": 3.981860551331232e-05,
+      "loss": 0.2669,
+      "step": 522
+    },
+    {
+      "epoch": 0.33472,
+      "grad_norm": 0.2922018702119113,
+      "learning_rate": 3.977591872587512e-05,
+      "loss": 0.2565,
+      "step": 523
+    },
+    {
+      "epoch": 0.33536,
+      "grad_norm": 0.2611827785563305,
+      "learning_rate": 3.973316562954539e-05,
+      "loss": 0.2516,
+      "step": 524
+    },
+    {
+      "epoch": 0.336,
+      "grad_norm": 0.2984964297936148,
+      "learning_rate": 3.969034641618332e-05,
+      "loss": 0.2852,
+      "step": 525
+    },
+    {
+      "epoch": 0.33664,
+      "grad_norm": 0.2892462655028235,
+      "learning_rate": 3.964746127794579e-05,
+      "loss": 0.2635,
+      "step": 526
+    },
+    {
+      "epoch": 0.33728,
+      "grad_norm": 0.28835600723222543,
+      "learning_rate": 3.9604510407285524e-05,
+      "loss": 0.2511,
+      "step": 527
+    },
+    {
+      "epoch": 0.33792,
+      "grad_norm": 0.2999026080900571,
+      "learning_rate": 3.956149399695027e-05,
+      "loss": 0.2529,
+      "step": 528
+    },
+    {
+      "epoch": 0.33856,
+      "grad_norm": 0.3174683122687009,
+      "learning_rate": 3.951841223998186e-05,
+      "loss": 0.2803,
+      "step": 529
+    },
+    {
+      "epoch": 0.3392,
+      "grad_norm": 0.26907149760511656,
+      "learning_rate": 3.947526532971538e-05,
+      "loss": 0.2452,
+      "step": 530
+    },
+    {
+      "epoch": 0.33984,
+      "grad_norm": 0.2771670066399074,
+      "learning_rate": 3.943205345977831e-05,
+      "loss": 0.2598,
+      "step": 531
+    },
+    {
+      "epoch": 0.34048,
+      "grad_norm": 0.28877064076702313,
+      "learning_rate": 3.9388776824089636e-05,
+      "loss": 0.2559,
+      "step": 532
+    },
+    {
+      "epoch": 0.34112,
+      "grad_norm": 0.2649860095131696,
+      "learning_rate": 3.9345435616858995e-05,
+      "loss": 0.2673,
+      "step": 533
+    },
+    {
+      "epoch": 0.34176,
+      "grad_norm": 0.269531839762295,
+      "learning_rate": 3.930203003258579e-05,
+      "loss": 0.2569,
+      "step": 534
+    },
+    {
+      "epoch": 0.3424,
+      "grad_norm": 0.27891756261987266,
+      "learning_rate": 3.9258560266058335e-05,
+      "loss": 0.2702,
+      "step": 535
+    },
+    {
+      "epoch": 0.34304,
+      "grad_norm": 0.27053829922547884,
+      "learning_rate": 3.921502651235297e-05,
+      "loss": 0.2535,
+      "step": 536
+    },
+    {
+      "epoch": 0.34368,
+      "grad_norm": 0.28011073116796703,
+      "learning_rate": 3.917142896683318e-05,
+      "loss": 0.272,
+      "step": 537
+    },
+    {
+      "epoch": 0.34432,
+      "grad_norm": 0.270867487872639,
+      "learning_rate": 3.912776782514872e-05,
+      "loss": 0.2543,
+      "step": 538
+    },
+    {
+      "epoch": 0.34496,
+      "grad_norm": 0.2746991664504782,
+      "learning_rate": 3.908404328323475e-05,
+      "loss": 0.2728,
+      "step": 539
+    },
+    {
+      "epoch": 0.3456,
+      "grad_norm": 0.27939140218767883,
+      "learning_rate": 3.904025553731094e-05,
+      "loss": 0.2418,
+      "step": 540
+    },
+    {
+      "epoch": 0.34624,
+      "grad_norm": 0.2731998143935589,
+      "learning_rate": 3.8996404783880606e-05,
+      "loss": 0.2433,
+      "step": 541
+    },
+    {
+      "epoch": 0.34688,
+      "grad_norm": 0.2750465650133765,
+      "learning_rate": 3.895249121972981e-05,
+      "loss": 0.2422,
+      "step": 542
+    },
+    {
+      "epoch": 0.34752,
+      "grad_norm": 0.27336449849709915,
+      "learning_rate": 3.890851504192649e-05,
+      "loss": 0.2689,
+      "step": 543
+    },
+    {
+      "epoch": 0.34816,
+      "grad_norm": 0.29586657387061377,
+      "learning_rate": 3.8864476447819565e-05,
+      "loss": 0.2465,
+      "step": 544
+    },
+    {
+      "epoch": 0.3488,
+      "grad_norm": 0.270200591770297,
+      "learning_rate": 3.8820375635038067e-05,
+      "loss": 0.2573,
+      "step": 545
+    },
+    {
+      "epoch": 0.34944,
+      "grad_norm": 0.3190425370253469,
+      "learning_rate": 3.877621280149022e-05,
+      "loss": 0.2678,
+      "step": 546
+    },
+    {
+      "epoch": 0.35008,
+      "grad_norm": 0.26043543170976835,
+      "learning_rate": 3.87319881453626e-05,
+      "loss": 0.2354,
+      "step": 547
+    },
+    {
+      "epoch": 0.35072,
+      "grad_norm": 0.2704011594558705,
+      "learning_rate": 3.868770186511921e-05,
+      "loss": 0.2626,
+      "step": 548
+    },
+    {
+      "epoch": 0.35136,
+      "grad_norm": 0.29304403560572506,
+      "learning_rate": 3.864335415950059e-05,
+      "loss": 0.2437,
+      "step": 549
+    },
+    {
+      "epoch": 0.352,
+      "grad_norm": 0.2655479075710523,
+      "learning_rate": 3.859894522752297e-05,
+      "loss": 0.2503,
+      "step": 550
+    },
+    {
+      "epoch": 0.35264,
+      "grad_norm": 0.312630229855102,
+      "learning_rate": 3.8554475268477274e-05,
+      "loss": 0.257,
+      "step": 551
+    },
+    {
+      "epoch": 0.35328,
+      "grad_norm": 0.274670625897055,
+      "learning_rate": 3.850994448192836e-05,
+      "loss": 0.2596,
+      "step": 552
+    },
+    {
+      "epoch": 0.35392,
+      "grad_norm": 0.2474262363659263,
+      "learning_rate": 3.846535306771404e-05,
+      "loss": 0.2497,
+      "step": 553
+    },
+    {
+      "epoch": 0.35456,
+      "grad_norm": 0.30813525842938966,
+      "learning_rate": 3.842070122594417e-05,
+      "loss": 0.2645,
+      "step": 554
+    },
+    {
+      "epoch": 0.3552,
+      "grad_norm": 0.27841535526184324,
+      "learning_rate": 3.837598915699981e-05,
+      "loss": 0.2464,
+      "step": 555
+    },
+    {
+      "epoch": 0.35584,
+      "grad_norm": 0.2938358535555391,
+      "learning_rate": 3.83312170615323e-05,
+      "loss": 0.2387,
+      "step": 556
+    },
+    {
+      "epoch": 0.35648,
+      "grad_norm": 0.29756211437074354,
+      "learning_rate": 3.8286385140462334e-05,
+      "loss": 0.2465,
+      "step": 557
+    },
+    {
+      "epoch": 0.35712,
+      "grad_norm": 0.30217380986318165,
+      "learning_rate": 3.8241493594979114e-05,
+      "loss": 0.264,
+      "step": 558
+    },
+    {
+      "epoch": 0.35776,
+      "grad_norm": 0.28961179785597074,
+      "learning_rate": 3.8196542626539396e-05,
+      "loss": 0.2641,
+      "step": 559
+    },
+    {
+      "epoch": 0.3584,
+      "grad_norm": 0.30695303659565276,
+      "learning_rate": 3.815153243686658e-05,
+      "loss": 0.2643,
+      "step": 560
+    },
+    {
+      "epoch": 0.35904,
+      "grad_norm": 0.29395099772794475,
+      "learning_rate": 3.8106463227949885e-05,
+      "loss": 0.2643,
+      "step": 561
+    },
+    {
+      "epoch": 0.35968,
+      "grad_norm": 0.29219915519201384,
+      "learning_rate": 3.806133520204335e-05,
+      "loss": 0.2358,
+      "step": 562
+    },
+    {
+      "epoch": 0.36032,
+      "grad_norm": 0.2771159950154576,
+      "learning_rate": 3.801614856166495e-05,
+      "loss": 0.2444,
+      "step": 563
+    },
+    {
+      "epoch": 0.36096,
+      "grad_norm": 0.3140877640793322,
+      "learning_rate": 3.797090350959574e-05,
+      "loss": 0.2479,
+      "step": 564
+    },
+    {
+      "epoch": 0.3616,
+      "grad_norm": 0.28818249284565073,
+      "learning_rate": 3.792560024887887e-05,
+      "loss": 0.2538,
+      "step": 565
+    },
+    {
+      "epoch": 0.36224,
+      "grad_norm": 0.2783892260751085,
+      "learning_rate": 3.788023898281873e-05,
+      "loss": 0.2523,
+      "step": 566
+    },
+    {
+      "epoch": 0.36288,
+      "grad_norm": 0.3092513006484571,
+      "learning_rate": 3.783481991498001e-05,
+      "loss": 0.2575,
+      "step": 567
+    },
+    {
+      "epoch": 0.36352,
+      "grad_norm": 0.28561921777773985,
+      "learning_rate": 3.7789343249186805e-05,
+      "loss": 0.2657,
+      "step": 568
+    },
+    {
+      "epoch": 0.36416,
+      "grad_norm": 0.29745794354352423,
+      "learning_rate": 3.774380918952165e-05,
+      "loss": 0.2746,
+      "step": 569
+    },
+    {
+      "epoch": 0.3648,
+      "grad_norm": 0.28619753080969124,
+      "learning_rate": 3.76982179403247e-05,
+      "loss": 0.238,
+      "step": 570
+    },
+    {
+      "epoch": 0.36544,
+      "grad_norm": 0.2865777627431609,
+      "learning_rate": 3.7652569706192694e-05,
+      "loss": 0.2672,
+      "step": 571
+    },
+    {
+      "epoch": 0.36608,
+      "grad_norm": 0.2662051134951306,
+      "learning_rate": 3.760686469197815e-05,
+      "loss": 0.2493,
+      "step": 572
+    },
+    {
+      "epoch": 0.36672,
+      "grad_norm": 0.3122204747602635,
+      "learning_rate": 3.756110310278835e-05,
+      "loss": 0.249,
+      "step": 573
+    },
+    {
+      "epoch": 0.36736,
+      "grad_norm": 0.2551416288395675,
+      "learning_rate": 3.7515285143984524e-05,
+      "loss": 0.2426,
+      "step": 574
+    },
+    {
+      "epoch": 0.368,
+      "grad_norm": 0.2982227316906956,
+      "learning_rate": 3.746941102118082e-05,
+      "loss": 0.2509,
+      "step": 575
+    },
+    {
+      "epoch": 0.36864,
+      "grad_norm": 0.24907998630323142,
+      "learning_rate": 3.742348094024342e-05,
+      "loss": 0.2731,
+      "step": 576
+    },
+    {
+      "epoch": 0.36928,
+      "grad_norm": 0.308035846935991,
+      "learning_rate": 3.7377495107289664e-05,
+      "loss": 0.2294,
+      "step": 577
+    },
+    {
+      "epoch": 0.36992,
+      "grad_norm": 0.2981772132955674,
+      "learning_rate": 3.7331453728687054e-05,
+      "loss": 0.2521,
+      "step": 578
+    },
+    {
+      "epoch": 0.37056,
+      "grad_norm": 0.31976646891891425,
+      "learning_rate": 3.7285357011052394e-05,
+      "loss": 0.2398,
+      "step": 579
+    },
+    {
+      "epoch": 0.3712,
+      "grad_norm": 0.30332985531986056,
+      "learning_rate": 3.72392051612508e-05,
+      "loss": 0.2692,
+      "step": 580
+    },
+    {
+      "epoch": 0.37184,
+      "grad_norm": 0.27711012195719126,
+      "learning_rate": 3.719299838639481e-05,
+      "loss": 0.2317,
+      "step": 581
+    },
+    {
+      "epoch": 0.37248,
+      "grad_norm": 0.4275716591537019,
+      "learning_rate": 3.714673689384346e-05,
+      "loss": 0.2737,
+      "step": 582
+    },
+    {
+      "epoch": 0.37312,
+      "grad_norm": 0.3026702601663526,
+      "learning_rate": 3.710042089120131e-05,
+      "loss": 0.2483,
+      "step": 583
+    },
+    {
+      "epoch": 0.37376,
+      "grad_norm": 0.29631681543119553,
+      "learning_rate": 3.705405058631756e-05,
+      "loss": 0.2632,
+      "step": 584
+    },
+    {
+      "epoch": 0.3744,
+      "grad_norm": 0.34902565005008385,
+      "learning_rate": 3.700762618728508e-05,
+      "loss": 0.2575,
+      "step": 585
+    },
+    {
+      "epoch": 0.37504,
+      "grad_norm": 0.30672309870176107,
+      "learning_rate": 3.696114790243955e-05,
+      "loss": 0.2645,
+      "step": 586
+    },
+    {
+      "epoch": 0.37568,
+      "grad_norm": 0.29876169974205874,
+      "learning_rate": 3.691461594035841e-05,
+      "loss": 0.2797,
+      "step": 587
+    },
+    {
+      "epoch": 0.37632,
+      "grad_norm": 0.2705265427737359,
+      "learning_rate": 3.686803050986e-05,
+      "loss": 0.2488,
+      "step": 588
+    },
+    {
+      "epoch": 0.37696,
+      "grad_norm": 0.36485954853209873,
+      "learning_rate": 3.6821391820002596e-05,
+      "loss": 0.2445,
+      "step": 589
+    },
+    {
+      "epoch": 0.3776,
+      "grad_norm": 0.2511551627718809,
+      "learning_rate": 3.677470008008352e-05,
+      "loss": 0.2461,
+      "step": 590
+    },
+    {
+      "epoch": 0.37824,
+      "grad_norm": 0.3221816057866106,
+      "learning_rate": 3.672795549963814e-05,
+      "loss": 0.2459,
+      "step": 591
+    },
+    {
+      "epoch": 0.37888,
+      "grad_norm": 0.29469731622849443,
+      "learning_rate": 3.668115828843894e-05,
+      "loss": 0.2507,
+      "step": 592
+    },
+    {
+      "epoch": 0.37952,
+      "grad_norm": 0.2752002164016696,
+      "learning_rate": 3.6634308656494605e-05,
+      "loss": 0.2515,
+      "step": 593
+    },
+    {
+      "epoch": 0.38016,
+      "grad_norm": 0.2937245665619183,
+      "learning_rate": 3.6587406814049075e-05,
+      "loss": 0.2414,
+      "step": 594
+    },
+    {
+      "epoch": 0.3808,
+      "grad_norm": 0.2623481228772293,
+      "learning_rate": 3.6540452971580566e-05,
+      "loss": 0.2592,
+      "step": 595
+    },
+    {
+      "epoch": 0.38144,
+      "grad_norm": 0.3278516786328869,
+      "learning_rate": 3.649344733980068e-05,
+      "loss": 0.2558,
+      "step": 596
+    },
+    {
+      "epoch": 0.38208,
+      "grad_norm": 0.30694577299986786,
+      "learning_rate": 3.6446390129653405e-05,
+      "loss": 0.2508,
+      "step": 597
+    },
+    {
+      "epoch": 0.38272,
+      "grad_norm": 0.2754762052588,
+      "learning_rate": 3.63992815523142e-05,
+      "loss": 0.2411,
+      "step": 598
+    },
+    {
+      "epoch": 0.38336,
+      "grad_norm": 0.2927492207692471,
+      "learning_rate": 3.6352121819189075e-05,
+      "loss": 0.2426,
+      "step": 599
+    },
+    {
+      "epoch": 0.384,
+      "grad_norm": 0.33037655660389065,
+      "learning_rate": 3.630491114191354e-05,
+      "loss": 0.2415,
+      "step": 600
+    },
+    {
+      "epoch": 0.38464,
+      "grad_norm": 0.29805367907913555,
+      "learning_rate": 3.62576497323518e-05,
+      "loss": 0.2803,
+      "step": 601
+    },
+    {
+      "epoch": 0.38528,
+      "grad_norm": 0.27588386664863995,
+      "learning_rate": 3.621033780259566e-05,
+      "loss": 0.2385,
+      "step": 602
+    },
+    {
+      "epoch": 0.38592,
+      "grad_norm": 0.32212522458208825,
+      "learning_rate": 3.616297556496369e-05,
+      "loss": 0.2387,
+      "step": 603
+    },
+    {
+      "epoch": 0.38656,
+      "grad_norm": 0.29947124111938617,
+      "learning_rate": 3.6115563232000206e-05,
+      "loss": 0.2357,
+      "step": 604
+    },
+    {
+      "epoch": 0.3872,
+      "grad_norm": 0.27587166347065994,
+      "learning_rate": 3.6068101016474314e-05,
+      "loss": 0.2386,
+      "step": 605
+    },
+    {
+      "epoch": 0.38784,
+      "grad_norm": 0.3128647806581742,
+      "learning_rate": 3.602058913137902e-05,
+      "loss": 0.2549,
+      "step": 606
+    },
+    {
+      "epoch": 0.38848,
+      "grad_norm": 0.2629559804900656,
+      "learning_rate": 3.597302778993018e-05,
+      "loss": 0.2544,
+      "step": 607
+    },
+    {
+      "epoch": 0.38912,
+      "grad_norm": 0.29509601479077685,
+      "learning_rate": 3.592541720556561e-05,
+      "loss": 0.247,
+      "step": 608
+    },
+    {
+      "epoch": 0.38976,
+      "grad_norm": 0.2750936321953339,
+      "learning_rate": 3.5877757591944137e-05,
+      "loss": 0.2551,
+      "step": 609
+    },
+    {
+      "epoch": 0.3904,
+      "grad_norm": 0.2794601458455836,
+      "learning_rate": 3.583004916294458e-05,
+      "loss": 0.2587,
+      "step": 610
+    },
+    {
+      "epoch": 0.39104,
+      "grad_norm": 0.27724909989488333,
+      "learning_rate": 3.578229213266483e-05,
+      "loss": 0.2532,
+      "step": 611
+    },
+    {
+      "epoch": 0.39168,
+      "grad_norm": 0.2959448489747847,
+      "learning_rate": 3.573448671542088e-05,
+      "loss": 0.2507,
+      "step": 612
+    },
+    {
+      "epoch": 0.39232,
+      "grad_norm": 0.3149037372365962,
+      "learning_rate": 3.568663312574589e-05,
+      "loss": 0.2359,
+      "step": 613
+    },
+    {
+      "epoch": 0.39296,
+      "grad_norm": 0.28544817752125,
+      "learning_rate": 3.563873157838917e-05,
+      "loss": 0.2591,
+      "step": 614
+    },
+    {
+      "epoch": 0.3936,
+      "grad_norm": 0.2821951863213839,
+      "learning_rate": 3.5590782288315266e-05,
+      "loss": 0.271,
+      "step": 615
+    },
+    {
+      "epoch": 0.39424,
+      "grad_norm": 0.3062006780490601,
+      "learning_rate": 3.5542785470702965e-05,
+      "loss": 0.2584,
+      "step": 616
+    },
+    {
+      "epoch": 0.39488,
+      "grad_norm": 0.26982808623927396,
+      "learning_rate": 3.549474134094434e-05,
+      "loss": 0.2607,
+      "step": 617
+    },
+    {
+      "epoch": 0.39552,
+      "grad_norm": 0.28882535101383605,
+      "learning_rate": 3.54466501146438e-05,
+      "loss": 0.2439,
+      "step": 618
+    },
+    {
+      "epoch": 0.39616,
+      "grad_norm": 0.3054230159989776,
+      "learning_rate": 3.539851200761707e-05,
+      "loss": 0.283,
+      "step": 619
+    },
+    {
+      "epoch": 0.3968,
+      "grad_norm": 0.2803994922266339,
+      "learning_rate": 3.535032723589031e-05,
+      "loss": 0.2401,
+      "step": 620
+    },
+    {
+      "epoch": 0.39744,
+      "grad_norm": 0.2829193733985107,
+      "learning_rate": 3.5302096015699035e-05,
+      "loss": 0.2532,
+      "step": 621
+    },
+    {
+      "epoch": 0.39808,
+      "grad_norm": 0.29493309388688765,
+      "learning_rate": 3.5253818563487266e-05,
+      "loss": 0.2502,
+      "step": 622
+    },
+    {
+      "epoch": 0.39872,
+      "grad_norm": 0.2616815315666944,
+      "learning_rate": 3.5205495095906444e-05,
+      "loss": 0.2515,
+      "step": 623
+    },
+    {
+      "epoch": 0.39936,
+      "grad_norm": 0.2679431721589359,
+      "learning_rate": 3.5157125829814534e-05,
+      "loss": 0.2436,
+      "step": 624
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.25554211659718934,
+      "learning_rate": 3.510871098227503e-05,
+      "loss": 0.2324,
+      "step": 625
+    },
+    {
+      "epoch": 0.40064,
+      "grad_norm": 0.2733276800637646,
+      "learning_rate": 3.506025077055597e-05,
+      "loss": 0.2914,
+      "step": 626
+    },
+    {
+      "epoch": 0.40128,
+      "grad_norm": 0.26328591986272304,
+      "learning_rate": 3.501174541212898e-05,
+      "loss": 0.2424,
+      "step": 627
+    },
+    {
+      "epoch": 0.40192,
+      "grad_norm": 0.2644108716838522,
+      "learning_rate": 3.496319512466827e-05,
+      "loss": 0.2555,
+      "step": 628
+    },
+    {
+      "epoch": 0.40256,
+      "grad_norm": 0.27240010844795565,
+      "learning_rate": 3.4914600126049696e-05,
+      "loss": 0.2458,
+      "step": 629
+    },
+    {
+      "epoch": 0.4032,
+      "grad_norm": 0.275592750641108,
+      "learning_rate": 3.486596063434975e-05,
+      "loss": 0.243,
+      "step": 630
+    },
+    {
+      "epoch": 0.40384,
+      "grad_norm": 0.28461986974911496,
+      "learning_rate": 3.481727686784458e-05,
+      "loss": 0.2442,
+      "step": 631
+    },
+    {
+      "epoch": 0.40448,
+      "grad_norm": 0.38358418504867753,
+      "learning_rate": 3.476854904500905e-05,
+      "loss": 0.2476,
+      "step": 632
+    },
+    {
+      "epoch": 0.40512,
+      "grad_norm": 0.27772995022705443,
+      "learning_rate": 3.471977738451572e-05,
+      "loss": 0.2532,
+      "step": 633
+    },
+    {
+      "epoch": 0.40576,
+      "grad_norm": 0.3063163796379174,
+      "learning_rate": 3.467096210523387e-05,
+      "loss": 0.2528,
+      "step": 634
+    },
+    {
+      "epoch": 0.4064,
+      "grad_norm": 0.2558103094662568,
+      "learning_rate": 3.4622103426228534e-05,
+      "loss": 0.2617,
+      "step": 635
+    },
+    {
+      "epoch": 0.40704,
+      "grad_norm": 0.27035171229910127,
+      "learning_rate": 3.4573201566759515e-05,
+      "loss": 0.2459,
+      "step": 636
+    },
+    {
+      "epoch": 0.40768,
+      "grad_norm": 0.2829589875667949,
+      "learning_rate": 3.4524256746280374e-05,
+      "loss": 0.2553,
+      "step": 637
+    },
+    {
+      "epoch": 0.40832,
+      "grad_norm": 0.276488759483485,
+      "learning_rate": 3.447526918443748e-05,
+      "loss": 0.2494,
+      "step": 638
+    },
+    {
+      "epoch": 0.40896,
+      "grad_norm": 0.2785085755001678,
+      "learning_rate": 3.442623910106901e-05,
+      "loss": 0.2308,
+      "step": 639
+    },
+    {
+      "epoch": 0.4096,
+      "grad_norm": 0.2582594146057901,
+      "learning_rate": 3.4377166716203945e-05,
+      "loss": 0.2401,
+      "step": 640
+    },
+    {
+      "epoch": 0.41024,
+      "grad_norm": 0.29532893078531347,
+      "learning_rate": 3.4328052250061124e-05,
+      "loss": 0.2464,
+      "step": 641
+    },
+    {
+      "epoch": 0.41088,
+      "grad_norm": 0.29066328186544355,
+      "learning_rate": 3.4278895923048226e-05,
+      "loss": 0.2556,
+      "step": 642
+    },
+    {
+      "epoch": 0.41152,
+      "grad_norm": 0.2687576511766121,
+      "learning_rate": 3.4229697955760767e-05,
+      "loss": 0.2287,
+      "step": 643
+    },
+    {
+      "epoch": 0.41216,
+      "grad_norm": 0.2835568843137537,
+      "learning_rate": 3.4180458568981136e-05,
+      "loss": 0.2797,
+      "step": 644
+    },
+    {
+      "epoch": 0.4128,
+      "grad_norm": 0.2629010441848825,
+      "learning_rate": 3.4131177983677624e-05,
+      "loss": 0.226,
+      "step": 645
+    },
+    {
+      "epoch": 0.41344,
+      "grad_norm": 0.28062825574605843,
+      "learning_rate": 3.4081856421003364e-05,
+      "loss": 0.2367,
+      "step": 646
+    },
+    {
+      "epoch": 0.41408,
+      "grad_norm": 0.270318316756242,
+      "learning_rate": 3.403249410229542e-05,
+      "loss": 0.2467,
+      "step": 647
+    },
+    {
+      "epoch": 0.41472,
+      "grad_norm": 0.316058725045681,
+      "learning_rate": 3.398309124907372e-05,
+      "loss": 0.2567,
+      "step": 648
+    },
+    {
+      "epoch": 0.41536,
+      "grad_norm": 0.24447817887936557,
+      "learning_rate": 3.393364808304012e-05,
+      "loss": 0.2718,
+      "step": 649
+    },
+    {
+      "epoch": 0.416,
+      "grad_norm": 0.2894191788618239,
+      "learning_rate": 3.3884164826077384e-05,
+      "loss": 0.2449,
+      "step": 650
+    },
+    {
+      "epoch": 0.41664,
+      "grad_norm": 0.2912762195649435,
+      "learning_rate": 3.383464170024817e-05,
+      "loss": 0.2441,
+      "step": 651
+    },
+    {
+      "epoch": 0.41728,
+      "grad_norm": 0.28705421045348156,
+      "learning_rate": 3.378507892779408e-05,
+      "loss": 0.2707,
+      "step": 652
+    },
+    {
+      "epoch": 0.41792,
+      "grad_norm": 0.29055146905014817,
+      "learning_rate": 3.37354767311346e-05,
+      "loss": 0.2536,
+      "step": 653
+    },
+    {
+      "epoch": 0.41856,
+      "grad_norm": 0.26018867708452015,
+      "learning_rate": 3.368583533286617e-05,
+      "loss": 0.2355,
+      "step": 654
+    },
+    {
+      "epoch": 0.4192,
+      "grad_norm": 0.26817580454089857,
+      "learning_rate": 3.363615495576115e-05,
+      "loss": 0.2566,
+      "step": 655
+    },
+    {
+      "epoch": 0.41984,
+      "grad_norm": 0.2762550908141269,
+      "learning_rate": 3.3586435822766785e-05,
+      "loss": 0.2566,
+      "step": 656
+    },
+    {
+      "epoch": 0.42048,
+      "grad_norm": 0.27020213820169053,
+      "learning_rate": 3.35366781570043e-05,
+      "loss": 0.2359,
+      "step": 657
+    },
+    {
+      "epoch": 0.42112,
+      "grad_norm": 0.302439548213349,
+      "learning_rate": 3.3486882181767785e-05,
+      "loss": 0.2548,
+      "step": 658
+    },
+    {
+      "epoch": 0.42176,
+      "grad_norm": 0.300389758723506,
+      "learning_rate": 3.34370481205233e-05,
+      "loss": 0.2618,
+      "step": 659
+    },
+    {
+      "epoch": 0.4224,
+      "grad_norm": 0.2713457188069892,
+      "learning_rate": 3.3387176196907776e-05,
+      "loss": 0.2607,
+      "step": 660
+    },
+    {
+      "epoch": 0.42304,
+      "grad_norm": 0.29789023190014424,
+      "learning_rate": 3.333726663472809e-05,
+      "loss": 0.2415,
+      "step": 661
+    },
+    {
+      "epoch": 0.42368,
+      "grad_norm": 0.2773543670343778,
+      "learning_rate": 3.328731965796e-05,
+      "loss": 0.2446,
+      "step": 662
+    },
+    {
+      "epoch": 0.42432,
+      "grad_norm": 0.2783765933442811,
+      "learning_rate": 3.32373354907472e-05,
+      "loss": 0.2339,
+      "step": 663
+    },
+    {
+      "epoch": 0.42496,
+      "grad_norm": 0.25339658241046037,
+      "learning_rate": 3.3187314357400244e-05,
+      "loss": 0.2346,
+      "step": 664
+    },
+    {
+      "epoch": 0.4256,
+      "grad_norm": 0.2648778149152874,
+      "learning_rate": 3.31372564823956e-05,
+      "loss": 0.2644,
+      "step": 665
+    },
+    {
+      "epoch": 0.42624,
+      "grad_norm": 0.2862572466968127,
+      "learning_rate": 3.308716209037462e-05,
+      "loss": 0.2548,
+      "step": 666
+    },
+    {
+      "epoch": 0.42688,
+      "grad_norm": 0.2586423249267937,
+      "learning_rate": 3.303703140614251e-05,
+      "loss": 0.2531,
+      "step": 667
+    },
+    {
+      "epoch": 0.42752,
+      "grad_norm": 0.4114872279281381,
+      "learning_rate": 3.2986864654667356e-05,
+      "loss": 0.2405,
+      "step": 668
+    },
+    {
+      "epoch": 0.42816,
+      "grad_norm": 0.24446086855481872,
+      "learning_rate": 3.293666206107911e-05,
+      "loss": 0.2515,
+      "step": 669
+    },
+    {
+      "epoch": 0.4288,
+      "grad_norm": 0.2791407834679966,
+      "learning_rate": 3.2886423850668555e-05,
+      "loss": 0.2493,
+      "step": 670
+    },
+    {
+      "epoch": 0.42944,
+      "grad_norm": 0.2741780919630588,
+      "learning_rate": 3.283615024888632e-05,
+      "loss": 0.2402,
+      "step": 671
+    },
+    {
+      "epoch": 0.43008,
+      "grad_norm": 0.293985795428857,
+      "learning_rate": 3.278584148134185e-05,
+      "loss": 0.2682,
+      "step": 672
+    },
+    {
+      "epoch": 0.43072,
+      "grad_norm": 0.2915497388084258,
+      "learning_rate": 3.2735497773802406e-05,
+      "loss": 0.2686,
+      "step": 673
+    },
+    {
+      "epoch": 0.43136,
+      "grad_norm": 0.2794624861425458,
+      "learning_rate": 3.268511935219204e-05,
+      "loss": 0.2418,
+      "step": 674
+    },
+    {
+      "epoch": 0.432,
+      "grad_norm": 0.2644683704239771,
+      "learning_rate": 3.263470644259059e-05,
+      "loss": 0.2225,
+      "step": 675
+    },
+    {
+      "epoch": 0.43264,
+      "grad_norm": 0.29839214475582065,
+      "learning_rate": 3.258425927123266e-05,
+      "loss": 0.2665,
+      "step": 676
+    },
+    {
+      "epoch": 0.43328,
+      "grad_norm": 0.2804235795912977,
+      "learning_rate": 3.253377806450664e-05,
+      "loss": 0.2491,
+      "step": 677
+    },
+    {
+      "epoch": 0.43392,
+      "grad_norm": 0.2552634500085142,
+      "learning_rate": 3.248326304895361e-05,
+      "loss": 0.2256,
+      "step": 678
+    },
+    {
+      "epoch": 0.43456,
+      "grad_norm": 0.27227200788700007,
+      "learning_rate": 3.2432714451266395e-05,
+      "loss": 0.2609,
+      "step": 679
+    },
+    {
+      "epoch": 0.4352,
+      "grad_norm": 0.27105089367623647,
+      "learning_rate": 3.238213249828853e-05,
+      "loss": 0.2587,
+      "step": 680
+    },
+    {
+      "epoch": 0.43584,
+      "grad_norm": 0.26509214505181616,
+      "learning_rate": 3.233151741701322e-05,
+      "loss": 0.2498,
+      "step": 681
+    },
+    {
+      "epoch": 0.43648,
+      "grad_norm": 0.2587788903976874,
+      "learning_rate": 3.228086943458235e-05,
+      "loss": 0.2562,
+      "step": 682
+    },
+    {
+      "epoch": 0.43712,
+      "grad_norm": 0.25091179033958944,
+      "learning_rate": 3.223018877828546e-05,
+      "loss": 0.2214,
+      "step": 683
+    },
+    {
+      "epoch": 0.43776,
+      "grad_norm": 0.2793253170166926,
+      "learning_rate": 3.2179475675558694e-05,
+      "loss": 0.2463,
+      "step": 684
+    },
+    {
+      "epoch": 0.4384,
+      "grad_norm": 0.2407880517047661,
+      "learning_rate": 3.212873035398383e-05,
+      "loss": 0.2493,
+      "step": 685
+    },
+    {
+      "epoch": 0.43904,
+      "grad_norm": 0.26757070589299814,
+      "learning_rate": 3.207795304128721e-05,
+      "loss": 0.2293,
+      "step": 686
+    },
+    {
+      "epoch": 0.43968,
+      "grad_norm": 0.26723528784093614,
+      "learning_rate": 3.2027143965338765e-05,
+      "loss": 0.2484,
+      "step": 687
+    },
+    {
+      "epoch": 0.44032,
+      "grad_norm": 0.2656705855889354,
+      "learning_rate": 3.1976303354150935e-05,
+      "loss": 0.2236,
+      "step": 688
+    },
+    {
+      "epoch": 0.44096,
+      "grad_norm": 0.2439985371889559,
+      "learning_rate": 3.192543143587771e-05,
+      "loss": 0.2342,
+      "step": 689
+    },
+    {
+      "epoch": 0.4416,
+      "grad_norm": 0.26235220172661927,
+      "learning_rate": 3.187452843881357e-05,
+      "loss": 0.248,
+      "step": 690
+    },
+    {
+      "epoch": 0.44224,
+      "grad_norm": 0.25083797196574825,
+      "learning_rate": 3.182359459139243e-05,
+      "loss": 0.2262,
+      "step": 691
+    },
+    {
+      "epoch": 0.44288,
+      "grad_norm": 0.3227207880943149,
+      "learning_rate": 3.1772630122186695e-05,
+      "loss": 0.249,
+      "step": 692
+    },
+    {
+      "epoch": 0.44352,
+      "grad_norm": 0.2510582734287034,
+      "learning_rate": 3.172163525990617e-05,
+      "loss": 0.2642,
+      "step": 693
+    },
+    {
+      "epoch": 0.44416,
+      "grad_norm": 0.23405484365904264,
+      "learning_rate": 3.167061023339706e-05,
+      "loss": 0.25,
+      "step": 694
+    },
+    {
+      "epoch": 0.4448,
+      "grad_norm": 0.25693251630337804,
+      "learning_rate": 3.161955527164092e-05,
+      "loss": 0.2509,
+      "step": 695
+    },
+    {
+      "epoch": 0.44544,
+      "grad_norm": 0.24145057284019783,
+      "learning_rate": 3.1568470603753666e-05,
+      "loss": 0.2369,
+      "step": 696
+    },
+    {
+      "epoch": 0.44608,
+      "grad_norm": 0.2656422680687307,
+      "learning_rate": 3.1517356458984485e-05,
+      "loss": 0.2483,
+      "step": 697
+    },
+    {
+      "epoch": 0.44672,
+      "grad_norm": 0.2511244867492243,
+      "learning_rate": 3.14662130667149e-05,
+      "loss": 0.2218,
+      "step": 698
+    },
+    {
+      "epoch": 0.44736,
+      "grad_norm": 0.27645698946904024,
+      "learning_rate": 3.141504065645765e-05,
+      "loss": 0.2597,
+      "step": 699
+    },
+    {
+      "epoch": 0.448,
+      "grad_norm": 0.2345290138656015,
+      "learning_rate": 3.1363839457855704e-05,
+      "loss": 0.2471,
+      "step": 700
+    },
+    {
+      "epoch": 0.44864,
+      "grad_norm": 0.2550749778653545,
+      "learning_rate": 3.131260970068122e-05,
+      "loss": 0.2474,
+      "step": 701
+    },
+    {
+      "epoch": 0.44928,
+      "grad_norm": 0.26631197327016853,
+      "learning_rate": 3.1261351614834536e-05,
+      "loss": 0.2383,
+      "step": 702
+    },
+    {
+      "epoch": 0.44992,
+      "grad_norm": 0.22169396341875647,
+      "learning_rate": 3.121006543034307e-05,
+      "loss": 0.2601,
+      "step": 703
+    },
+    {
+      "epoch": 0.45056,
+      "grad_norm": 0.2388958834756383,
+      "learning_rate": 3.11587513773604e-05,
+      "loss": 0.2224,
+      "step": 704
+    },
+    {
+      "epoch": 0.4512,
+      "grad_norm": 0.24836043244890216,
+      "learning_rate": 3.1107409686165105e-05,
+      "loss": 0.2499,
+      "step": 705
+    },
+    {
+      "epoch": 0.45184,
+      "grad_norm": 0.2641275652959496,
+      "learning_rate": 3.1056040587159856e-05,
+      "loss": 0.2273,
+      "step": 706
+    },
+    {
+      "epoch": 0.45248,
+      "grad_norm": 0.2653747554412634,
+      "learning_rate": 3.100464431087027e-05,
+      "loss": 0.2348,
+      "step": 707
+    },
+    {
+      "epoch": 0.45312,
+      "grad_norm": 0.2554298939523283,
+      "learning_rate": 3.095322108794396e-05,
+      "loss": 0.2352,
+      "step": 708
+    },
+    {
+      "epoch": 0.45376,
+      "grad_norm": 0.26559974477527737,
+      "learning_rate": 3.0901771149149434e-05,
+      "loss": 0.227,
+      "step": 709
+    },
+    {
+      "epoch": 0.4544,
+      "grad_norm": 0.24105288644141198,
+      "learning_rate": 3.0850294725375105e-05,
+      "loss": 0.2456,
+      "step": 710
+    },
+    {
+      "epoch": 0.45504,
+      "grad_norm": 0.25639595036381535,
+      "learning_rate": 3.079879204762827e-05,
+      "loss": 0.2335,
+      "step": 711
+    },
+    {
+      "epoch": 0.45568,
+      "grad_norm": 0.2666175090966313,
+      "learning_rate": 3.074726334703399e-05,
+      "loss": 0.2375,
+      "step": 712
+    },
+    {
+      "epoch": 0.45632,
+      "grad_norm": 0.24609688887166636,
+      "learning_rate": 3.069570885483414e-05,
+      "loss": 0.2497,
+      "step": 713
+    },
+    {
+      "epoch": 0.45696,
+      "grad_norm": 0.26685315906241225,
+      "learning_rate": 3.064412880238636e-05,
+      "loss": 0.2498,
+      "step": 714
+    },
+    {
+      "epoch": 0.4576,
+      "grad_norm": 0.2688506170038658,
+      "learning_rate": 3.059252342116293e-05,
+      "loss": 0.2481,
+      "step": 715
+    },
+    {
+      "epoch": 0.45824,
+      "grad_norm": 0.2594010390645574,
+      "learning_rate": 3.054089294274985e-05,
+      "loss": 0.2468,
+      "step": 716
+    },
+    {
+      "epoch": 0.45888,
+      "grad_norm": 0.28084138121726687,
+      "learning_rate": 3.048923759884573e-05,
+      "loss": 0.2362,
+      "step": 717
+    },
+    {
+      "epoch": 0.45952,
+      "grad_norm": 0.2923747466796738,
+      "learning_rate": 3.043755762126077e-05,
+      "loss": 0.2619,
+      "step": 718
+    },
+    {
+      "epoch": 0.46016,
+      "grad_norm": 0.2575669206483763,
+      "learning_rate": 3.038585324191571e-05,
+      "loss": 0.2529,
+      "step": 719
+    },
+    {
+      "epoch": 0.4608,
+      "grad_norm": 0.24997801681198326,
+      "learning_rate": 3.0334124692840815e-05,
+      "loss": 0.2558,
+      "step": 720
+    },
+    {
+      "epoch": 0.46144,
+      "grad_norm": 0.28797967225037385,
+      "learning_rate": 3.0282372206174785e-05,
+      "loss": 0.2498,
+      "step": 721
+    },
+    {
+      "epoch": 0.46208,
+      "grad_norm": 0.29276072080394017,
+      "learning_rate": 3.023059601416376e-05,
+      "loss": 0.2492,
+      "step": 722
+    },
+    {
+      "epoch": 0.46272,
+      "grad_norm": 0.24024625168643718,
+      "learning_rate": 3.017879634916027e-05,
+      "loss": 0.24,
+      "step": 723
+    },
+    {
+      "epoch": 0.46336,
+      "grad_norm": 0.250522628591233,
+      "learning_rate": 3.012697344362217e-05,
+      "loss": 0.2354,
+      "step": 724
+    },
+    {
+      "epoch": 0.464,
+      "grad_norm": 0.28028684047213753,
+      "learning_rate": 3.0075127530111606e-05,
+      "loss": 0.2248,
+      "step": 725
+    },
+    {
+      "epoch": 0.46464,
+      "grad_norm": 0.2791381712788782,
+      "learning_rate": 3.0023258841294e-05,
+      "loss": 0.2425,
+      "step": 726
+    },
+    {
+      "epoch": 0.46528,
+      "grad_norm": 0.2659960218700461,
+      "learning_rate": 2.9971367609936945e-05,
+      "loss": 0.2526,
+      "step": 727
+    },
+    {
+      "epoch": 0.46592,
+      "grad_norm": 0.2772736125798831,
+      "learning_rate": 2.9919454068909218e-05,
+      "loss": 0.2247,
+      "step": 728
+    },
+    {
+      "epoch": 0.46656,
+      "grad_norm": 0.3124309146865046,
+      "learning_rate": 2.986751845117972e-05,
+      "loss": 0.2569,
+      "step": 729
+    },
+    {
+      "epoch": 0.4672,
+      "grad_norm": 0.2626897535417429,
+      "learning_rate": 2.981556098981641e-05,
+      "loss": 0.2806,
+      "step": 730
+    },
+    {
+      "epoch": 0.46784,
+      "grad_norm": 0.29956166216034524,
+      "learning_rate": 2.9763581917985274e-05,
+      "loss": 0.2636,
+      "step": 731
+    },
+    {
+      "epoch": 0.46848,
+      "grad_norm": 0.2911940048402578,
+      "learning_rate": 2.9711581468949295e-05,
+      "loss": 0.2581,
+      "step": 732
+    },
+    {
+      "epoch": 0.46912,
+      "grad_norm": 0.3066474268110775,
+      "learning_rate": 2.965955987606736e-05,
+      "loss": 0.2713,
+      "step": 733
+    },
+    {
+      "epoch": 0.46976,
+      "grad_norm": 0.2948399561745436,
+      "learning_rate": 2.9607517372793268e-05,
+      "loss": 0.2504,
+      "step": 734
+    },
+    {
+      "epoch": 0.4704,
+      "grad_norm": 0.29922556962835833,
+      "learning_rate": 2.9555454192674638e-05,
+      "loss": 0.2506,
+      "step": 735
+    },
+    {
+      "epoch": 0.47104,
+      "grad_norm": 0.3090688707981366,
+      "learning_rate": 2.9503370569351895e-05,
+      "loss": 0.2616,
+      "step": 736
+    },
+    {
+      "epoch": 0.47168,
+      "grad_norm": 0.2761500484620512,
+      "learning_rate": 2.9451266736557202e-05,
+      "loss": 0.2577,
+      "step": 737
+    },
+    {
+      "epoch": 0.47232,
+      "grad_norm": 0.254217878635331,
+      "learning_rate": 2.9399142928113403e-05,
+      "loss": 0.2189,
+      "step": 738
+    },
+    {
+      "epoch": 0.47296,
+      "grad_norm": 0.26296653857821706,
+      "learning_rate": 2.9346999377932992e-05,
+      "loss": 0.2634,
+      "step": 739
+    },
+    {
+      "epoch": 0.4736,
+      "grad_norm": 0.2572341383024019,
+      "learning_rate": 2.929483632001706e-05,
+      "loss": 0.2417,
+      "step": 740
+    },
+    {
+      "epoch": 0.47424,
+      "grad_norm": 0.28892208828624116,
+      "learning_rate": 2.9242653988454237e-05,
+      "loss": 0.2445,
+      "step": 741
+    },
+    {
+      "epoch": 0.47488,
+      "grad_norm": 0.28102463284648577,
+      "learning_rate": 2.919045261741965e-05,
+      "loss": 0.2388,
+      "step": 742
+    },
+    {
+      "epoch": 0.47552,
+      "grad_norm": 0.24662076667210664,
+      "learning_rate": 2.9138232441173867e-05,
+      "loss": 0.2383,
+      "step": 743
+    },
+    {
+      "epoch": 0.47616,
+      "grad_norm": 0.29423089365699157,
+      "learning_rate": 2.9085993694061853e-05,
+      "loss": 0.2425,
+      "step": 744
+    },
+    {
+      "epoch": 0.4768,
+      "grad_norm": 0.2957483091439502,
+      "learning_rate": 2.9033736610511887e-05,
+      "loss": 0.2617,
+      "step": 745
+    },
+    {
+      "epoch": 0.47744,
+      "grad_norm": 0.28663575693261895,
+      "learning_rate": 2.898146142503456e-05,
+      "loss": 0.2633,
+      "step": 746
+    },
+    {
+      "epoch": 0.47808,
+      "grad_norm": 0.2603089054365686,
+      "learning_rate": 2.8929168372221693e-05,
+      "loss": 0.2496,
+      "step": 747
+    },
+    {
+      "epoch": 0.47872,
+      "grad_norm": 0.29032157339351555,
+      "learning_rate": 2.88768576867453e-05,
+      "loss": 0.2562,
+      "step": 748
+    },
+    {
+      "epoch": 0.47936,
+      "grad_norm": 0.25757922984484416,
+      "learning_rate": 2.8824529603356497e-05,
+      "loss": 0.2233,
+      "step": 749
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.29211519919338724,
+      "learning_rate": 2.8772184356884496e-05,
+      "loss": 0.2602,
+      "step": 750
+    },
+    {
+      "epoch": 0.48064,
+      "grad_norm": 0.26552159476837556,
+      "learning_rate": 2.8719822182235516e-05,
+      "loss": 0.2551,
+      "step": 751
+    },
+    {
+      "epoch": 0.48128,
+      "grad_norm": 0.2713416231519419,
+      "learning_rate": 2.8667443314391762e-05,
+      "loss": 0.2291,
+      "step": 752
+    },
+    {
+      "epoch": 0.48192,
+      "grad_norm": 0.2688122633975986,
+      "learning_rate": 2.8615047988410337e-05,
+      "loss": 0.2426,
+      "step": 753
+    },
+    {
+      "epoch": 0.48256,
+      "grad_norm": 0.2552031893223486,
+      "learning_rate": 2.8562636439422202e-05,
+      "loss": 0.243,
+      "step": 754
+    },
+    {
+      "epoch": 0.4832,
+      "grad_norm": 0.25644480909672956,
+      "learning_rate": 2.8510208902631135e-05,
+      "loss": 0.2507,
+      "step": 755
+    },
+    {
+      "epoch": 0.48384,
+      "grad_norm": 0.2650303461926079,
+      "learning_rate": 2.845776561331265e-05,
+      "loss": 0.2524,
+      "step": 756
+    },
+    {
+      "epoch": 0.48448,
+      "grad_norm": 0.2841325794413843,
+      "learning_rate": 2.8405306806812947e-05,
+      "loss": 0.2246,
+      "step": 757
+    },
+    {
+      "epoch": 0.48512,
+      "grad_norm": 0.27963333968219384,
+      "learning_rate": 2.835283271854788e-05,
+      "loss": 0.2627,
+      "step": 758
+    },
+    {
+      "epoch": 0.48576,
+      "grad_norm": 0.2469872711729761,
+      "learning_rate": 2.8300343584001853e-05,
+      "loss": 0.2456,
+      "step": 759
+    },
+    {
+      "epoch": 0.4864,
+      "grad_norm": 0.2539236034985644,
+      "learning_rate": 2.8247839638726825e-05,
+      "loss": 0.2586,
+      "step": 760
+    },
+    {
+      "epoch": 0.48704,
+      "grad_norm": 0.25645384647167085,
+      "learning_rate": 2.8195321118341204e-05,
+      "loss": 0.2404,
+      "step": 761
+    },
+    {
+      "epoch": 0.48768,
+      "grad_norm": 0.3015394883426219,
+      "learning_rate": 2.8142788258528813e-05,
+      "loss": 0.2571,
+      "step": 762
+    },
+    {
+      "epoch": 0.48832,
+      "grad_norm": 0.26599432217072383,
+      "learning_rate": 2.8090241295037804e-05,
+      "loss": 0.2534,
+      "step": 763
+    },
+    {
+      "epoch": 0.48896,
+      "grad_norm": 0.26116413484343565,
+      "learning_rate": 2.8037680463679648e-05,
+      "loss": 0.2473,
+      "step": 764
+    },
+    {
+      "epoch": 0.4896,
+      "grad_norm": 0.25013902597670157,
+      "learning_rate": 2.7985106000328033e-05,
+      "loss": 0.2406,
+      "step": 765
+    },
+    {
+      "epoch": 0.49024,
+      "grad_norm": 0.27203497182718117,
+      "learning_rate": 2.793251814091782e-05,
+      "loss": 0.2364,
+      "step": 766
+    },
+    {
+      "epoch": 0.49088,
+      "grad_norm": 0.2809161478373533,
+      "learning_rate": 2.787991712144401e-05,
+      "loss": 0.2598,
+      "step": 767
+    },
+    {
+      "epoch": 0.49152,
+      "grad_norm": 0.2594480739530536,
+      "learning_rate": 2.7827303177960638e-05,
+      "loss": 0.2655,
+      "step": 768
+    },
+    {
+      "epoch": 0.49216,
+      "grad_norm": 0.2613564843549674,
+      "learning_rate": 2.7774676546579737e-05,
+      "loss": 0.2518,
+      "step": 769
+    },
+    {
+      "epoch": 0.4928,
+      "grad_norm": 0.2628312412666247,
+      "learning_rate": 2.7722037463470297e-05,
+      "loss": 0.2314,
+      "step": 770
+    },
+    {
+      "epoch": 0.49344,
+      "grad_norm": 0.24433922482979678,
+      "learning_rate": 2.7669386164857174e-05,
+      "loss": 0.216,
+      "step": 771
+    },
+    {
+      "epoch": 0.49408,
+      "grad_norm": 0.2811567735489263,
+      "learning_rate": 2.7616722887020034e-05,
+      "loss": 0.2317,
+      "step": 772
+    },
+    {
+      "epoch": 0.49472,
+      "grad_norm": 0.2988300663760754,
+      "learning_rate": 2.7564047866292336e-05,
+      "loss": 0.2353,
+      "step": 773
+    },
+    {
+      "epoch": 0.49536,
+      "grad_norm": 0.2462499230606909,
+      "learning_rate": 2.7511361339060193e-05,
+      "loss": 0.2579,
+      "step": 774
+    },
+    {
+      "epoch": 0.496,
+      "grad_norm": 0.23990263065434209,
+      "learning_rate": 2.7458663541761376e-05,
+      "loss": 0.2535,
+      "step": 775
+    },
+    {
+      "epoch": 0.49664,
+      "grad_norm": 0.2604844585372496,
+      "learning_rate": 2.740595471088424e-05,
+      "loss": 0.2329,
+      "step": 776
+    },
+    {
+      "epoch": 0.49728,
+      "grad_norm": 0.5977257472918388,
+      "learning_rate": 2.735323508296664e-05,
+      "loss": 0.2542,
+      "step": 777
+    },
+    {
+      "epoch": 0.49792,
+      "grad_norm": 0.2584304116675002,
+      "learning_rate": 2.7300504894594892e-05,
+      "loss": 0.2422,
+      "step": 778
+    },
+    {
+      "epoch": 0.49856,
+      "grad_norm": 0.25230670266535726,
+      "learning_rate": 2.7247764382402696e-05,
+      "loss": 0.2697,
+      "step": 779
+    },
+    {
+      "epoch": 0.4992,
+      "grad_norm": 0.25239364609932186,
+      "learning_rate": 2.7195013783070093e-05,
+      "loss": 0.2626,
+      "step": 780
+    },
+    {
+      "epoch": 0.49984,
+      "grad_norm": 0.26187905841301573,
+      "learning_rate": 2.7142253333322387e-05,
+      "loss": 0.2486,
+      "step": 781
+    },
+    {
+      "epoch": 0.50048,
+      "grad_norm": 0.24964884815156177,
+      "learning_rate": 2.7089483269929074e-05,
+      "loss": 0.2514,
+      "step": 782
+    },
+    {
+      "epoch": 0.50112,
+      "grad_norm": 0.238711792939048,
+      "learning_rate": 2.703670382970281e-05,
+      "loss": 0.2349,
+      "step": 783
+    },
+    {
+      "epoch": 0.50176,
+      "grad_norm": 0.24042897666733604,
+      "learning_rate": 2.698391524949832e-05,
+      "loss": 0.2398,
+      "step": 784
+    },
+    {
+      "epoch": 0.5024,
+      "grad_norm": 0.2566913545681311,
+      "learning_rate": 2.693111776621136e-05,
+      "loss": 0.2406,
+      "step": 785
+    },
+    {
+      "epoch": 0.50304,
+      "grad_norm": 0.22830002906868818,
+      "learning_rate": 2.6878311616777618e-05,
+      "loss": 0.2455,
+      "step": 786
+    },
+    {
+      "epoch": 0.50368,
+      "grad_norm": 0.27212470243007647,
+      "learning_rate": 2.68254970381717e-05,
+      "loss": 0.2438,
+      "step": 787
+    },
+    {
+      "epoch": 0.50432,
+      "grad_norm": 0.24664299011968796,
+      "learning_rate": 2.6772674267406005e-05,
+      "loss": 0.2629,
+      "step": 788
+    },
+    {
+      "epoch": 0.50496,
+      "grad_norm": 0.24016014731888177,
+      "learning_rate": 2.6719843541529727e-05,
+      "loss": 0.2542,
+      "step": 789
+    },
+    {
+      "epoch": 0.5056,
+      "grad_norm": 0.2838669576909251,
+      "learning_rate": 2.666700509762774e-05,
+      "loss": 0.2351,
+      "step": 790
+    },
+    {
+      "epoch": 0.50624,
+      "grad_norm": 0.2847600127908522,
+      "learning_rate": 2.6614159172819564e-05,
+      "loss": 0.2508,
+      "step": 791
+    },
+    {
+      "epoch": 0.50688,
+      "grad_norm": 0.26634679184834414,
+      "learning_rate": 2.6561306004258284e-05,
+      "loss": 0.2535,
+      "step": 792
+    },
+    {
+      "epoch": 0.50752,
+      "grad_norm": 0.24952395558141793,
+      "learning_rate": 2.6508445829129498e-05,
+      "loss": 0.2556,
+      "step": 793
+    },
+    {
+      "epoch": 0.50816,
+      "grad_norm": 0.2654129783364512,
+      "learning_rate": 2.645557888465024e-05,
+      "loss": 0.2275,
+      "step": 794
+    },
+    {
+      "epoch": 0.5088,
+      "grad_norm": 0.2694040389228673,
+      "learning_rate": 2.640270540806793e-05,
+      "loss": 0.231,
+      "step": 795
+    },
+    {
+      "epoch": 0.50944,
+      "grad_norm": 0.3175921083819656,
+      "learning_rate": 2.6349825636659292e-05,
+      "loss": 0.2753,
+      "step": 796
+    },
+    {
+      "epoch": 0.51008,
+      "grad_norm": 0.2573609623414005,
+      "learning_rate": 2.629693980772931e-05,
+      "loss": 0.2391,
+      "step": 797
+    },
+    {
+      "epoch": 0.51072,
+      "grad_norm": 0.23879176748287295,
+      "learning_rate": 2.6244048158610148e-05,
+      "loss": 0.2517,
+      "step": 798
+    },
+    {
+      "epoch": 0.51136,
+      "grad_norm": 0.26796463054690156,
+      "learning_rate": 2.619115092666008e-05,
+      "loss": 0.232,
+      "step": 799
+    },
+    {
+      "epoch": 0.512,
+      "grad_norm": 0.2966374077475493,
+      "learning_rate": 2.6138248349262446e-05,
+      "loss": 0.2467,
+      "step": 800
+    },
+    {
+      "epoch": 0.51264,
+      "grad_norm": 0.2559019628328055,
+      "learning_rate": 2.608534066382457e-05,
+      "loss": 0.2583,
+      "step": 801
+    },
+    {
+      "epoch": 0.51328,
+      "grad_norm": 0.23331645091348954,
+      "learning_rate": 2.60324281077767e-05,
+      "loss": 0.2304,
+      "step": 802
+    },
+    {
+      "epoch": 0.51392,
+      "grad_norm": 0.272771565582496,
+      "learning_rate": 2.5979510918570943e-05,
+      "loss": 0.2286,
+      "step": 803
+    },
+    {
+      "epoch": 0.51456,
+      "grad_norm": 0.27375943881269327,
+      "learning_rate": 2.592658933368019e-05,
+      "loss": 0.239,
+      "step": 804
+    },
+    {
+      "epoch": 0.5152,
+      "grad_norm": 0.25468027988189307,
+      "learning_rate": 2.5873663590597065e-05,
+      "loss": 0.2224,
+      "step": 805
+    },
+    {
+      "epoch": 0.51584,
+      "grad_norm": 0.23691066477988892,
+      "learning_rate": 2.582073392683284e-05,
+      "loss": 0.2348,
+      "step": 806
+    },
+    {
+      "epoch": 0.51648,
+      "grad_norm": 0.2721627464869961,
+      "learning_rate": 2.5767800579916412e-05,
+      "loss": 0.2322,
+      "step": 807
+    },
+    {
+      "epoch": 0.51712,
+      "grad_norm": 0.2791338963336496,
+      "learning_rate": 2.5714863787393173e-05,
+      "loss": 0.2417,
+      "step": 808
+    },
+    {
+      "epoch": 0.51776,
+      "grad_norm": 0.2716225839252745,
+      "learning_rate": 2.5661923786823994e-05,
+      "loss": 0.2168,
+      "step": 809
+    },
+    {
+      "epoch": 0.5184,
+      "grad_norm": 0.29319302582792406,
+      "learning_rate": 2.5608980815784144e-05,
+      "loss": 0.248,
+      "step": 810
+    },
+    {
+      "epoch": 0.51904,
+      "grad_norm": 0.23430479943354784,
+      "learning_rate": 2.5556035111862213e-05,
+      "loss": 0.2448,
+      "step": 811
+    },
+    {
+      "epoch": 0.51968,
+      "grad_norm": 0.2744979882247978,
+      "learning_rate": 2.550308691265907e-05,
+      "loss": 0.2554,
+      "step": 812
+    },
+    {
+      "epoch": 0.52032,
+      "grad_norm": 0.28406329312925144,
+      "learning_rate": 2.545013645578676e-05,
+      "loss": 0.2586,
+      "step": 813
+    },
+    {
+      "epoch": 0.52096,
+      "grad_norm": 0.2551623881443743,
+      "learning_rate": 2.5397183978867484e-05,
+      "loss": 0.2384,
+      "step": 814
+    },
+    {
+      "epoch": 0.5216,
+      "grad_norm": 0.2580716099842203,
+      "learning_rate": 2.5344229719532486e-05,
+      "loss": 0.2408,
+      "step": 815
+    },
+    {
+      "epoch": 0.52224,
+      "grad_norm": 0.2716922403653216,
+      "learning_rate": 2.529127391542103e-05,
+      "loss": 0.2248,
+      "step": 816
+    },
+    {
+      "epoch": 0.52288,
+      "grad_norm": 0.2837517113091113,
+      "learning_rate": 2.5238316804179296e-05,
+      "loss": 0.2455,
+      "step": 817
+    },
+    {
+      "epoch": 0.52352,
+      "grad_norm": 0.2878366005738279,
+      "learning_rate": 2.5185358623459332e-05,
+      "loss": 0.2355,
+      "step": 818
+    },
+    {
+      "epoch": 0.52416,
+      "grad_norm": 0.2508463586449631,
+      "learning_rate": 2.513239961091799e-05,
+      "loss": 0.2166,
+      "step": 819
+    },
+    {
+      "epoch": 0.5248,
+      "grad_norm": 0.2514543116452259,
+      "learning_rate": 2.507944000421586e-05,
+      "loss": 0.225,
+      "step": 820
+    },
+    {
+      "epoch": 0.52544,
+      "grad_norm": 0.23838166341891015,
+      "learning_rate": 2.502648004101619e-05,
+      "loss": 0.2346,
+      "step": 821
+    },
+    {
+      "epoch": 0.52608,
+      "grad_norm": 0.29590958995133615,
+      "learning_rate": 2.497351995898382e-05,
+      "loss": 0.2394,
+      "step": 822
+    },
+    {
+      "epoch": 0.52672,
+      "grad_norm": 0.2435661972740144,
+      "learning_rate": 2.492055999578415e-05,
+      "loss": 0.2454,
+      "step": 823
+    },
+    {
+      "epoch": 0.52736,
+      "grad_norm": 0.2629808140926325,
+      "learning_rate": 2.486760038908201e-05,
+      "loss": 0.2358,
+      "step": 824
+    },
+    {
+      "epoch": 0.528,
+      "grad_norm": 0.2496659525076201,
+      "learning_rate": 2.481464137654068e-05,
+      "loss": 0.2572,
+      "step": 825
+    },
+    {
+      "epoch": 0.52864,
+      "grad_norm": 0.24943051024145046,
+      "learning_rate": 2.476168319582071e-05,
+      "loss": 0.2272,
+      "step": 826
+    },
+    {
+      "epoch": 0.52928,
+      "grad_norm": 0.2550396348334548,
+      "learning_rate": 2.4708726084578978e-05,
+      "loss": 0.236,
+      "step": 827
+    },
+    {
+      "epoch": 0.52992,
+      "grad_norm": 0.25505946249281997,
+      "learning_rate": 2.465577028046752e-05,
+      "loss": 0.2422,
+      "step": 828
+    },
+    {
+      "epoch": 0.53056,
+      "grad_norm": 0.249271035350032,
+      "learning_rate": 2.4602816021132515e-05,
+      "loss": 0.2286,
+      "step": 829
+    },
+    {
+      "epoch": 0.5312,
+      "grad_norm": 0.2595497603633753,
+      "learning_rate": 2.4549863544213243e-05,
+      "loss": 0.2411,
+      "step": 830
+    },
+    {
+      "epoch": 0.53184,
+      "grad_norm": 0.26723166451301167,
+      "learning_rate": 2.449691308734093e-05,
+      "loss": 0.2715,
+      "step": 831
+    },
+    {
+      "epoch": 0.53248,
+      "grad_norm": 0.2484633693197436,
+      "learning_rate": 2.444396488813779e-05,
+      "loss": 0.2459,
+      "step": 832
+    },
+    {
+      "epoch": 0.53312,
+      "grad_norm": 0.2524686819806887,
+      "learning_rate": 2.4391019184215862e-05,
+      "loss": 0.2876,
+      "step": 833
+    },
+    {
+      "epoch": 0.53376,
+      "grad_norm": 0.2692967304813649,
+      "learning_rate": 2.4338076213176012e-05,
+      "loss": 0.2381,
+      "step": 834
+    },
+    {
+      "epoch": 0.5344,
+      "grad_norm": 0.27066578064017544,
+      "learning_rate": 2.4285136212606836e-05,
+      "loss": 0.2242,
+      "step": 835
+    },
+    {
+      "epoch": 0.53504,
+      "grad_norm": 0.2438814881714075,
+      "learning_rate": 2.42321994200836e-05,
+      "loss": 0.2303,
+      "step": 836
+    },
+    {
+      "epoch": 0.53568,
+      "grad_norm": 0.2722275688117108,
+      "learning_rate": 2.417926607316716e-05,
+      "loss": 0.2448,
+      "step": 837
+    },
+    {
+      "epoch": 0.53632,
+      "grad_norm": 0.2539349871159882,
+      "learning_rate": 2.412633640940295e-05,
+      "loss": 0.2246,
+      "step": 838
+    },
+    {
+      "epoch": 0.53696,
+      "grad_norm": 0.2450942412448625,
+      "learning_rate": 2.4073410666319816e-05,
+      "loss": 0.2483,
+      "step": 839
+    },
+    {
+      "epoch": 0.5376,
+      "grad_norm": 0.24046377830291707,
+      "learning_rate": 2.4020489081429063e-05,
+      "loss": 0.2555,
+      "step": 840
+    },
+    {
+      "epoch": 0.53824,
+      "grad_norm": 0.2530918591131236,
+      "learning_rate": 2.3967571892223305e-05,
+      "loss": 0.2231,
+      "step": 841
+    },
+    {
+      "epoch": 0.53888,
+      "grad_norm": 0.22192550890977153,
+      "learning_rate": 2.3914659336175428e-05,
+      "loss": 0.2283,
+      "step": 842
+    },
+    {
+      "epoch": 0.53952,
+      "grad_norm": 0.26225082453693915,
+      "learning_rate": 2.386175165073756e-05,
+      "loss": 0.2298,
+      "step": 843
+    },
+    {
+      "epoch": 0.54016,
+      "grad_norm": 0.26464899265715924,
+      "learning_rate": 2.3808849073339923e-05,
+      "loss": 0.2361,
+      "step": 844
+    },
+    {
+      "epoch": 0.5408,
+      "grad_norm": 0.2841667761256595,
+      "learning_rate": 2.375595184138986e-05,
+      "loss": 0.2404,
+      "step": 845
+    },
+    {
+      "epoch": 0.54144,
+      "grad_norm": 0.22303490743225543,
+      "learning_rate": 2.3703060192270696e-05,
+      "loss": 0.2317,
+      "step": 846
+    },
+    {
+      "epoch": 0.54208,
+      "grad_norm": 0.23424479953468394,
+      "learning_rate": 2.365017436334072e-05,
+      "loss": 0.2306,
+      "step": 847
+    },
+    {
+      "epoch": 0.54272,
+      "grad_norm": 0.24494905507576614,
+      "learning_rate": 2.3597294591932077e-05,
+      "loss": 0.233,
+      "step": 848
+    },
+    {
+      "epoch": 0.54336,
+      "grad_norm": 0.2509652948143267,
+      "learning_rate": 2.3544421115349772e-05,
+      "loss": 0.2304,
+      "step": 849
+    },
+    {
+      "epoch": 0.544,
+      "grad_norm": 0.2544926498276418,
+      "learning_rate": 2.3491554170870504e-05,
+      "loss": 0.2128,
+      "step": 850
+    },
+    {
+      "epoch": 0.54464,
+      "grad_norm": 0.27906813385984347,
+      "learning_rate": 2.343869399574172e-05,
+      "loss": 0.2334,
+      "step": 851
+    },
+    {
+      "epoch": 0.54528,
+      "grad_norm": 0.2607125470447707,
+      "learning_rate": 2.3385840827180445e-05,
+      "loss": 0.2397,
+      "step": 852
+    },
+    {
+      "epoch": 0.54592,
+      "grad_norm": 0.2838209360177119,
+      "learning_rate": 2.333299490237226e-05,
+      "loss": 0.2512,
+      "step": 853
+    },
+    {
+      "epoch": 0.54656,
+      "grad_norm": 0.2743757757237935,
+      "learning_rate": 2.3280156458470282e-05,
+      "loss": 0.2396,
+      "step": 854
+    },
+    {
+      "epoch": 0.5472,
+      "grad_norm": 0.26262382562949216,
+      "learning_rate": 2.3227325732593997e-05,
+      "loss": 0.2344,
+      "step": 855
+    },
+    {
+      "epoch": 0.54784,
+      "grad_norm": 0.25132172823400895,
+      "learning_rate": 2.317450296182831e-05,
+      "loss": 0.2527,
+      "step": 856
+    },
+    {
+      "epoch": 0.54848,
+      "grad_norm": 0.2709068292103184,
+      "learning_rate": 2.3121688383222384e-05,
+      "loss": 0.2264,
+      "step": 857
+    },
+    {
+      "epoch": 0.54912,
+      "grad_norm": 0.25611652073359836,
+      "learning_rate": 2.306888223378865e-05,
+      "loss": 0.246,
+      "step": 858
+    },
+    {
+      "epoch": 0.54976,
+      "grad_norm": 0.24829165113802293,
+      "learning_rate": 2.3016084750501685e-05,
+      "loss": 0.2499,
+      "step": 859
+    },
+    {
+      "epoch": 0.5504,
+      "grad_norm": 0.25718141898362745,
+      "learning_rate": 2.2963296170297204e-05,
+      "loss": 0.2428,
+      "step": 860
+    },
+    {
+      "epoch": 0.55104,
+      "grad_norm": 0.251334804450548,
+      "learning_rate": 2.2910516730070932e-05,
+      "loss": 0.2508,
+      "step": 861
+    },
+    {
+      "epoch": 0.55168,
+      "grad_norm": 0.25911057587539127,
+      "learning_rate": 2.2857746666677626e-05,
+      "loss": 0.2127,
+      "step": 862
+    },
+    {
+      "epoch": 0.55232,
+      "grad_norm": 0.24325904115468286,
+      "learning_rate": 2.280498621692991e-05,
+      "loss": 0.2341,
+      "step": 863
+    },
+    {
+      "epoch": 0.55296,
+      "grad_norm": 0.2364393515160929,
+      "learning_rate": 2.275223561759731e-05,
+      "loss": 0.2467,
+      "step": 864
+    },
+    {
+      "epoch": 0.5536,
+      "grad_norm": 0.2709372463182954,
+      "learning_rate": 2.2699495105405117e-05,
+      "loss": 0.2413,
+      "step": 865
+    },
+    {
+      "epoch": 0.55424,
+      "grad_norm": 0.25370125098514806,
+      "learning_rate": 2.264676491703336e-05,
+      "loss": 0.2274,
+      "step": 866
+    },
+    {
+      "epoch": 0.55488,
+      "grad_norm": 0.27606834553829646,
+      "learning_rate": 2.2594045289115764e-05,
+      "loss": 0.2257,
+      "step": 867
+    },
+    {
+      "epoch": 0.55552,
+      "grad_norm": 0.24708188794738328,
+      "learning_rate": 2.254133645823862e-05,
+      "loss": 0.2325,
+      "step": 868
+    },
+    {
+      "epoch": 0.55616,
+      "grad_norm": 0.49705005766307636,
+      "learning_rate": 2.2488638660939816e-05,
+      "loss": 0.25,
+      "step": 869
+    },
+    {
+      "epoch": 0.5568,
+      "grad_norm": 0.2553166538382622,
+      "learning_rate": 2.243595213370767e-05,
+      "loss": 0.2351,
+      "step": 870
+    },
+    {
+      "epoch": 0.55744,
+      "grad_norm": 0.2602246110794164,
+      "learning_rate": 2.238327711297997e-05,
+      "loss": 0.247,
+      "step": 871
+    },
+    {
+      "epoch": 0.55808,
+      "grad_norm": 0.2410395840841737,
+      "learning_rate": 2.2330613835142832e-05,
+      "loss": 0.2403,
+      "step": 872
+    },
+    {
+      "epoch": 0.55872,
+      "grad_norm": 0.25882744184520884,
+      "learning_rate": 2.2277962536529712e-05,
+      "loss": 0.2146,
+      "step": 873
+    },
+    {
+      "epoch": 0.55936,
+      "grad_norm": 0.24449717518726702,
+      "learning_rate": 2.2225323453420265e-05,
+      "loss": 0.2213,
+      "step": 874
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.25742794024642224,
+      "learning_rate": 2.2172696822039374e-05,
+      "loss": 0.2399,
+      "step": 875
+    },
+    {
+      "epoch": 0.56064,
+      "grad_norm": 0.2291359936845207,
+      "learning_rate": 2.2120082878556e-05,
+      "loss": 0.2196,
+      "step": 876
+    },
+    {
+      "epoch": 0.56128,
+      "grad_norm": 0.26545394800922195,
+      "learning_rate": 2.206748185908219e-05,
+      "loss": 0.2178,
+      "step": 877
+    },
+    {
+      "epoch": 0.56192,
+      "grad_norm": 0.2561710517840164,
+      "learning_rate": 2.2014893999671976e-05,
+      "loss": 0.2535,
+      "step": 878
+    },
+    {
+      "epoch": 0.56256,
+      "grad_norm": 0.27914607898196286,
+      "learning_rate": 2.1962319536320354e-05,
+      "loss": 0.2451,
+      "step": 879
+    },
+    {
+      "epoch": 0.5632,
+      "grad_norm": 0.25972131311467844,
+      "learning_rate": 2.1909758704962198e-05,
+      "loss": 0.2157,
+      "step": 880
+    },
+    {
+      "epoch": 0.56384,
+      "grad_norm": 0.237962980492853,
+      "learning_rate": 2.185721174147119e-05,
+      "loss": 0.2282,
+      "step": 881
+    },
+    {
+      "epoch": 0.56448,
+      "grad_norm": 0.25762548209762043,
+      "learning_rate": 2.1804678881658798e-05,
+      "loss": 0.2301,
+      "step": 882
+    },
+    {
+      "epoch": 0.56512,
+      "grad_norm": 0.2699542335600971,
+      "learning_rate": 2.1752160361273174e-05,
+      "loss": 0.2301,
+      "step": 883
+    },
+    {
+      "epoch": 0.56576,
+      "grad_norm": 0.2604286630504809,
+      "learning_rate": 2.1699656415998153e-05,
+      "loss": 0.2267,
+      "step": 884
+    },
+    {
+      "epoch": 0.5664,
+      "grad_norm": 0.29153377537799685,
+      "learning_rate": 2.164716728145213e-05,
+      "loss": 0.2568,
+      "step": 885
+    },
+    {
+      "epoch": 0.56704,
+      "grad_norm": 0.24065800983461674,
+      "learning_rate": 2.1594693193187062e-05,
+      "loss": 0.2349,
+      "step": 886
+    },
+    {
+      "epoch": 0.56768,
+      "grad_norm": 0.23988585134839563,
+      "learning_rate": 2.1542234386687354e-05,
+      "loss": 0.2437,
+      "step": 887
+    },
+    {
+      "epoch": 0.56832,
+      "grad_norm": 0.2470409007453363,
+      "learning_rate": 2.148979109736887e-05,
+      "loss": 0.2554,
+      "step": 888
+    },
+    {
+      "epoch": 0.56896,
+      "grad_norm": 0.24742552820439054,
+      "learning_rate": 2.1437363560577803e-05,
+      "loss": 0.2387,
+      "step": 889
+    },
+    {
+      "epoch": 0.5696,
+      "grad_norm": 0.233730304427317,
+      "learning_rate": 2.1384952011589675e-05,
+      "loss": 0.2526,
+      "step": 890
+    },
+    {
+      "epoch": 0.57024,
+      "grad_norm": 0.23016357844616137,
+      "learning_rate": 2.1332556685608243e-05,
+      "loss": 0.2519,
+      "step": 891
+    },
+    {
+      "epoch": 0.57088,
+      "grad_norm": 0.22523388261465116,
+      "learning_rate": 2.128017781776448e-05,
+      "loss": 0.2307,
+      "step": 892
+    },
+    {
+      "epoch": 0.57152,
+      "grad_norm": 0.2310630062616413,
+      "learning_rate": 2.122781564311551e-05,
+      "loss": 0.2287,
+      "step": 893
+    },
+    {
+      "epoch": 0.57216,
+      "grad_norm": 0.24955885974829986,
+      "learning_rate": 2.1175470396643505e-05,
+      "loss": 0.2354,
+      "step": 894
+    },
+    {
+      "epoch": 0.5728,
+      "grad_norm": 0.3047535916708344,
+      "learning_rate": 2.1123142313254708e-05,
+      "loss": 0.2398,
+      "step": 895
+    },
+    {
+      "epoch": 0.57344,
+      "grad_norm": 0.27225561708977664,
+      "learning_rate": 2.1070831627778303e-05,
+      "loss": 0.2516,
+      "step": 896
+    },
+    {
+      "epoch": 0.57408,
+      "grad_norm": 0.26463285324894,
+      "learning_rate": 2.101853857496545e-05,
+      "loss": 0.2421,
+      "step": 897
+    },
+    {
+      "epoch": 0.57472,
+      "grad_norm": 0.2475498008102138,
+      "learning_rate": 2.0966263389488123e-05,
+      "loss": 0.2337,
+      "step": 898
+    },
+    {
+      "epoch": 0.57536,
+      "grad_norm": 0.2717603361886218,
+      "learning_rate": 2.091400630593816e-05,
+      "loss": 0.2116,
+      "step": 899
+    },
+    {
+      "epoch": 0.576,
+      "grad_norm": 0.26228840740494686,
+      "learning_rate": 2.086176755882614e-05,
+      "loss": 0.2048,
+      "step": 900
+    },
+    {
+      "epoch": 0.57664,
+      "grad_norm": 0.3007453395429832,
+      "learning_rate": 2.0809547382580357e-05,
+      "loss": 0.2495,
+      "step": 901
+    },
+    {
+      "epoch": 0.57728,
+      "grad_norm": 0.23319995677496175,
+      "learning_rate": 2.075734601154577e-05,
+      "loss": 0.2295,
+      "step": 902
+    },
+    {
+      "epoch": 0.57792,
+      "grad_norm": 0.25256488605645533,
+      "learning_rate": 2.070516367998294e-05,
+      "loss": 0.2208,
+      "step": 903
+    },
+    {
+      "epoch": 0.57856,
+      "grad_norm": 0.2802794061326995,
+      "learning_rate": 2.065300062206701e-05,
+      "loss": 0.2399,
+      "step": 904
+    },
+    {
+      "epoch": 0.5792,
+      "grad_norm": 0.24590290065315673,
+      "learning_rate": 2.06008570718866e-05,
+      "loss": 0.2261,
+      "step": 905
+    },
+    {
+      "epoch": 0.57984,
+      "grad_norm": 0.2497522545133733,
+      "learning_rate": 2.0548733263442804e-05,
+      "loss": 0.2325,
+      "step": 906
+    },
+    {
+      "epoch": 0.58048,
+      "grad_norm": 0.2438963405514945,
+      "learning_rate": 2.04966294306481e-05,
+      "loss": 0.2421,
+      "step": 907
+    },
+    {
+      "epoch": 0.58112,
+      "grad_norm": 0.23142789378701548,
+      "learning_rate": 2.0444545807325364e-05,
+      "loss": 0.2382,
+      "step": 908
+    },
+    {
+      "epoch": 0.58176,
+      "grad_norm": 0.2635171192503007,
+      "learning_rate": 2.0392482627206734e-05,
+      "loss": 0.2465,
+      "step": 909
+    },
+    {
+      "epoch": 0.5824,
+      "grad_norm": 0.24061216761382034,
+      "learning_rate": 2.034044012393265e-05,
+      "loss": 0.2276,
+      "step": 910
+    },
+    {
+      "epoch": 0.58304,
+      "grad_norm": 0.23328312340267662,
+      "learning_rate": 2.028841853105071e-05,
+      "loss": 0.2173,
+      "step": 911
+    },
+    {
+      "epoch": 0.58368,
+      "grad_norm": 0.2368340947684322,
+      "learning_rate": 2.0236418082014728e-05,
+      "loss": 0.2322,
+      "step": 912
+    },
+    {
+      "epoch": 0.58432,
+      "grad_norm": 0.2811697336050109,
+      "learning_rate": 2.01844390101836e-05,
+      "loss": 0.2439,
+      "step": 913
+    },
+    {
+      "epoch": 0.58496,
+      "grad_norm": 0.23710863921249056,
+      "learning_rate": 2.013248154882029e-05,
+      "loss": 0.2375,
+      "step": 914
+    },
+    {
+      "epoch": 0.5856,
+      "grad_norm": 0.25772754012713756,
+      "learning_rate": 2.0080545931090784e-05,
+      "loss": 0.2511,
+      "step": 915
+    },
+    {
+      "epoch": 0.58624,
+      "grad_norm": 0.34812928934454995,
+      "learning_rate": 2.0028632390063057e-05,
+      "loss": 0.2567,
+      "step": 916
+    },
+    {
+      "epoch": 0.58688,
+      "grad_norm": 0.23874891742804483,
+      "learning_rate": 1.9976741158706002e-05,
+      "loss": 0.2234,
+      "step": 917
+    },
+    {
+      "epoch": 0.58752,
+      "grad_norm": 0.2570008574553706,
+      "learning_rate": 1.992487246988839e-05,
+      "loss": 0.2425,
+      "step": 918
+    },
+    {
+      "epoch": 0.58816,
+      "grad_norm": 0.2592824155426394,
+      "learning_rate": 1.9873026556377835e-05,
+      "loss": 0.2547,
+      "step": 919
+    },
+    {
+      "epoch": 0.5888,
+      "grad_norm": 0.25791906553166993,
+      "learning_rate": 1.9821203650839728e-05,
+      "loss": 0.2507,
+      "step": 920
+    },
+    {
+      "epoch": 0.58944,
+      "grad_norm": 0.2519619902132567,
+      "learning_rate": 1.9769403985836242e-05,
+      "loss": 0.22,
+      "step": 921
+    },
+    {
+      "epoch": 0.59008,
+      "grad_norm": 0.2158346037440381,
+      "learning_rate": 1.971762779382522e-05,
+      "loss": 0.2437,
+      "step": 922
+    },
+    {
+      "epoch": 0.59072,
+      "grad_norm": 0.2352764274570805,
+      "learning_rate": 1.9665875307159194e-05,
+      "loss": 0.238,
+      "step": 923
+    },
+    {
+      "epoch": 0.59136,
+      "grad_norm": 0.272979967431664,
+      "learning_rate": 1.961414675808429e-05,
+      "loss": 0.2165,
+      "step": 924
+    },
+    {
+      "epoch": 0.592,
+      "grad_norm": 0.2519012978032965,
+      "learning_rate": 1.956244237873924e-05,
+      "loss": 0.2332,
+      "step": 925
+    },
+    {
+      "epoch": 0.59264,
+      "grad_norm": 0.2415099987459759,
+      "learning_rate": 1.9510762401154274e-05,
+      "loss": 0.2264,
+      "step": 926
+    },
+    {
+      "epoch": 0.59328,
+      "grad_norm": 0.262755443221329,
+      "learning_rate": 1.945910705725016e-05,
+      "loss": 0.2092,
+      "step": 927
+    },
+    {
+      "epoch": 0.59392,
+      "grad_norm": 0.23263565799380795,
+      "learning_rate": 1.940747657883708e-05,
+      "loss": 0.209,
+      "step": 928
+    },
+    {
+      "epoch": 0.59456,
+      "grad_norm": 0.23006448167906013,
+      "learning_rate": 1.9355871197613647e-05,
+      "loss": 0.2264,
+      "step": 929
+    },
+    {
+      "epoch": 0.5952,
+      "grad_norm": 0.24808480980233624,
+      "learning_rate": 1.9304291145165858e-05,
+      "loss": 0.2418,
+      "step": 930
+    },
+    {
+      "epoch": 0.59584,
+      "grad_norm": 0.24167255232909088,
+      "learning_rate": 1.9252736652966015e-05,
+      "loss": 0.2095,
+      "step": 931
+    },
+    {
+      "epoch": 0.59648,
+      "grad_norm": 0.36749034216763266,
+      "learning_rate": 1.920120795237174e-05,
+      "loss": 0.2443,
+      "step": 932
+    },
+    {
+      "epoch": 0.59712,
+      "grad_norm": 0.25896873770943857,
+      "learning_rate": 1.914970527462489e-05,
+      "loss": 0.219,
+      "step": 933
+    },
+    {
+      "epoch": 0.59776,
+      "grad_norm": 0.23106699553522572,
+      "learning_rate": 1.909822885085058e-05,
+      "loss": 0.2395,
+      "step": 934
+    },
+    {
+      "epoch": 0.5984,
+      "grad_norm": 0.24559530418670644,
+      "learning_rate": 1.9046778912056047e-05,
+      "loss": 0.2164,
+      "step": 935
+    },
+    {
+      "epoch": 0.59904,
+      "grad_norm": 0.25977484660332073,
+      "learning_rate": 1.8995355689129734e-05,
+      "loss": 0.2401,
+      "step": 936
+    },
+    {
+      "epoch": 0.59968,
+      "grad_norm": 0.2427018895740078,
+      "learning_rate": 1.894395941284015e-05,
+      "loss": 0.2404,
+      "step": 937
+    },
+    {
+      "epoch": 0.60032,
+      "grad_norm": 0.249114093571263,
+      "learning_rate": 1.88925903138349e-05,
+      "loss": 0.2272,
+      "step": 938
+    },
+    {
+      "epoch": 0.60096,
+      "grad_norm": 0.25487846095563155,
+      "learning_rate": 1.884124862263961e-05,
+      "loss": 0.2444,
+      "step": 939
+    },
+    {
+      "epoch": 0.6016,
+      "grad_norm": 0.2772681154769936,
+      "learning_rate": 1.8789934569656932e-05,
+      "loss": 0.236,
+      "step": 940
+    },
+    {
+      "epoch": 0.60224,
+      "grad_norm": 0.26276401305661673,
+      "learning_rate": 1.8738648385165476e-05,
+      "loss": 0.2509,
+      "step": 941
+    },
+    {
+      "epoch": 0.60288,
+      "grad_norm": 0.2421643902947639,
+      "learning_rate": 1.868739029931878e-05,
+      "loss": 0.2265,
+      "step": 942
+    },
+    {
+      "epoch": 0.60352,
+      "grad_norm": 0.25209773308876127,
+      "learning_rate": 1.8636160542144305e-05,
+      "loss": 0.2462,
+      "step": 943
+    },
+    {
+      "epoch": 0.60416,
+      "grad_norm": 0.2504374986113504,
+      "learning_rate": 1.858495934354235e-05,
+      "loss": 0.2516,
+      "step": 944
+    },
+    {
+      "epoch": 0.6048,
+      "grad_norm": 0.2371903842169047,
+      "learning_rate": 1.8533786933285106e-05,
+      "loss": 0.2231,
+      "step": 945
+    },
+    {
+      "epoch": 0.60544,
+      "grad_norm": 0.2764862594272394,
+      "learning_rate": 1.8482643541015514e-05,
+      "loss": 0.2383,
+      "step": 946
+    },
+    {
+      "epoch": 0.60608,
+      "grad_norm": 0.2575549133224725,
+      "learning_rate": 1.843152939624635e-05,
+      "loss": 0.2323,
+      "step": 947
+    },
+    {
+      "epoch": 0.60672,
+      "grad_norm": 0.23295470309229857,
+      "learning_rate": 1.8380444728359086e-05,
+      "loss": 0.2275,
+      "step": 948
+    },
+    {
+      "epoch": 0.60736,
+      "grad_norm": 0.25271507492186174,
+      "learning_rate": 1.832938976660295e-05,
+      "loss": 0.2264,
+      "step": 949
+    },
+    {
+      "epoch": 0.608,
+      "grad_norm": 0.2652561072464238,
+      "learning_rate": 1.827836474009383e-05,
+      "loss": 0.2239,
+      "step": 950
+    },
+    {
+      "epoch": 0.60864,
+      "grad_norm": 0.22608022892434435,
+      "learning_rate": 1.8227369877813314e-05,
+      "loss": 0.2377,
+      "step": 951
+    },
+    {
+      "epoch": 0.60928,
+      "grad_norm": 0.24446247123420378,
+      "learning_rate": 1.8176405408607574e-05,
+      "loss": 0.2269,
+      "step": 952
+    },
+    {
+      "epoch": 0.60992,
+      "grad_norm": 0.2490136365512997,
+      "learning_rate": 1.812547156118644e-05,
+      "loss": 0.2399,
+      "step": 953
+    },
+    {
+      "epoch": 0.61056,
+      "grad_norm": 0.250733094731068,
+      "learning_rate": 1.8074568564122293e-05,
+      "loss": 0.2285,
+      "step": 954
+    },
+    {
+      "epoch": 0.6112,
+      "grad_norm": 0.2377411755095795,
+      "learning_rate": 1.8023696645849064e-05,
+      "loss": 0.2076,
+      "step": 955
+    },
+    {
+      "epoch": 0.61184,
+      "grad_norm": 0.2443525802339012,
+      "learning_rate": 1.7972856034661244e-05,
+      "loss": 0.2373,
+      "step": 956
+    },
+    {
+      "epoch": 0.61248,
+      "grad_norm": 0.24506635992781736,
+      "learning_rate": 1.7922046958712786e-05,
+      "loss": 0.2306,
+      "step": 957
+    },
+    {
+      "epoch": 0.61312,
+      "grad_norm": 0.22738852829598277,
+      "learning_rate": 1.7871269646016177e-05,
+      "loss": 0.2336,
+      "step": 958
+    },
+    {
+      "epoch": 0.61376,
+      "grad_norm": 0.2544653921193952,
+      "learning_rate": 1.782052432444131e-05,
+      "loss": 0.2302,
+      "step": 959
+    },
+    {
+      "epoch": 0.6144,
+      "grad_norm": 0.2555663332082816,
+      "learning_rate": 1.7769811221714544e-05,
+      "loss": 0.2196,
+      "step": 960
+    },
+    {
+      "epoch": 0.61504,
+      "grad_norm": 0.22266677466207177,
+      "learning_rate": 1.7719130565417653e-05,
+      "loss": 0.2475,
+      "step": 961
+    },
+    {
+      "epoch": 0.61568,
+      "grad_norm": 0.2389157150564247,
+      "learning_rate": 1.7668482582986793e-05,
+      "loss": 0.2215,
+      "step": 962
+    },
+    {
+      "epoch": 0.61632,
+      "grad_norm": 0.2271967944606995,
+      "learning_rate": 1.7617867501711478e-05,
+      "loss": 0.2128,
+      "step": 963
+    },
+    {
+      "epoch": 0.61696,
+      "grad_norm": 0.23211617956670375,
+      "learning_rate": 1.7567285548733614e-05,
+      "loss": 0.2372,
+      "step": 964
+    },
+    {
+      "epoch": 0.6176,
+      "grad_norm": 0.2169848262071649,
+      "learning_rate": 1.7516736951046394e-05,
+      "loss": 0.2126,
+      "step": 965
+    },
+    {
+      "epoch": 0.61824,
+      "grad_norm": 0.249168880968389,
+      "learning_rate": 1.746622193549336e-05,
+      "loss": 0.2349,
+      "step": 966
+    },
+    {
+      "epoch": 0.61888,
+      "grad_norm": 0.23837849411260273,
+      "learning_rate": 1.7415740728767337e-05,
+      "loss": 0.2188,
+      "step": 967
+    },
+    {
+      "epoch": 0.61952,
+      "grad_norm": 0.24563388739452638,
+      "learning_rate": 1.7365293557409413e-05,
+      "loss": 0.2459,
+      "step": 968
+    },
+    {
+      "epoch": 0.62016,
+      "grad_norm": 0.21499816162977467,
+      "learning_rate": 1.7314880647807967e-05,
+      "loss": 0.2268,
+      "step": 969
+    },
+    {
+      "epoch": 0.6208,
+      "grad_norm": 0.2579882292507303,
+      "learning_rate": 1.7264502226197593e-05,
+      "loss": 0.2137,
+      "step": 970
+    },
+    {
+      "epoch": 0.62144,
+      "grad_norm": 0.24874653864057025,
+      "learning_rate": 1.7214158518658152e-05,
+      "loss": 0.2168,
+      "step": 971
+    },
+    {
+      "epoch": 0.62208,
+      "grad_norm": 0.24597202210241592,
+      "learning_rate": 1.716384975111368e-05,
+      "loss": 0.2248,
+      "step": 972
+    },
+    {
+      "epoch": 0.62272,
+      "grad_norm": 0.272136761134001,
+      "learning_rate": 1.711357614933145e-05,
+      "loss": 0.2013,
+      "step": 973
+    },
+    {
+      "epoch": 0.62336,
+      "grad_norm": 0.24891181430030945,
+      "learning_rate": 1.7063337938920895e-05,
+      "loss": 0.2268,
+      "step": 974
+    },
+    {
+      "epoch": 0.624,
+      "grad_norm": 0.2737551998037743,
+      "learning_rate": 1.7013135345332653e-05,
+      "loss": 0.2244,
+      "step": 975
+    },
+    {
+      "epoch": 0.62464,
+      "grad_norm": 0.24070930758029294,
+      "learning_rate": 1.6962968593857502e-05,
+      "loss": 0.234,
+      "step": 976
+    },
+    {
+      "epoch": 0.62528,
+      "grad_norm": 0.2502924992088562,
+      "learning_rate": 1.6912837909625395e-05,
+      "loss": 0.2521,
+      "step": 977
+    },
+    {
+      "epoch": 0.62592,
+      "grad_norm": 0.23633257011183453,
+      "learning_rate": 1.6862743517604407e-05,
+      "loss": 0.2463,
+      "step": 978
+    },
+    {
+      "epoch": 0.62656,
+      "grad_norm": 0.2625709662809829,
+      "learning_rate": 1.6812685642599758e-05,
+      "loss": 0.2361,
+      "step": 979
+    },
+    {
+      "epoch": 0.6272,
+      "grad_norm": 0.2460886586797407,
+      "learning_rate": 1.6762664509252806e-05,
+      "loss": 0.2443,
+      "step": 980
+    },
+    {
+      "epoch": 0.62784,
+      "grad_norm": 0.2326551082107272,
+      "learning_rate": 1.6712680342039994e-05,
+      "loss": 0.2123,
+      "step": 981
+    },
+    {
+      "epoch": 0.62848,
+      "grad_norm": 0.3289541327924151,
+      "learning_rate": 1.6662733365271916e-05,
+      "loss": 0.2326,
+      "step": 982
+    },
+    {
+      "epoch": 0.62912,
+      "grad_norm": 0.2404629024405418,
+      "learning_rate": 1.661282380309223e-05,
+      "loss": 0.2402,
+      "step": 983
+    },
+    {
+      "epoch": 0.62976,
+      "grad_norm": 0.23238643115966054,
+      "learning_rate": 1.6562951879476708e-05,
+      "loss": 0.2499,
+      "step": 984
+    },
+    {
+      "epoch": 0.6304,
+      "grad_norm": 0.2487557824774618,
+      "learning_rate": 1.6513117818232217e-05,
+      "loss": 0.2213,
+      "step": 985
+    },
+    {
+      "epoch": 0.63104,
+      "grad_norm": 0.2720891226041973,
+      "learning_rate": 1.6463321842995717e-05,
+      "loss": 0.2056,
+      "step": 986
+    },
+    {
+      "epoch": 0.63168,
+      "grad_norm": 0.2419524975201308,
+      "learning_rate": 1.641356417723322e-05,
+      "loss": 0.2194,
+      "step": 987
+    },
+    {
+      "epoch": 0.63232,
+      "grad_norm": 0.2456654914470876,
+      "learning_rate": 1.6363845044238868e-05,
+      "loss": 0.2299,
+      "step": 988
+    },
+    {
+      "epoch": 0.63296,
+      "grad_norm": 0.24606571662060764,
+      "learning_rate": 1.631416466713383e-05,
+      "loss": 0.2343,
+      "step": 989
+    },
+    {
+      "epoch": 0.6336,
+      "grad_norm": 0.25976929437351914,
+      "learning_rate": 1.62645232688654e-05,
+      "loss": 0.2124,
+      "step": 990
+    },
+    {
+      "epoch": 0.63424,
+      "grad_norm": 0.22966378183764538,
+      "learning_rate": 1.621492107220593e-05,
+      "loss": 0.2249,
+      "step": 991
+    },
+    {
+      "epoch": 0.63488,
+      "grad_norm": 0.24560027003327328,
+      "learning_rate": 1.6165358299751825e-05,
+      "loss": 0.2421,
+      "step": 992
+    },
+    {
+      "epoch": 0.63552,
+      "grad_norm": 0.2615147248972779,
+      "learning_rate": 1.6115835173922618e-05,
+      "loss": 0.2286,
+      "step": 993
+    },
+    {
+      "epoch": 0.63616,
+      "grad_norm": 0.2498872142446554,
+      "learning_rate": 1.6066351916959876e-05,
+      "loss": 0.2385,
+      "step": 994
+    },
+    {
+      "epoch": 0.6368,
+      "grad_norm": 0.24668966466931394,
+      "learning_rate": 1.6016908750926286e-05,
+      "loss": 0.2279,
+      "step": 995
+    },
+    {
+      "epoch": 0.63744,
+      "grad_norm": 0.2448153409198341,
+      "learning_rate": 1.596750589770459e-05,
+      "loss": 0.251,
+      "step": 996
+    },
+    {
+      "epoch": 0.63808,
+      "grad_norm": 0.2255213910797428,
+      "learning_rate": 1.5918143578996646e-05,
+      "loss": 0.229,
+      "step": 997
+    },
+    {
+      "epoch": 0.63872,
+      "grad_norm": 0.26670669610797537,
+      "learning_rate": 1.5868822016322386e-05,
+      "loss": 0.2292,
+      "step": 998
+    },
+    {
+      "epoch": 0.63936,
+      "grad_norm": 0.27422960837817995,
+      "learning_rate": 1.5819541431018874e-05,
+      "loss": 0.2467,
+      "step": 999
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.2594796851816953,
+      "learning_rate": 1.577030204423924e-05,
+      "loss": 0.2321,
+      "step": 1000
+    },
+    {
+      "epoch": 0.64064,
+      "grad_norm": 0.251125173314332,
+      "learning_rate": 1.5721104076951783e-05,
+      "loss": 0.2188,
+      "step": 1001
+    },
+    {
+      "epoch": 0.64128,
+      "grad_norm": 0.27041270020727115,
+      "learning_rate": 1.5671947749938875e-05,
+      "loss": 0.2412,
+      "step": 1002
+    },
+    {
+      "epoch": 0.64192,
+      "grad_norm": 0.2587525546327185,
+      "learning_rate": 1.562283328379605e-05,
+      "loss": 0.2344,
+      "step": 1003
+    },
+    {
+      "epoch": 0.64256,
+      "grad_norm": 0.2558126129444334,
+      "learning_rate": 1.5573760898931e-05,
+      "loss": 0.2175,
+      "step": 1004
+    },
+    {
+      "epoch": 0.6432,
+      "grad_norm": 0.25305386195310686,
+      "learning_rate": 1.5524730815562518e-05,
+      "loss": 0.2312,
+      "step": 1005
+    },
+    {
+      "epoch": 0.64384,
+      "grad_norm": 0.25270340970103905,
+      "learning_rate": 1.5475743253719628e-05,
+      "loss": 0.217,
+      "step": 1006
+    },
+    {
+      "epoch": 0.64448,
+      "grad_norm": 0.2605550420385158,
+      "learning_rate": 1.542679843324049e-05,
+      "loss": 0.2342,
+      "step": 1007
+    },
+    {
+      "epoch": 0.64512,
+      "grad_norm": 0.23968724291537247,
+      "learning_rate": 1.5377896573771465e-05,
+      "loss": 0.2401,
+      "step": 1008
+    },
+    {
+      "epoch": 0.64576,
+      "grad_norm": 0.2454287940522169,
+      "learning_rate": 1.5329037894766133e-05,
+      "loss": 0.2502,
+      "step": 1009
+    },
+    {
+      "epoch": 0.6464,
+      "grad_norm": 0.30833838890513654,
+      "learning_rate": 1.528022261548429e-05,
+      "loss": 0.2358,
+      "step": 1010
+    },
+    {
+      "epoch": 0.64704,
+      "grad_norm": 0.23544190960239791,
+      "learning_rate": 1.5231450954990955e-05,
+      "loss": 0.2428,
+      "step": 1011
+    },
+    {
+      "epoch": 0.64768,
+      "grad_norm": 0.2444795119322928,
+      "learning_rate": 1.518272313215543e-05,
+      "loss": 0.2325,
+      "step": 1012
+    },
+    {
+      "epoch": 0.64832,
+      "grad_norm": 0.24444722344736422,
+      "learning_rate": 1.513403936565026e-05,
+      "loss": 0.2226,
+      "step": 1013
+    },
+    {
+      "epoch": 0.64896,
+      "grad_norm": 0.21722105721574908,
+      "learning_rate": 1.5085399873950313e-05,
+      "loss": 0.2129,
+      "step": 1014
+    },
+    {
+      "epoch": 0.6496,
+      "grad_norm": 0.2662609811233179,
+      "learning_rate": 1.5036804875331734e-05,
+      "loss": 0.2668,
+      "step": 1015
+    },
+    {
+      "epoch": 0.65024,
+      "grad_norm": 0.21367906357204217,
+      "learning_rate": 1.498825458787102e-05,
+      "loss": 0.2363,
+      "step": 1016
+    },
+    {
+      "epoch": 0.65088,
+      "grad_norm": 0.24687360474551862,
+      "learning_rate": 1.4939749229444028e-05,
+      "loss": 0.2277,
+      "step": 1017
+    },
+    {
+      "epoch": 0.65152,
+      "grad_norm": 0.32162107665269024,
+      "learning_rate": 1.4891289017724967e-05,
+      "loss": 0.2261,
+      "step": 1018
+    },
+    {
+      "epoch": 0.65216,
+      "grad_norm": 0.2227680806268295,
+      "learning_rate": 1.4842874170185467e-05,
+      "loss": 0.2367,
+      "step": 1019
+    },
+    {
+      "epoch": 0.6528,
+      "grad_norm": 0.23176735014984065,
+      "learning_rate": 1.4794504904093559e-05,
+      "loss": 0.221,
+      "step": 1020
+    },
+    {
+      "epoch": 0.65344,
+      "grad_norm": 0.240409278759745,
+      "learning_rate": 1.4746181436512747e-05,
+      "loss": 0.25,
+      "step": 1021
+    },
+    {
+      "epoch": 0.65408,
+      "grad_norm": 0.29139858507432276,
+      "learning_rate": 1.4697903984300964e-05,
+      "loss": 0.2087,
+      "step": 1022
+    },
+    {
+      "epoch": 0.65472,
+      "grad_norm": 0.2486310034892573,
+      "learning_rate": 1.4649672764109704e-05,
+      "loss": 0.2346,
+      "step": 1023
+    },
+    {
+      "epoch": 0.65536,
+      "grad_norm": 0.2582009940090924,
+      "learning_rate": 1.460148799238293e-05,
+      "loss": 0.2273,
+      "step": 1024
+    },
+    {
+      "epoch": 0.656,
+      "grad_norm": 0.27507557105821945,
+      "learning_rate": 1.4553349885356213e-05,
+      "loss": 0.2355,
+      "step": 1025
+    },
+    {
+      "epoch": 0.65664,
+      "grad_norm": 0.2272595636884127,
+      "learning_rate": 1.4505258659055665e-05,
+      "loss": 0.2122,
+      "step": 1026
+    },
+    {
+      "epoch": 0.65728,
+      "grad_norm": 0.2685299474659218,
+      "learning_rate": 1.4457214529297043e-05,
+      "loss": 0.2524,
+      "step": 1027
+    },
+    {
+      "epoch": 0.65792,
+      "grad_norm": 0.24478045620656394,
+      "learning_rate": 1.4409217711684737e-05,
+      "loss": 0.2271,
+      "step": 1028
+    },
+    {
+      "epoch": 0.65856,
+      "grad_norm": 0.2438608867949604,
+      "learning_rate": 1.4361268421610829e-05,
+      "loss": 0.2082,
+      "step": 1029
+    },
+    {
+      "epoch": 0.6592,
+      "grad_norm": 0.23950650796557532,
+      "learning_rate": 1.431336687425411e-05,
+      "loss": 0.242,
+      "step": 1030
+    },
+    {
+      "epoch": 0.65984,
+      "grad_norm": 0.21923187858425636,
+      "learning_rate": 1.4265513284579118e-05,
+      "loss": 0.2202,
+      "step": 1031
+    },
+    {
+      "epoch": 0.66048,
+      "grad_norm": 0.23108649237258344,
+      "learning_rate": 1.4217707867335179e-05,
+      "loss": 0.2138,
+      "step": 1032
+    },
+    {
+      "epoch": 0.66112,
+      "grad_norm": 0.22514082591088475,
+      "learning_rate": 1.4169950837055424e-05,
+      "loss": 0.2337,
+      "step": 1033
+    },
+    {
+      "epoch": 0.66176,
+      "grad_norm": 0.24557807845051935,
+      "learning_rate": 1.4122242408055864e-05,
+      "loss": 0.2303,
+      "step": 1034
+    },
+    {
+      "epoch": 0.6624,
+      "grad_norm": 0.26955216666831283,
+      "learning_rate": 1.4074582794434388e-05,
+      "loss": 0.2401,
+      "step": 1035
+    },
+    {
+      "epoch": 0.66304,
+      "grad_norm": 0.31465882803473016,
+      "learning_rate": 1.4026972210069839e-05,
+      "loss": 0.2223,
+      "step": 1036
+    },
+    {
+      "epoch": 0.66368,
+      "grad_norm": 0.23017556870547434,
+      "learning_rate": 1.3979410868620987e-05,
+      "loss": 0.2334,
+      "step": 1037
+    },
+    {
+      "epoch": 0.66432,
+      "grad_norm": 0.2304501081180642,
+      "learning_rate": 1.3931898983525694e-05,
+      "loss": 0.2458,
+      "step": 1038
+    },
+    {
+      "epoch": 0.66496,
+      "grad_norm": 0.23734751196071166,
+      "learning_rate": 1.3884436767999798e-05,
+      "loss": 0.2442,
+      "step": 1039
+    },
+    {
+      "epoch": 0.6656,
+      "grad_norm": 0.21870149517594278,
+      "learning_rate": 1.3837024435036309e-05,
+      "loss": 0.2102,
+      "step": 1040
+    },
+    {
+      "epoch": 0.66624,
+      "grad_norm": 0.2573044798183855,
+      "learning_rate": 1.3789662197404346e-05,
+      "loss": 0.2147,
+      "step": 1041
+    },
+    {
+      "epoch": 0.66688,
+      "grad_norm": 0.25790649176441605,
+      "learning_rate": 1.3742350267648202e-05,
+      "loss": 0.2295,
+      "step": 1042
+    },
+    {
+      "epoch": 0.66752,
+      "grad_norm": 0.22127521722926796,
+      "learning_rate": 1.3695088858086464e-05,
+      "loss": 0.2356,
+      "step": 1043
+    },
+    {
+      "epoch": 0.66816,
+      "grad_norm": 0.22325505943109972,
+      "learning_rate": 1.3647878180810924e-05,
+      "loss": 0.2164,
+      "step": 1044
+    },
+    {
+      "epoch": 0.6688,
+      "grad_norm": 0.2409802391363893,
+      "learning_rate": 1.36007184476858e-05,
+      "loss": 0.2413,
+      "step": 1045
+    },
+    {
+      "epoch": 0.66944,
+      "grad_norm": 0.25444344445905537,
+      "learning_rate": 1.3553609870346595e-05,
+      "loss": 0.2425,
+      "step": 1046
+    },
+    {
+      "epoch": 0.67008,
+      "grad_norm": 0.24868149873617512,
+      "learning_rate": 1.3506552660199328e-05,
+      "loss": 0.2215,
+      "step": 1047
+    },
+    {
+      "epoch": 0.67072,
+      "grad_norm": 0.23414858316139175,
+      "learning_rate": 1.3459547028419438e-05,
+      "loss": 0.2214,
+      "step": 1048
+    },
+    {
+      "epoch": 0.67136,
+      "grad_norm": 0.22922138158714406,
+      "learning_rate": 1.3412593185950934e-05,
+      "loss": 0.2102,
+      "step": 1049
+    },
+    {
+      "epoch": 0.672,
+      "grad_norm": 0.2255749212351229,
+      "learning_rate": 1.33656913435054e-05,
+      "loss": 0.247,
+      "step": 1050
+    },
+    {
+      "epoch": 0.67264,
+      "grad_norm": 0.2415131133238682,
+      "learning_rate": 1.3318841711561067e-05,
+      "loss": 0.2356,
+      "step": 1051
+    },
+    {
+      "epoch": 0.67328,
+      "grad_norm": 0.23142719398025727,
+      "learning_rate": 1.3272044500361865e-05,
+      "loss": 0.2153,
+      "step": 1052
+    },
+    {
+      "epoch": 0.67392,
+      "grad_norm": 0.25073453268026724,
+      "learning_rate": 1.3225299919916479e-05,
+      "loss": 0.2452,
+      "step": 1053
+    },
+    {
+      "epoch": 0.67456,
+      "grad_norm": 0.25629355291923234,
+      "learning_rate": 1.3178608179997406e-05,
+      "loss": 0.2345,
+      "step": 1054
+    },
+    {
+      "epoch": 0.6752,
+      "grad_norm": 0.24676785243199625,
+      "learning_rate": 1.3131969490140012e-05,
+      "loss": 0.2146,
+      "step": 1055
+    },
+    {
+      "epoch": 0.67584,
+      "grad_norm": 0.23912531318178484,
+      "learning_rate": 1.3085384059641597e-05,
+      "loss": 0.2236,
+      "step": 1056
+    },
+    {
+      "epoch": 0.67648,
+      "grad_norm": 0.2529286224469849,
+      "learning_rate": 1.303885209756045e-05,
+      "loss": 0.2249,
+      "step": 1057
+    },
+    {
+      "epoch": 0.67712,
+      "grad_norm": 0.23133087336687283,
+      "learning_rate": 1.2992373812714913e-05,
+      "loss": 0.2125,
+      "step": 1058
+    },
+    {
+      "epoch": 0.67776,
+      "grad_norm": 0.24904546196447905,
+      "learning_rate": 1.2945949413682451e-05,
+      "loss": 0.2275,
+      "step": 1059
+    },
+    {
+      "epoch": 0.6784,
+      "grad_norm": 0.23545034103097306,
+      "learning_rate": 1.2899579108798709e-05,
+      "loss": 0.2259,
+      "step": 1060
+    },
+    {
+      "epoch": 0.67904,
+      "grad_norm": 0.22841850179856155,
+      "learning_rate": 1.2853263106156549e-05,
+      "loss": 0.215,
+      "step": 1061
+    },
+    {
+      "epoch": 0.67968,
+      "grad_norm": 0.24647355165743123,
+      "learning_rate": 1.28070016136052e-05,
+      "loss": 0.23,
+      "step": 1062
+    },
+    {
+      "epoch": 0.68032,
+      "grad_norm": 0.2300144795779541,
+      "learning_rate": 1.2760794838749202e-05,
+      "loss": 0.2376,
+      "step": 1063
+    },
+    {
+      "epoch": 0.68096,
+      "grad_norm": 0.2360570555777092,
+      "learning_rate": 1.2714642988947617e-05,
+      "loss": 0.2084,
+      "step": 1064
+    },
+    {
+      "epoch": 0.6816,
+      "grad_norm": 0.24658002367964216,
+      "learning_rate": 1.2668546271312954e-05,
+      "loss": 0.2361,
+      "step": 1065
+    },
+    {
+      "epoch": 0.68224,
+      "grad_norm": 0.23397987240843635,
+      "learning_rate": 1.262250489271034e-05,
+      "loss": 0.2216,
+      "step": 1066
+    },
+    {
+      "epoch": 0.68288,
+      "grad_norm": 0.21849138633593948,
+      "learning_rate": 1.2576519059756591e-05,
+      "loss": 0.2313,
+      "step": 1067
+    },
+    {
+      "epoch": 0.68352,
+      "grad_norm": 0.2676662034284032,
+      "learning_rate": 1.2530588978819182e-05,
+      "loss": 0.213,
+      "step": 1068
+    },
+    {
+      "epoch": 0.68416,
+      "grad_norm": 0.2392956841913877,
+      "learning_rate": 1.2484714856015478e-05,
+      "loss": 0.2231,
+      "step": 1069
+    },
+    {
+      "epoch": 0.6848,
+      "grad_norm": 0.2299969754799877,
+      "learning_rate": 1.243889689721165e-05,
+      "loss": 0.2237,
+      "step": 1070
+    },
+    {
+      "epoch": 0.68544,
+      "grad_norm": 0.26249271558966797,
+      "learning_rate": 1.2393135308021863e-05,
+      "loss": 0.2357,
+      "step": 1071
+    },
+    {
+      "epoch": 0.68608,
+      "grad_norm": 0.26030120939009876,
+      "learning_rate": 1.2347430293807318e-05,
+      "loss": 0.2328,
+      "step": 1072
+    },
+    {
+      "epoch": 0.68672,
+      "grad_norm": 0.23928311030536506,
+      "learning_rate": 1.2301782059675315e-05,
+      "loss": 0.2043,
+      "step": 1073
+    },
+    {
+      "epoch": 0.68736,
+      "grad_norm": 0.23842176304873977,
+      "learning_rate": 1.2256190810478355e-05,
+      "loss": 0.2167,
+      "step": 1074
+    },
+    {
+      "epoch": 0.688,
+      "grad_norm": 0.2231113970544429,
+      "learning_rate": 1.2210656750813204e-05,
+      "loss": 0.2161,
+      "step": 1075
+    },
+    {
+      "epoch": 0.68864,
+      "grad_norm": 0.24152708496370323,
+      "learning_rate": 1.2165180085019989e-05,
+      "loss": 0.2134,
+      "step": 1076
+    },
+    {
+      "epoch": 0.68928,
+      "grad_norm": 0.23934610249237756,
+      "learning_rate": 1.2119761017181273e-05,
+      "loss": 0.211,
+      "step": 1077
+    },
+    {
+      "epoch": 0.68992,
+      "grad_norm": 0.22269457671319887,
+      "learning_rate": 1.2074399751121135e-05,
+      "loss": 0.2139,
+      "step": 1078
+    },
+    {
+      "epoch": 0.69056,
+      "grad_norm": 0.24127394916502604,
+      "learning_rate": 1.2029096490404265e-05,
+      "loss": 0.2143,
+      "step": 1079
+    },
+    {
+      "epoch": 0.6912,
+      "grad_norm": 0.2314027877200604,
+      "learning_rate": 1.1983851438335052e-05,
+      "loss": 0.2426,
+      "step": 1080
+    },
+    {
+      "epoch": 0.69184,
+      "grad_norm": 0.23219723459760594,
+      "learning_rate": 1.193866479795666e-05,
+      "loss": 0.2166,
+      "step": 1081
+    },
+    {
+      "epoch": 0.69248,
+      "grad_norm": 0.23202391814070256,
+      "learning_rate": 1.1893536772050115e-05,
+      "loss": 0.2173,
+      "step": 1082
+    },
+    {
+      "epoch": 0.69312,
+      "grad_norm": 0.26099654560611124,
+      "learning_rate": 1.1848467563133417e-05,
+      "loss": 0.2085,
+      "step": 1083
+    },
+    {
+      "epoch": 0.69376,
+      "grad_norm": 0.2478437160004748,
+      "learning_rate": 1.1803457373460622e-05,
+      "loss": 0.2212,
+      "step": 1084
+    },
+    {
+      "epoch": 0.6944,
+      "grad_norm": 0.24092641967627243,
+      "learning_rate": 1.1758506405020887e-05,
+      "loss": 0.2168,
+      "step": 1085
+    },
+    {
+      "epoch": 0.69504,
+      "grad_norm": 0.22970315985315906,
+      "learning_rate": 1.1713614859537675e-05,
+      "loss": 0.228,
+      "step": 1086
+    },
+    {
+      "epoch": 0.69568,
+      "grad_norm": 0.24086157787563756,
+      "learning_rate": 1.1668782938467706e-05,
+      "loss": 0.192,
+      "step": 1087
+    },
+    {
+      "epoch": 0.69632,
+      "grad_norm": 0.2516127808779685,
+      "learning_rate": 1.1624010843000201e-05,
+      "loss": 0.204,
+      "step": 1088
+    },
+    {
+      "epoch": 0.69696,
+      "grad_norm": 0.24996411478912844,
+      "learning_rate": 1.1579298774055843e-05,
+      "loss": 0.219,
+      "step": 1089
+    },
+    {
+      "epoch": 0.6976,
+      "grad_norm": 0.24437058816496582,
+      "learning_rate": 1.1534646932285962e-05,
+      "loss": 0.24,
+      "step": 1090
+    },
+    {
+      "epoch": 0.69824,
+      "grad_norm": 0.22500894621156375,
+      "learning_rate": 1.1490055518071643e-05,
+      "loss": 0.2244,
+      "step": 1091
+    },
+    {
+      "epoch": 0.69888,
+      "grad_norm": 0.23636669242762376,
+      "learning_rate": 1.1445524731522722e-05,
+      "loss": 0.2144,
+      "step": 1092
+    },
+    {
+      "epoch": 0.69952,
+      "grad_norm": 0.23513524357862164,
+      "learning_rate": 1.1401054772477045e-05,
+      "loss": 0.2075,
+      "step": 1093
+    },
+    {
+      "epoch": 0.70016,
+      "grad_norm": 0.25368935547064475,
+      "learning_rate": 1.1356645840499413e-05,
+      "loss": 0.2351,
+      "step": 1094
+    },
+    {
+      "epoch": 0.7008,
+      "grad_norm": 0.24205150817660032,
+      "learning_rate": 1.13122981348808e-05,
+      "loss": 0.2138,
+      "step": 1095
+    },
+    {
+      "epoch": 0.70144,
+      "grad_norm": 0.2615794629550789,
+      "learning_rate": 1.1268011854637409e-05,
+      "loss": 0.2286,
+      "step": 1096
+    },
+    {
+      "epoch": 0.70208,
+      "grad_norm": 0.2593375619419983,
+      "learning_rate": 1.1223787198509791e-05,
+      "loss": 0.2303,
+      "step": 1097
+    },
+    {
+      "epoch": 0.70272,
+      "grad_norm": 0.2709928850464376,
+      "learning_rate": 1.1179624364961946e-05,
+      "loss": 0.2344,
+      "step": 1098
+    },
+    {
+      "epoch": 0.70336,
+      "grad_norm": 0.26415947364642406,
+      "learning_rate": 1.1135523552180441e-05,
+      "loss": 0.2254,
+      "step": 1099
+    },
+    {
+      "epoch": 0.704,
+      "grad_norm": 0.23497108785863402,
+      "learning_rate": 1.1091484958073513e-05,
+      "loss": 0.2011,
+      "step": 1100
+    },
+    {
+      "epoch": 0.70464,
+      "grad_norm": 0.25247549078652853,
+      "learning_rate": 1.1047508780270194e-05,
+      "loss": 0.2125,
+      "step": 1101
+    },
+    {
+      "epoch": 0.70528,
+      "grad_norm": 0.257276927516065,
+      "learning_rate": 1.1003595216119395e-05,
+      "loss": 0.2367,
+      "step": 1102
+    },
+    {
+      "epoch": 0.70592,
+      "grad_norm": 0.22939217181385815,
+      "learning_rate": 1.0959744462689062e-05,
+      "loss": 0.2283,
+      "step": 1103
+    },
+    {
+      "epoch": 0.70656,
+      "grad_norm": 0.24122307567844234,
+      "learning_rate": 1.0915956716765251e-05,
+      "loss": 0.2255,
+      "step": 1104
+    },
+    {
+      "epoch": 0.7072,
+      "grad_norm": 0.23487347492943667,
+      "learning_rate": 1.0872232174851282e-05,
+      "loss": 0.2187,
+      "step": 1105
+    },
+    {
+      "epoch": 0.70784,
+      "grad_norm": 0.21942698538490377,
+      "learning_rate": 1.0828571033166823e-05,
+      "loss": 0.2336,
+      "step": 1106
+    },
+    {
+      "epoch": 0.70848,
+      "grad_norm": 0.25445594653527487,
+      "learning_rate": 1.0784973487647029e-05,
+      "loss": 0.2255,
+      "step": 1107
+    },
+    {
+      "epoch": 0.70912,
+      "grad_norm": 0.24552884289202662,
+      "learning_rate": 1.0741439733941674e-05,
+      "loss": 0.2265,
+      "step": 1108
+    },
+    {
+      "epoch": 0.70976,
+      "grad_norm": 0.24741341654602492,
+      "learning_rate": 1.0697969967414212e-05,
+      "loss": 0.2328,
+      "step": 1109
+    },
+    {
+      "epoch": 0.7104,
+      "grad_norm": 0.21952355784938485,
+      "learning_rate": 1.0654564383141021e-05,
+      "loss": 0.2138,
+      "step": 1110
+    },
+    {
+      "epoch": 0.71104,
+      "grad_norm": 0.22351479238879535,
+      "learning_rate": 1.061122317591037e-05,
+      "loss": 0.213,
+      "step": 1111
+    },
+    {
+      "epoch": 0.71168,
+      "grad_norm": 0.25152081373546625,
+      "learning_rate": 1.0567946540221701e-05,
+      "loss": 0.2282,
+      "step": 1112
+    },
+    {
+      "epoch": 0.71232,
+      "grad_norm": 0.24109401693501464,
+      "learning_rate": 1.0524734670284631e-05,
+      "loss": 0.2324,
+      "step": 1113
+    },
+    {
+      "epoch": 0.71296,
+      "grad_norm": 0.24729300210883084,
+      "learning_rate": 1.048158776001815e-05,
+      "loss": 0.2439,
+      "step": 1114
+    },
+    {
+      "epoch": 0.7136,
+      "grad_norm": 0.23383364014683763,
+      "learning_rate": 1.0438506003049737e-05,
+      "loss": 0.2447,
+      "step": 1115
+    },
+    {
+      "epoch": 0.71424,
+      "grad_norm": 0.24856586374372389,
+      "learning_rate": 1.0395489592714472e-05,
+      "loss": 0.223,
+      "step": 1116
+    },
+    {
+      "epoch": 0.71488,
+      "grad_norm": 0.27347144582789584,
+      "learning_rate": 1.0352538722054225e-05,
+      "loss": 0.2026,
+      "step": 1117
+    },
+    {
+      "epoch": 0.71552,
+      "grad_norm": 0.234331513050963,
+      "learning_rate": 1.0309653583816688e-05,
+      "loss": 0.2267,
+      "step": 1118
+    },
+    {
+      "epoch": 0.71616,
+      "grad_norm": 0.24376631600636176,
+      "learning_rate": 1.0266834370454612e-05,
+      "loss": 0.2305,
+      "step": 1119
+    },
+    {
+      "epoch": 0.7168,
+      "grad_norm": 0.22716142524028313,
+      "learning_rate": 1.0224081274124884e-05,
+      "loss": 0.2179,
+      "step": 1120
+    },
+    {
+      "epoch": 0.71744,
+      "grad_norm": 0.23934307222706375,
+      "learning_rate": 1.018139448668769e-05,
+      "loss": 0.2227,
+      "step": 1121
+    },
+    {
+      "epoch": 0.71808,
+      "grad_norm": 0.25324563992500065,
+      "learning_rate": 1.013877419970564e-05,
+      "loss": 0.2477,
+      "step": 1122
+    },
+    {
+      "epoch": 0.71872,
+      "grad_norm": 0.2476691072531496,
+      "learning_rate": 1.0096220604442921e-05,
+      "loss": 0.2304,
+      "step": 1123
+    },
+    {
+      "epoch": 0.71936,
+      "grad_norm": 0.2361471680546047,
+      "learning_rate": 1.0053733891864428e-05,
+      "loss": 0.2173,
+      "step": 1124
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.21784797724300706,
+      "learning_rate": 1.001131425263491e-05,
+      "loss": 0.23,
+      "step": 1125
+    },
+    {
+      "epoch": 0.72064,
+      "grad_norm": 0.24350185601333185,
+      "learning_rate": 9.968961877118116e-06,
+      "loss": 0.2033,
+      "step": 1126
+    },
+    {
+      "epoch": 0.72128,
+      "grad_norm": 0.2349437611897262,
+      "learning_rate": 9.926676955375944e-06,
+      "loss": 0.2394,
+      "step": 1127
+    },
+    {
+      "epoch": 0.72192,
+      "grad_norm": 0.2323461974295896,
+      "learning_rate": 9.884459677167579e-06,
+      "loss": 0.2302,
+      "step": 1128
+    },
+    {
+      "epoch": 0.72256,
+      "grad_norm": 0.24796804636850636,
+      "learning_rate": 9.84231023194865e-06,
+      "loss": 0.2258,
+      "step": 1129
+    },
+    {
+      "epoch": 0.7232,
+      "grad_norm": 0.2359200888166391,
+      "learning_rate": 9.800228808870379e-06,
+      "loss": 0.213,
+      "step": 1130
+    },
+    {
+      "epoch": 0.72384,
+      "grad_norm": 0.23276646327001516,
+      "learning_rate": 9.758215596778722e-06,
+      "loss": 0.2188,
+      "step": 1131
+    },
+    {
+      "epoch": 0.72448,
+      "grad_norm": 0.23852131735997478,
+      "learning_rate": 9.716270784213543e-06,
+      "loss": 0.2203,
+      "step": 1132
+    },
+    {
+      "epoch": 0.72512,
+      "grad_norm": 0.3186712004967345,
+      "learning_rate": 9.674394559407718e-06,
+      "loss": 0.2312,
+      "step": 1133
+    },
+    {
+      "epoch": 0.72576,
+      "grad_norm": 0.23845955234210406,
+      "learning_rate": 9.63258711028639e-06,
+      "loss": 0.2306,
+      "step": 1134
+    },
+    {
+      "epoch": 0.7264,
+      "grad_norm": 0.2501633767336366,
+      "learning_rate": 9.59084862446599e-06,
+      "loss": 0.2261,
+      "step": 1135
+    },
+    {
+      "epoch": 0.72704,
+      "grad_norm": 0.2426029681228632,
+      "learning_rate": 9.549179289253533e-06,
+      "loss": 0.2351,
+      "step": 1136
+    },
+    {
+      "epoch": 0.72768,
+      "grad_norm": 0.4381014753520497,
+      "learning_rate": 9.507579291645654e-06,
+      "loss": 0.2345,
+      "step": 1137
+    },
+    {
+      "epoch": 0.72832,
+      "grad_norm": 0.24829712206880325,
+      "learning_rate": 9.466048818327868e-06,
+      "loss": 0.2218,
+      "step": 1138
+    },
+    {
+      "epoch": 0.72896,
+      "grad_norm": 0.2256828881929371,
+      "learning_rate": 9.42458805567367e-06,
+      "loss": 0.2149,
+      "step": 1139
+    },
+    {
+      "epoch": 0.7296,
+      "grad_norm": 0.2683328694411975,
+      "learning_rate": 9.383197189743717e-06,
+      "loss": 0.202,
+      "step": 1140
+    },
+    {
+      "epoch": 0.73024,
+      "grad_norm": 0.23327018930794263,
+      "learning_rate": 9.341876406285027e-06,
+      "loss": 0.2103,
+      "step": 1141
+    },
+    {
+      "epoch": 0.73088,
+      "grad_norm": 0.22911945516789914,
+      "learning_rate": 9.300625890730066e-06,
+      "loss": 0.2484,
+      "step": 1142
+    },
+    {
+      "epoch": 0.73152,
+      "grad_norm": 0.23613541073187627,
+      "learning_rate": 9.259445828195995e-06,
+      "loss": 0.1989,
+      "step": 1143
+    },
+    {
+      "epoch": 0.73216,
+      "grad_norm": 0.2279871010899268,
+      "learning_rate": 9.218336403483797e-06,
+      "loss": 0.2277,
+      "step": 1144
+    },
+    {
+      "epoch": 0.7328,
+      "grad_norm": 0.21799813294931664,
+      "learning_rate": 9.17729780107746e-06,
+      "loss": 0.2162,
+      "step": 1145
+    },
+    {
+      "epoch": 0.73344,
+      "grad_norm": 0.24265201972066888,
+      "learning_rate": 9.136330205143153e-06,
+      "loss": 0.2295,
+      "step": 1146
+    },
+    {
+      "epoch": 0.73408,
+      "grad_norm": 0.23351669638890205,
+      "learning_rate": 9.095433799528388e-06,
+      "loss": 0.2218,
+      "step": 1147
+    },
+    {
+      "epoch": 0.73472,
+      "grad_norm": 0.25986001347199694,
+      "learning_rate": 9.0546087677612e-06,
+      "loss": 0.24,
+      "step": 1148
+    },
+    {
+      "epoch": 0.73536,
+      "grad_norm": 0.2869469344969374,
+      "learning_rate": 9.013855293049328e-06,
+      "loss": 0.2335,
+      "step": 1149
+    },
+    {
+      "epoch": 0.736,
+      "grad_norm": 0.23982207230566563,
+      "learning_rate": 8.973173558279389e-06,
+      "loss": 0.2261,
+      "step": 1150
+    },
+    {
+      "epoch": 0.73664,
+      "grad_norm": 0.25914493590582866,
+      "learning_rate": 8.932563746016064e-06,
+      "loss": 0.2357,
+      "step": 1151
+    },
+    {
+      "epoch": 0.73728,
+      "grad_norm": 0.24128802191873547,
+      "learning_rate": 8.892026038501242e-06,
+      "loss": 0.2036,
+      "step": 1152
+    },
+    {
+      "epoch": 0.73792,
+      "grad_norm": 0.24901580726093003,
+      "learning_rate": 8.851560617653273e-06,
+      "loss": 0.2123,
+      "step": 1153
+    },
+    {
+      "epoch": 0.73856,
+      "grad_norm": 0.22314334513810227,
+      "learning_rate": 8.811167665066095e-06,
+      "loss": 0.2332,
+      "step": 1154
+    },
+    {
+      "epoch": 0.7392,
+      "grad_norm": 0.24908159050452208,
+      "learning_rate": 8.770847362008427e-06,
+      "loss": 0.2289,
+      "step": 1155
+    },
+    {
+      "epoch": 0.73984,
+      "grad_norm": 0.3051006440704254,
+      "learning_rate": 8.73059988942298e-06,
+      "loss": 0.2421,
+      "step": 1156
+    },
+    {
+      "epoch": 0.74048,
+      "grad_norm": 0.23335754529616762,
+      "learning_rate": 8.690425427925591e-06,
+      "loss": 0.207,
+      "step": 1157
+    },
+    {
+      "epoch": 0.74112,
+      "grad_norm": 0.237316907054958,
+      "learning_rate": 8.650324157804515e-06,
+      "loss": 0.2301,
+      "step": 1158
+    },
+    {
+      "epoch": 0.74176,
+      "grad_norm": 0.25233886865351574,
+      "learning_rate": 8.610296259019479e-06,
+      "loss": 0.2293,
+      "step": 1159
+    },
+    {
+      "epoch": 0.7424,
+      "grad_norm": 0.23991185636703868,
+      "learning_rate": 8.570341911201011e-06,
+      "loss": 0.2432,
+      "step": 1160
+    },
+    {
+      "epoch": 0.74304,
+      "grad_norm": 0.23719495299587995,
+      "learning_rate": 8.53046129364952e-06,
+      "loss": 0.2406,
+      "step": 1161
+    },
+    {
+      "epoch": 0.74368,
+      "grad_norm": 0.24611542047809729,
+      "learning_rate": 8.490654585334562e-06,
+      "loss": 0.1941,
+      "step": 1162
+    },
+    {
+      "epoch": 0.74432,
+      "grad_norm": 0.23899311502661658,
+      "learning_rate": 8.450921964894017e-06,
+      "loss": 0.2327,
+      "step": 1163
+    },
+    {
+      "epoch": 0.74496,
+      "grad_norm": 0.2576114411204611,
+      "learning_rate": 8.411263610633283e-06,
+      "loss": 0.2412,
+      "step": 1164
+    },
+    {
+      "epoch": 0.7456,
+      "grad_norm": 0.22553622707903853,
+      "learning_rate": 8.371679700524476e-06,
+      "loss": 0.2165,
+      "step": 1165
+    },
+    {
+      "epoch": 0.74624,
+      "grad_norm": 0.22217964115194208,
+      "learning_rate": 8.332170412205637e-06,
+      "loss": 0.2248,
+      "step": 1166
+    },
+    {
+      "epoch": 0.74688,
+      "grad_norm": 0.2219067698202354,
+      "learning_rate": 8.29273592297993e-06,
+      "loss": 0.2172,
+      "step": 1167
+    },
+    {
+      "epoch": 0.74752,
+      "grad_norm": 0.2188161709024101,
+      "learning_rate": 8.25337640981485e-06,
+      "loss": 0.2041,
+      "step": 1168
+    },
+    {
+      "epoch": 0.74816,
+      "grad_norm": 0.22371194602316422,
+      "learning_rate": 8.214092049341427e-06,
+      "loss": 0.2131,
+      "step": 1169
+    },
+    {
+      "epoch": 0.7488,
+      "grad_norm": 0.22074333149806447,
+      "learning_rate": 8.174883017853429e-06,
+      "loss": 0.2056,
+      "step": 1170
+    },
+    {
+      "epoch": 0.74944,
+      "grad_norm": 0.23363590876098603,
+      "learning_rate": 8.13574949130658e-06,
+      "loss": 0.2061,
+      "step": 1171
+    },
+    {
+      "epoch": 0.75008,
+      "grad_norm": 0.2575584769487215,
+      "learning_rate": 8.096691645317767e-06,
+      "loss": 0.2376,
+      "step": 1172
+    },
+    {
+      "epoch": 0.75072,
+      "grad_norm": 0.252720295492527,
+      "learning_rate": 8.057709655164242e-06,
+      "loss": 0.2186,
+      "step": 1173
+    },
+    {
+      "epoch": 0.75136,
+      "grad_norm": 0.21586281460898304,
+      "learning_rate": 8.018803695782853e-06,
+      "loss": 0.2049,
+      "step": 1174
+    },
+    {
+      "epoch": 0.752,
+      "grad_norm": 0.23035320065808185,
+      "learning_rate": 7.979973941769257e-06,
+      "loss": 0.2295,
+      "step": 1175
+    },
+    {
+      "epoch": 0.75264,
+      "grad_norm": 0.2421794800112151,
+      "learning_rate": 7.941220567377089e-06,
+      "loss": 0.2421,
+      "step": 1176
+    },
+    {
+      "epoch": 0.75328,
+      "grad_norm": 0.21830541101271783,
+      "learning_rate": 7.902543746517277e-06,
+      "loss": 0.2287,
+      "step": 1177
+    },
+    {
+      "epoch": 0.75392,
+      "grad_norm": 0.25560658369492234,
+      "learning_rate": 7.863943652757166e-06,
+      "loss": 0.2272,
+      "step": 1178
+    },
+    {
+      "epoch": 0.75456,
+      "grad_norm": 0.23128581371738616,
+      "learning_rate": 7.825420459319797e-06,
+      "loss": 0.2221,
+      "step": 1179
+    },
+    {
+      "epoch": 0.7552,
+      "grad_norm": 0.2397924464434262,
+      "learning_rate": 7.786974339083105e-06,
+      "loss": 0.2161,
+      "step": 1180
+    },
+    {
+      "epoch": 0.75584,
+      "grad_norm": 0.24722945285723155,
+      "learning_rate": 7.748605464579128e-06,
+      "loss": 0.2121,
+      "step": 1181
+    },
+    {
+      "epoch": 0.75648,
+      "grad_norm": 0.24487163431041922,
+      "learning_rate": 7.710314007993302e-06,
+      "loss": 0.2392,
+      "step": 1182
+    },
+    {
+      "epoch": 0.75712,
+      "grad_norm": 0.2663607132358166,
+      "learning_rate": 7.672100141163577e-06,
+      "loss": 0.2241,
+      "step": 1183
+    },
+    {
+      "epoch": 0.75776,
+      "grad_norm": 0.21574064551083452,
+      "learning_rate": 7.633964035579776e-06,
+      "loss": 0.227,
+      "step": 1184
+    },
+    {
+      "epoch": 0.7584,
+      "grad_norm": 0.22235132512931682,
+      "learning_rate": 7.595905862382705e-06,
+      "loss": 0.2389,
+      "step": 1185
+    },
+    {
+      "epoch": 0.75904,
+      "grad_norm": 0.2235683342421712,
+      "learning_rate": 7.557925792363457e-06,
+      "loss": 0.2213,
+      "step": 1186
+    },
+    {
+      "epoch": 0.75968,
+      "grad_norm": 0.24087158387016824,
+      "learning_rate": 7.5200239959626325e-06,
+      "loss": 0.2118,
+      "step": 1187
+    },
+    {
+      "epoch": 0.76032,
+      "grad_norm": 0.26677780873719925,
+      "learning_rate": 7.48220064326956e-06,
+      "loss": 0.2218,
+      "step": 1188
+    },
+    {
+      "epoch": 0.76096,
+      "grad_norm": 0.21548096876347833,
+      "learning_rate": 7.444455904021549e-06,
+      "loss": 0.227,
+      "step": 1189
+    },
+    {
+      "epoch": 0.7616,
+      "grad_norm": 0.2575652398220675,
+      "learning_rate": 7.4067899476031154e-06,
+      "loss": 0.2107,
+      "step": 1190
+    },
+    {
+      "epoch": 0.76224,
+      "grad_norm": 0.21071320245963931,
+      "learning_rate": 7.369202943045228e-06,
+      "loss": 0.2005,
+      "step": 1191
+    },
+    {
+      "epoch": 0.76288,
+      "grad_norm": 0.21003990105143006,
+      "learning_rate": 7.331695059024546e-06,
+      "loss": 0.2118,
+      "step": 1192
+    },
+    {
+      "epoch": 0.76352,
+      "grad_norm": 0.2835839824014075,
+      "learning_rate": 7.29426646386267e-06,
+      "loss": 0.2216,
+      "step": 1193
+    },
+    {
+      "epoch": 0.76416,
+      "grad_norm": 0.22826459808701152,
+      "learning_rate": 7.256917325525378e-06,
+      "loss": 0.2305,
+      "step": 1194
+    },
+    {
+      "epoch": 0.7648,
+      "grad_norm": 0.23642740014825986,
+      "learning_rate": 7.219647811621874e-06,
+      "loss": 0.2113,
+      "step": 1195
+    },
+    {
+      "epoch": 0.76544,
+      "grad_norm": 0.24970432900903092,
+      "learning_rate": 7.18245808940404e-06,
+      "loss": 0.218,
+      "step": 1196
+    },
+    {
+      "epoch": 0.76608,
+      "grad_norm": 0.20914458315959308,
+      "learning_rate": 7.145348325765683e-06,
+      "loss": 0.2096,
+      "step": 1197
+    },
+    {
+      "epoch": 0.76672,
+      "grad_norm": 0.24295600915188692,
+      "learning_rate": 7.108318687241777e-06,
+      "loss": 0.2162,
+      "step": 1198
+    },
+    {
+      "epoch": 0.76736,
+      "grad_norm": 0.2668793565791304,
+      "learning_rate": 7.071369340007744e-06,
+      "loss": 0.2273,
+      "step": 1199
+    },
+    {
+      "epoch": 0.768,
+      "grad_norm": 0.23253327949163685,
+      "learning_rate": 7.034500449878648e-06,
+      "loss": 0.2462,
+      "step": 1200
+    },
+    {
+      "epoch": 0.76864,
+      "grad_norm": 0.252448254870111,
+      "learning_rate": 6.9977121823085464e-06,
+      "loss": 0.2307,
+      "step": 1201
+    },
+    {
+      "epoch": 0.76928,
+      "grad_norm": 0.2959955620127609,
+      "learning_rate": 6.961004702389637e-06,
+      "loss": 0.2275,
+      "step": 1202
+    },
+    {
+      "epoch": 0.76992,
+      "grad_norm": 0.25938376118159756,
+      "learning_rate": 6.9243781748516145e-06,
+      "loss": 0.2127,
+      "step": 1203
+    },
+    {
+      "epoch": 0.77056,
+      "grad_norm": 0.22848777183634242,
+      "learning_rate": 6.887832764060881e-06,
+      "loss": 0.2317,
+      "step": 1204
+    },
+    {
+      "epoch": 0.7712,
+      "grad_norm": 0.20505924148335786,
+      "learning_rate": 6.851368634019778e-06,
+      "loss": 0.2131,
+      "step": 1205
+    },
+    {
+      "epoch": 0.77184,
+      "grad_norm": 0.26843010093768244,
+      "learning_rate": 6.8149859483659515e-06,
+      "loss": 0.2126,
+      "step": 1206
+    },
+    {
+      "epoch": 0.77248,
+      "grad_norm": 0.24248131976447143,
+      "learning_rate": 6.778684870371496e-06,
+      "loss": 0.2246,
+      "step": 1207
+    },
+    {
+      "epoch": 0.77312,
+      "grad_norm": 0.3801285107189877,
+      "learning_rate": 6.742465562942335e-06,
+      "loss": 0.2321,
+      "step": 1208
+    },
+    {
+      "epoch": 0.77376,
+      "grad_norm": 0.2894453287797247,
+      "learning_rate": 6.706328188617386e-06,
+      "loss": 0.2101,
+      "step": 1209
+    },
+    {
+      "epoch": 0.7744,
+      "grad_norm": 0.22945081417174015,
+      "learning_rate": 6.6702729095679095e-06,
+      "loss": 0.2188,
+      "step": 1210
+    },
+    {
+      "epoch": 0.77504,
+      "grad_norm": 0.22961868282072362,
+      "learning_rate": 6.6342998875967485e-06,
+      "loss": 0.2126,
+      "step": 1211
+    },
+    {
+      "epoch": 0.77568,
+      "grad_norm": 0.24359372317851352,
+      "learning_rate": 6.598409284137602e-06,
+      "loss": 0.2367,
+      "step": 1212
+    },
+    {
+      "epoch": 0.77632,
+      "grad_norm": 0.21475194458018806,
+      "learning_rate": 6.5626012602543085e-06,
+      "loss": 0.2196,
+      "step": 1213
+    },
+    {
+      "epoch": 0.77696,
+      "grad_norm": 0.22300041578900634,
+      "learning_rate": 6.5268759766401174e-06,
+      "loss": 0.2061,
+      "step": 1214
+    },
+    {
+      "epoch": 0.7776,
+      "grad_norm": 0.24555482596347522,
+      "learning_rate": 6.491233593616971e-06,
+      "loss": 0.2137,
+      "step": 1215
+    },
+    {
+      "epoch": 0.77824,
+      "grad_norm": 0.24419669696500723,
+      "learning_rate": 6.455674271134787e-06,
+      "loss": 0.2107,
+      "step": 1216
+    },
+    {
+      "epoch": 0.77888,
+      "grad_norm": 0.22003884505706134,
+      "learning_rate": 6.420198168770733e-06,
+      "loss": 0.2389,
+      "step": 1217
+    },
+    {
+      "epoch": 0.77952,
+      "grad_norm": 0.24045675666383412,
+      "learning_rate": 6.384805445728517e-06,
+      "loss": 0.2077,
+      "step": 1218
+    },
+    {
+      "epoch": 0.78016,
+      "grad_norm": 0.22316484387720012,
+      "learning_rate": 6.349496260837675e-06,
+      "loss": 0.206,
+      "step": 1219
+    },
+    {
+      "epoch": 0.7808,
+      "grad_norm": 0.23540776334696803,
+      "learning_rate": 6.314270772552849e-06,
+      "loss": 0.2058,
+      "step": 1220
+    },
+    {
+      "epoch": 0.78144,
+      "grad_norm": 0.22966766584099912,
+      "learning_rate": 6.2791291389530825e-06,
+      "loss": 0.2045,
+      "step": 1221
+    },
+    {
+      "epoch": 0.78208,
+      "grad_norm": 0.2240899940551242,
+      "learning_rate": 6.244071517741112e-06,
+      "loss": 0.2166,
+      "step": 1222
+    },
+    {
+      "epoch": 0.78272,
+      "grad_norm": 0.22584809771662878,
+      "learning_rate": 6.209098066242666e-06,
+      "loss": 0.1893,
+      "step": 1223
+    },
+    {
+      "epoch": 0.78336,
+      "grad_norm": 0.23967568670298528,
+      "learning_rate": 6.174208941405718e-06,
+      "loss": 0.2362,
+      "step": 1224
+    },
+    {
+      "epoch": 0.784,
+      "grad_norm": 0.24358640659674474,
+      "learning_rate": 6.139404299799864e-06,
+      "loss": 0.2283,
+      "step": 1225
+    },
+    {
+      "epoch": 0.78464,
+      "grad_norm": 0.23485095037921896,
+      "learning_rate": 6.104684297615518e-06,
+      "loss": 0.2108,
+      "step": 1226
+    },
+    {
+      "epoch": 0.78528,
+      "grad_norm": 0.22624410501145345,
+      "learning_rate": 6.070049090663308e-06,
+      "loss": 0.2094,
+      "step": 1227
+    },
+    {
+      "epoch": 0.78592,
+      "grad_norm": 0.2608989034285192,
+      "learning_rate": 6.035498834373313e-06,
+      "loss": 0.2498,
+      "step": 1228
+    },
+    {
+      "epoch": 0.78656,
+      "grad_norm": 0.22110036947871828,
+      "learning_rate": 6.001033683794363e-06,
+      "loss": 0.2348,
+      "step": 1229
+    },
+    {
+      "epoch": 0.7872,
+      "grad_norm": 0.2124254456256694,
+      "learning_rate": 5.966653793593419e-06,
+      "loss": 0.2112,
+      "step": 1230
+    },
+    {
+      "epoch": 0.78784,
+      "grad_norm": 0.2343604692769573,
+      "learning_rate": 5.932359318054759e-06,
+      "loss": 0.2372,
+      "step": 1231
+    },
+    {
+      "epoch": 0.78848,
+      "grad_norm": 0.21792784674511467,
+      "learning_rate": 5.898150411079417e-06,
+      "loss": 0.2181,
+      "step": 1232
+    },
+    {
+      "epoch": 0.78912,
+      "grad_norm": 0.23317866352552133,
+      "learning_rate": 5.864027226184368e-06,
+      "loss": 0.2175,
+      "step": 1233
+    },
+    {
+      "epoch": 0.78976,
+      "grad_norm": 0.22182689476613604,
+      "learning_rate": 5.829989916501938e-06,
+      "loss": 0.2416,
+      "step": 1234
+    },
+    {
+      "epoch": 0.7904,
+      "grad_norm": 0.2125544591628715,
+      "learning_rate": 5.796038634779058e-06,
+      "loss": 0.2094,
+      "step": 1235
+    },
+    {
+      "epoch": 0.79104,
+      "grad_norm": 0.24318176263180385,
+      "learning_rate": 5.762173533376603e-06,
+      "loss": 0.2014,
+      "step": 1236
+    },
+    {
+      "epoch": 0.79168,
+      "grad_norm": 0.24449585502285096,
+      "learning_rate": 5.728394764268702e-06,
+      "loss": 0.2182,
+      "step": 1237
+    },
+    {
+      "epoch": 0.79232,
+      "grad_norm": 0.212375425559508,
+      "learning_rate": 5.694702479042055e-06,
+      "loss": 0.2157,
+      "step": 1238
+    },
+    {
+      "epoch": 0.79296,
+      "grad_norm": 0.24463705041225112,
+      "learning_rate": 5.661096828895254e-06,
+      "loss": 0.2253,
+      "step": 1239
+    },
+    {
+      "epoch": 0.7936,
+      "grad_norm": 0.2213101090236379,
+      "learning_rate": 5.627577964638109e-06,
+      "loss": 0.2099,
+      "step": 1240
+    },
+    {
+      "epoch": 0.79424,
+      "grad_norm": 0.22983280526878436,
+      "learning_rate": 5.594146036690961e-06,
+      "loss": 0.2106,
+      "step": 1241
+    },
+    {
+      "epoch": 0.79488,
+      "grad_norm": 0.21360866108460141,
+      "learning_rate": 5.560801195084018e-06,
+      "loss": 0.2226,
+      "step": 1242
+    },
+    {
+      "epoch": 0.79552,
+      "grad_norm": 0.23635622958477748,
+      "learning_rate": 5.527543589456668e-06,
+      "loss": 0.2211,
+      "step": 1243
+    },
+    {
+      "epoch": 0.79616,
+      "grad_norm": 0.22322101495943988,
+      "learning_rate": 5.494373369056826e-06,
+      "loss": 0.2096,
+      "step": 1244
+    },
+    {
+      "epoch": 0.7968,
+      "grad_norm": 0.24596462352726498,
+      "learning_rate": 5.461290682740247e-06,
+      "loss": 0.2227,
+      "step": 1245
+    },
+    {
+      "epoch": 0.79744,
+      "grad_norm": 0.23755939350134542,
+      "learning_rate": 5.428295678969872e-06,
+      "loss": 0.2389,
+      "step": 1246
+    },
+    {
+      "epoch": 0.79808,
+      "grad_norm": 0.250227656242041,
+      "learning_rate": 5.3953885058151525e-06,
+      "loss": 0.2248,
+      "step": 1247
+    },
+    {
+      "epoch": 0.79872,
+      "grad_norm": 0.22822877559585875,
+      "learning_rate": 5.3625693109513704e-06,
+      "loss": 0.2154,
+      "step": 1248
+    },
+    {
+      "epoch": 0.79936,
+      "grad_norm": 0.2504360656958186,
+      "learning_rate": 5.329838241659038e-06,
+      "loss": 0.2259,
+      "step": 1249
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.23888958973870866,
+      "learning_rate": 5.297195444823136e-06,
+      "loss": 0.2037,
+      "step": 1250
+    },
+    {
+      "epoch": 0.80064,
+      "grad_norm": 0.23155307902046116,
+      "learning_rate": 5.264641066932563e-06,
+      "loss": 0.2078,
+      "step": 1251
+    },
+    {
+      "epoch": 0.80128,
+      "grad_norm": 0.22442202407676448,
+      "learning_rate": 5.2321752540793855e-06,
+      "loss": 0.239,
+      "step": 1252
+    },
+    {
+      "epoch": 0.80192,
+      "grad_norm": 0.21100181579650684,
+      "learning_rate": 5.199798151958238e-06,
+      "loss": 0.2085,
+      "step": 1253
+    },
+    {
+      "epoch": 0.80256,
+      "grad_norm": 0.21891369156648757,
+      "learning_rate": 5.167509905865675e-06,
+      "loss": 0.2136,
+      "step": 1254
+    },
+    {
+      "epoch": 0.8032,
+      "grad_norm": 0.23405793643390937,
+      "learning_rate": 5.135310660699453e-06,
+      "loss": 0.212,
+      "step": 1255
+    },
+    {
+      "epoch": 0.80384,
+      "grad_norm": 0.27593157371009747,
+      "learning_rate": 5.103200560957974e-06,
+      "loss": 0.2294,
+      "step": 1256
+    },
+    {
+      "epoch": 0.80448,
+      "grad_norm": 0.38708872808005773,
+      "learning_rate": 5.071179750739544e-06,
+      "loss": 0.2126,
+      "step": 1257
+    },
+    {
+      "epoch": 0.80512,
+      "grad_norm": 0.24488074932262327,
+      "learning_rate": 5.039248373741801e-06,
+      "loss": 0.2467,
+      "step": 1258
+    },
+    {
+      "epoch": 0.80576,
+      "grad_norm": 0.21243688153602522,
+      "learning_rate": 5.007406573261022e-06,
+      "loss": 0.2138,
+      "step": 1259
+    },
+    {
+      "epoch": 0.8064,
+      "grad_norm": 0.23388239920554604,
+      "learning_rate": 4.975654492191509e-06,
+      "loss": 0.2161,
+      "step": 1260
+    },
+    {
+      "epoch": 0.80704,
+      "grad_norm": 0.2257775942613988,
+      "learning_rate": 4.943992273024933e-06,
+      "loss": 0.2014,
+      "step": 1261
+    },
+    {
+      "epoch": 0.80768,
+      "grad_norm": 0.19657893575408056,
+      "learning_rate": 4.912420057849693e-06,
+      "loss": 0.2261,
+      "step": 1262
+    },
+    {
+      "epoch": 0.80832,
+      "grad_norm": 0.23235701795648625,
+      "learning_rate": 4.88093798835029e-06,
+      "loss": 0.2088,
+      "step": 1263
+    },
+    {
+      "epoch": 0.80896,
+      "grad_norm": 0.22870370651478264,
+      "learning_rate": 4.849546205806679e-06,
+      "loss": 0.204,
+      "step": 1264
+    },
+    {
+      "epoch": 0.8096,
+      "grad_norm": 0.22524532252921017,
+      "learning_rate": 4.818244851093642e-06,
+      "loss": 0.22,
+      "step": 1265
+    },
+    {
+      "epoch": 0.81024,
+      "grad_norm": 0.23755634993386263,
+      "learning_rate": 4.787034064680157e-06,
+      "loss": 0.2072,
+      "step": 1266
+    },
+    {
+      "epoch": 0.81088,
+      "grad_norm": 0.2205974119770644,
+      "learning_rate": 4.755913986628757e-06,
+      "loss": 0.1955,
+      "step": 1267
+    },
+    {
+      "epoch": 0.81152,
+      "grad_norm": 0.23878480505086405,
+      "learning_rate": 4.724884756594922e-06,
+      "loss": 0.2213,
+      "step": 1268
+    },
+    {
+      "epoch": 0.81216,
+      "grad_norm": 0.22644611132255146,
+      "learning_rate": 4.693946513826428e-06,
+      "loss": 0.2066,
+      "step": 1269
+    },
+    {
+      "epoch": 0.8128,
+      "grad_norm": 0.23435476358782967,
+      "learning_rate": 4.663099397162734e-06,
+      "loss": 0.2152,
+      "step": 1270
+    },
+    {
+      "epoch": 0.81344,
+      "grad_norm": 0.22597050768455026,
+      "learning_rate": 4.632343545034376e-06,
+      "loss": 0.2313,
+      "step": 1271
+    },
+    {
+      "epoch": 0.81408,
+      "grad_norm": 0.2622797153339369,
+      "learning_rate": 4.6016790954622915e-06,
+      "loss": 0.221,
+      "step": 1272
+    },
+    {
+      "epoch": 0.81472,
+      "grad_norm": 0.21380031437179098,
+      "learning_rate": 4.5711061860572835e-06,
+      "loss": 0.2175,
+      "step": 1273
+    },
+    {
+      "epoch": 0.81536,
+      "grad_norm": 0.2570851013227341,
+      "learning_rate": 4.54062495401931e-06,
+      "loss": 0.2221,
+      "step": 1274
+    },
+    {
+      "epoch": 0.816,
+      "grad_norm": 0.24326649772484504,
+      "learning_rate": 4.510235536136962e-06,
+      "loss": 0.2176,
+      "step": 1275
+    },
+    {
+      "epoch": 0.81664,
+      "grad_norm": 0.24724927149179474,
+      "learning_rate": 4.479938068786757e-06,
+      "loss": 0.2529,
+      "step": 1276
+    },
+    {
+      "epoch": 0.81728,
+      "grad_norm": 0.22198457602234464,
+      "learning_rate": 4.449732687932595e-06,
+      "loss": 0.2214,
+      "step": 1277
+    },
+    {
+      "epoch": 0.81792,
+      "grad_norm": 0.1962706510510867,
+      "learning_rate": 4.419619529125143e-06,
+      "loss": 0.2339,
+      "step": 1278
+    },
+    {
+      "epoch": 0.81856,
+      "grad_norm": 0.2462915433604134,
+      "learning_rate": 4.389598727501165e-06,
+      "loss": 0.2116,
+      "step": 1279
+    },
+    {
+      "epoch": 0.8192,
+      "grad_norm": 0.2450893774550224,
+      "learning_rate": 4.359670417783007e-06,
+      "loss": 0.237,
+      "step": 1280
+    },
+    {
+      "epoch": 0.81984,
+      "grad_norm": 0.2326935222216056,
+      "learning_rate": 4.329834734277907e-06,
+      "loss": 0.2202,
+      "step": 1281
+    },
+    {
+      "epoch": 0.82048,
+      "grad_norm": 0.2176535552968833,
+      "learning_rate": 4.300091810877452e-06,
+      "loss": 0.2386,
+      "step": 1282
+    },
+    {
+      "epoch": 0.82112,
+      "grad_norm": 0.22683377003308944,
+      "learning_rate": 4.2704417810569524e-06,
+      "loss": 0.2176,
+      "step": 1283
+    },
+    {
+      "epoch": 0.82176,
+      "grad_norm": 0.22378672342506417,
+      "learning_rate": 4.240884777874843e-06,
+      "loss": 0.2402,
+      "step": 1284
+    },
+    {
+      "epoch": 0.8224,
+      "grad_norm": 0.22747323224483365,
+      "learning_rate": 4.21142093397209e-06,
+      "loss": 0.2285,
+      "step": 1285
+    },
+    {
+      "epoch": 0.82304,
+      "grad_norm": 0.24952869232848146,
+      "learning_rate": 4.182050381571598e-06,
+      "loss": 0.2153,
+      "step": 1286
+    },
+    {
+      "epoch": 0.82368,
+      "grad_norm": 0.2368790088312886,
+      "learning_rate": 4.1527732524776065e-06,
+      "loss": 0.214,
+      "step": 1287
+    },
+    {
+      "epoch": 0.82432,
+      "grad_norm": 0.21005408914353893,
+      "learning_rate": 4.123589678075118e-06,
+      "loss": 0.2026,
+      "step": 1288
+    },
+    {
+      "epoch": 0.82496,
+      "grad_norm": 0.21607068715865752,
+      "learning_rate": 4.094499789329284e-06,
+      "loss": 0.2104,
+      "step": 1289
+    },
+    {
+      "epoch": 0.8256,
+      "grad_norm": 0.21822116261547195,
+      "learning_rate": 4.0655037167848326e-06,
+      "loss": 0.2218,
+      "step": 1290
+    },
+    {
+      "epoch": 0.82624,
+      "grad_norm": 0.23958436004104197,
+      "learning_rate": 4.0366015905654805e-06,
+      "loss": 0.2236,
+      "step": 1291
+    },
+    {
+      "epoch": 0.82688,
+      "grad_norm": 0.22639217871463924,
+      "learning_rate": 4.00779354037335e-06,
+      "loss": 0.2162,
+      "step": 1292
+    },
+    {
+      "epoch": 0.82752,
+      "grad_norm": 0.2365878154004433,
+      "learning_rate": 3.979079695488383e-06,
+      "loss": 0.1979,
+      "step": 1293
+    },
+    {
+      "epoch": 0.82816,
+      "grad_norm": 0.2329143228892626,
+      "learning_rate": 3.950460184767757e-06,
+      "loss": 0.219,
+      "step": 1294
+    },
+    {
+      "epoch": 0.8288,
+      "grad_norm": 0.2736640870762084,
+      "learning_rate": 3.921935136645327e-06,
+      "loss": 0.2258,
+      "step": 1295
+    },
+    {
+      "epoch": 0.82944,
+      "grad_norm": 0.2387728195721327,
+      "learning_rate": 3.893504679131008e-06,
+      "loss": 0.2008,
+      "step": 1296
+    },
+    {
+      "epoch": 0.83008,
+      "grad_norm": 0.3025091557443113,
+      "learning_rate": 3.865168939810262e-06,
+      "loss": 0.2145,
+      "step": 1297
+    },
+    {
+      "epoch": 0.83072,
+      "grad_norm": 0.22622793831419782,
+      "learning_rate": 3.836928045843455e-06,
+      "loss": 0.2191,
+      "step": 1298
+    },
+    {
+      "epoch": 0.83136,
+      "grad_norm": 0.2317299593907126,
+      "learning_rate": 3.8087821239653625e-06,
+      "loss": 0.2184,
+      "step": 1299
+    },
+    {
+      "epoch": 0.832,
+      "grad_norm": 0.21199722754396097,
+      "learning_rate": 3.780731300484516e-06,
+      "loss": 0.2242,
+      "step": 1300
+    },
+    {
+      "epoch": 0.83264,
+      "grad_norm": 0.2262737481331807,
+      "learning_rate": 3.7527757012827126e-06,
+      "loss": 0.2188,
+      "step": 1301
+    },
+    {
+      "epoch": 0.83328,
+      "grad_norm": 0.21609800165740564,
+      "learning_rate": 3.724915451814401e-06,
+      "loss": 0.2025,
+      "step": 1302
+    },
+    {
+      "epoch": 0.83392,
+      "grad_norm": 0.23639901983878522,
+      "learning_rate": 3.6971506771061396e-06,
+      "loss": 0.213,
+      "step": 1303
+    },
+    {
+      "epoch": 0.83456,
+      "grad_norm": 0.24920717583114138,
+      "learning_rate": 3.6694815017560475e-06,
+      "loss": 0.2116,
+      "step": 1304
+    },
+    {
+      "epoch": 0.8352,
+      "grad_norm": 0.2287588885774272,
+      "learning_rate": 3.6419080499331987e-06,
+      "loss": 0.2061,
+      "step": 1305
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1562,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 261,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 598201326895104.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}