{
  "best_metric": 0.9301,
  "best_model_checkpoint": "checkpoint/swin-base/checkpoint-13320",
  "epoch": 300.0,
  "eval_steps": 500,
  "global_step": 99900,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.03,
      "grad_norm": 9.57372760772705,
      "learning_rate": 9.998998998999e-06,
      "loss": 4.6311,
      "step": 10
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.215463161468506,
      "learning_rate": 9.997997997998e-06,
      "loss": 4.57,
      "step": 20
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.255615711212158,
      "learning_rate": 9.996996996996998e-06,
      "loss": 4.5316,
      "step": 30
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.566373825073242,
      "learning_rate": 9.995995995995997e-06,
      "loss": 4.4767,
      "step": 40
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.302833557128906,
      "learning_rate": 9.994994994994995e-06,
      "loss": 4.4207,
      "step": 50
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.3884124755859375,
      "learning_rate": 9.993993993993994e-06,
      "loss": 4.3579,
      "step": 60
    },
    {
      "epoch": 0.21,
      "grad_norm": 10.630465507507324,
      "learning_rate": 9.992992992992994e-06,
      "loss": 4.2144,
      "step": 70
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.169319152832031,
      "learning_rate": 9.991991991991993e-06,
      "loss": 4.1091,
      "step": 80
    },
    {
      "epoch": 0.27,
      "grad_norm": 10.23892879486084,
      "learning_rate": 9.990990990990992e-06,
      "loss": 3.9887,
      "step": 90
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.140935897827148,
      "learning_rate": 9.989989989989992e-06,
      "loss": 3.838,
      "step": 100
    },
    {
      "epoch": 0.33,
      "grad_norm": 11.872363090515137,
      "learning_rate": 9.98898898898899e-06,
      "loss": 3.6626,
      "step": 110
    },
    {
      "epoch": 0.36,
      "grad_norm": 11.113883018493652,
      "learning_rate": 9.987987987987989e-06,
      "loss": 3.5326,
      "step": 120
    },
    {
      "epoch": 0.39,
      "grad_norm": 15.503419876098633,
      "learning_rate": 9.986986986986988e-06,
      "loss": 3.3042,
      "step": 130
    },
    {
      "epoch": 0.42,
      "grad_norm": 14.594585418701172,
      "learning_rate": 9.985985985985986e-06,
      "loss": 3.111,
      "step": 140
    },
    {
      "epoch": 0.45,
      "grad_norm": 31.187252044677734,
      "learning_rate": 9.984984984984985e-06,
      "loss": 2.9329,
      "step": 150
    },
    {
      "epoch": 0.48,
      "grad_norm": 20.547489166259766,
      "learning_rate": 9.983983983983985e-06,
      "loss": 2.8472,
      "step": 160
    },
    {
      "epoch": 0.51,
      "grad_norm": 20.63930892944336,
      "learning_rate": 9.982982982982984e-06,
      "loss": 2.6238,
      "step": 170
    },
    {
      "epoch": 0.54,
      "grad_norm": 24.311126708984375,
      "learning_rate": 9.981981981981982e-06,
      "loss": 2.4911,
      "step": 180
    },
    {
      "epoch": 0.57,
      "grad_norm": 13.25031566619873,
      "learning_rate": 9.980980980980983e-06,
      "loss": 2.4185,
      "step": 190
    },
    {
      "epoch": 0.6,
      "grad_norm": 43.531578063964844,
      "learning_rate": 9.979979979979981e-06,
      "loss": 2.3114,
      "step": 200
    },
    {
      "epoch": 0.63,
      "grad_norm": 12.955471992492676,
      "learning_rate": 9.97897897897898e-06,
      "loss": 2.2558,
      "step": 210
    },
    {
      "epoch": 0.66,
      "grad_norm": 15.336274147033691,
      "learning_rate": 9.977977977977978e-06,
      "loss": 2.1056,
      "step": 220
    },
    {
      "epoch": 0.69,
      "grad_norm": 12.100207328796387,
      "learning_rate": 9.976976976976977e-06,
      "loss": 2.0495,
      "step": 230
    },
    {
      "epoch": 0.72,
      "grad_norm": 21.610525131225586,
      "learning_rate": 9.975975975975977e-06,
      "loss": 2.0121,
      "step": 240
    },
    {
      "epoch": 0.75,
      "grad_norm": 12.644548416137695,
      "learning_rate": 9.974974974974976e-06,
      "loss": 1.8556,
      "step": 250
    },
    {
      "epoch": 0.78,
      "grad_norm": 15.277923583984375,
      "learning_rate": 9.973973973973974e-06,
      "loss": 1.8089,
      "step": 260
    },
    {
      "epoch": 0.81,
      "grad_norm": 12.686960220336914,
      "learning_rate": 9.972972972972975e-06,
      "loss": 1.7365,
      "step": 270
    },
    {
      "epoch": 0.84,
      "grad_norm": 13.061480522155762,
      "learning_rate": 9.971971971971973e-06,
      "loss": 1.7073,
      "step": 280
    },
    {
      "epoch": 0.87,
      "grad_norm": 14.452112197875977,
      "learning_rate": 9.970970970970972e-06,
      "loss": 1.6694,
      "step": 290
    },
    {
      "epoch": 0.9,
      "grad_norm": 17.65357780456543,
      "learning_rate": 9.96996996996997e-06,
      "loss": 1.6142,
      "step": 300
    },
    {
      "epoch": 0.93,
      "grad_norm": 17.804506301879883,
      "learning_rate": 9.968968968968969e-06,
      "loss": 1.5512,
      "step": 310
    },
    {
      "epoch": 0.96,
      "grad_norm": 21.067514419555664,
      "learning_rate": 9.96796796796797e-06,
      "loss": 1.5347,
      "step": 320
    },
    {
      "epoch": 0.99,
      "grad_norm": 22.17087173461914,
      "learning_rate": 9.966966966966968e-06,
      "loss": 1.4475,
      "step": 330
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8024,
      "eval_loss": 0.8144841194152832,
      "eval_runtime": 32.2937,
      "eval_samples_per_second": 309.658,
      "eval_steps_per_second": 1.239,
      "step": 333
    },
    {
      "epoch": 1.02,
      "grad_norm": 16.76087760925293,
      "learning_rate": 9.965965965965967e-06,
      "loss": 1.4671,
      "step": 340
    },
    {
      "epoch": 1.05,
      "grad_norm": 13.050296783447266,
      "learning_rate": 9.964964964964965e-06,
      "loss": 1.3778,
      "step": 350
    },
    {
      "epoch": 1.08,
      "grad_norm": 24.59625816345215,
      "learning_rate": 9.963963963963965e-06,
      "loss": 1.3997,
      "step": 360
    },
    {
      "epoch": 1.11,
      "grad_norm": 18.178476333618164,
      "learning_rate": 9.962962962962964e-06,
      "loss": 1.3204,
      "step": 370
    },
    {
      "epoch": 1.14,
      "grad_norm": 16.15305519104004,
      "learning_rate": 9.961961961961963e-06,
      "loss": 1.3254,
      "step": 380
    },
    {
      "epoch": 1.17,
      "grad_norm": 13.282833099365234,
      "learning_rate": 9.960960960960961e-06,
      "loss": 1.2781,
      "step": 390
    },
    {
      "epoch": 1.2,
      "grad_norm": 13.037567138671875,
      "learning_rate": 9.95995995995996e-06,
      "loss": 1.3349,
      "step": 400
    },
    {
      "epoch": 1.23,
      "grad_norm": 16.423372268676758,
      "learning_rate": 9.95895895895896e-06,
      "loss": 1.2962,
      "step": 410
    },
    {
      "epoch": 1.26,
      "grad_norm": 17.499189376831055,
      "learning_rate": 9.957957957957959e-06,
      "loss": 1.2996,
      "step": 420
    },
    {
      "epoch": 1.29,
      "grad_norm": 13.364770889282227,
      "learning_rate": 9.956956956956957e-06,
      "loss": 1.2447,
      "step": 430
    },
    {
      "epoch": 1.32,
      "grad_norm": 10.91217041015625,
      "learning_rate": 9.955955955955958e-06,
      "loss": 1.2044,
      "step": 440
    },
    {
      "epoch": 1.35,
      "grad_norm": 21.4530029296875,
      "learning_rate": 9.954954954954956e-06,
      "loss": 1.2419,
      "step": 450
    },
    {
      "epoch": 1.38,
      "grad_norm": 22.028942108154297,
      "learning_rate": 9.953953953953955e-06,
      "loss": 1.1525,
      "step": 460
    },
    {
      "epoch": 1.41,
      "grad_norm": 15.714077949523926,
      "learning_rate": 9.952952952952953e-06,
      "loss": 1.1201,
      "step": 470
    },
    {
      "epoch": 1.44,
      "grad_norm": 18.178543090820312,
      "learning_rate": 9.951951951951952e-06,
      "loss": 1.1787,
      "step": 480
    },
    {
      "epoch": 1.47,
      "grad_norm": 12.53498363494873,
      "learning_rate": 9.950950950950952e-06,
      "loss": 1.1371,
      "step": 490
    },
    {
      "epoch": 1.5,
      "grad_norm": 13.074543952941895,
      "learning_rate": 9.949949949949951e-06,
      "loss": 1.154,
      "step": 500
    },
    {
      "epoch": 1.53,
      "grad_norm": 12.679483413696289,
      "learning_rate": 9.94894894894895e-06,
      "loss": 1.2112,
      "step": 510
    },
    {
      "epoch": 1.56,
      "grad_norm": 20.075759887695312,
      "learning_rate": 9.94794794794795e-06,
      "loss": 1.2047,
      "step": 520
    },
    {
      "epoch": 1.59,
      "grad_norm": 23.097604751586914,
      "learning_rate": 9.946946946946948e-06,
      "loss": 1.0866,
      "step": 530
    },
    {
      "epoch": 1.62,
      "grad_norm": 11.857926368713379,
      "learning_rate": 9.945945945945947e-06,
      "loss": 1.1568,
      "step": 540
    },
    {
      "epoch": 1.65,
      "grad_norm": 15.009241104125977,
      "learning_rate": 9.944944944944946e-06,
      "loss": 1.0899,
      "step": 550
    },
    {
      "epoch": 1.68,
      "grad_norm": 21.914396286010742,
      "learning_rate": 9.943943943943944e-06,
      "loss": 0.9945,
      "step": 560
    },
    {
      "epoch": 1.71,
      "grad_norm": 15.895747184753418,
      "learning_rate": 9.942942942942944e-06,
      "loss": 1.1402,
      "step": 570
    },
    {
      "epoch": 1.74,
      "grad_norm": 22.441926956176758,
      "learning_rate": 9.941941941941943e-06,
      "loss": 1.0905,
      "step": 580
    },
    {
      "epoch": 1.77,
      "grad_norm": 16.437023162841797,
      "learning_rate": 9.940940940940942e-06,
      "loss": 1.0368,
      "step": 590
    },
    {
      "epoch": 1.8,
      "grad_norm": 12.452149391174316,
      "learning_rate": 9.93993993993994e-06,
      "loss": 1.1294,
      "step": 600
    },
    {
      "epoch": 1.83,
      "grad_norm": 50.94220733642578,
      "learning_rate": 9.93893893893894e-06,
      "loss": 1.0648,
      "step": 610
    },
    {
      "epoch": 1.86,
      "grad_norm": 22.72955322265625,
      "learning_rate": 9.937937937937939e-06,
      "loss": 1.0418,
      "step": 620
    },
    {
      "epoch": 1.89,
      "grad_norm": 15.258744239807129,
      "learning_rate": 9.936936936936938e-06,
      "loss": 1.0291,
      "step": 630
    },
    {
      "epoch": 1.92,
      "grad_norm": 13.402735710144043,
      "learning_rate": 9.935935935935936e-06,
      "loss": 1.0124,
      "step": 640
    },
    {
      "epoch": 1.95,
      "grad_norm": 19.616920471191406,
      "learning_rate": 9.934934934934935e-06,
      "loss": 1.065,
      "step": 650
    },
    {
      "epoch": 1.98,
      "grad_norm": 14.682768821716309,
      "learning_rate": 9.933933933933935e-06,
      "loss": 1.1153,
      "step": 660
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8751,
      "eval_loss": 0.4367116093635559,
      "eval_runtime": 30.112,
      "eval_samples_per_second": 332.093,
      "eval_steps_per_second": 1.328,
      "step": 666
    },
    {
      "epoch": 2.01,
      "grad_norm": 14.608451843261719,
      "learning_rate": 9.932932932932934e-06,
      "loss": 1.0625,
      "step": 670
    },
    {
      "epoch": 2.04,
      "grad_norm": 18.75455093383789,
      "learning_rate": 9.931931931931932e-06,
      "loss": 1.0292,
      "step": 680
    },
    {
      "epoch": 2.07,
      "grad_norm": 12.429720878601074,
      "learning_rate": 9.930930930930933e-06,
      "loss": 1.009,
      "step": 690
    },
    {
      "epoch": 2.1,
      "grad_norm": 15.66808032989502,
      "learning_rate": 9.929929929929931e-06,
      "loss": 0.9994,
      "step": 700
    },
    {
      "epoch": 2.13,
      "grad_norm": 20.903182983398438,
      "learning_rate": 9.92892892892893e-06,
      "loss": 0.9621,
      "step": 710
    },
    {
      "epoch": 2.16,
      "grad_norm": 10.713549613952637,
      "learning_rate": 9.927927927927928e-06,
      "loss": 0.9526,
      "step": 720
    },
    {
      "epoch": 2.19,
      "grad_norm": 19.999610900878906,
      "learning_rate": 9.926926926926927e-06,
      "loss": 1.0163,
      "step": 730
    },
    {
      "epoch": 2.22,
      "grad_norm": 10.432254791259766,
      "learning_rate": 9.925925925925927e-06,
      "loss": 1.0134,
      "step": 740
    },
    {
      "epoch": 2.25,
      "grad_norm": 9.605545043945312,
      "learning_rate": 9.924924924924926e-06,
      "loss": 0.9707,
      "step": 750
    },
    {
      "epoch": 2.28,
      "grad_norm": 17.136886596679688,
      "learning_rate": 9.923923923923925e-06,
      "loss": 0.9925,
      "step": 760
    },
    {
      "epoch": 2.31,
      "grad_norm": 10.676803588867188,
      "learning_rate": 9.922922922922925e-06,
      "loss": 0.9082,
      "step": 770
    },
    {
      "epoch": 2.34,
      "grad_norm": 13.845260620117188,
      "learning_rate": 9.921921921921923e-06,
      "loss": 0.891,
      "step": 780
    },
    {
      "epoch": 2.37,
      "grad_norm": 13.065056800842285,
      "learning_rate": 9.920920920920922e-06,
      "loss": 0.8955,
      "step": 790
    },
    {
      "epoch": 2.4,
      "grad_norm": 10.908167839050293,
      "learning_rate": 9.91991991991992e-06,
      "loss": 0.9711,
      "step": 800
    },
    {
      "epoch": 2.43,
      "grad_norm": 17.723506927490234,
      "learning_rate": 9.91891891891892e-06,
      "loss": 0.9859,
      "step": 810
    },
    {
      "epoch": 2.46,
      "grad_norm": 12.879526138305664,
      "learning_rate": 9.917917917917918e-06,
      "loss": 0.9318,
      "step": 820
    },
    {
      "epoch": 2.49,
      "grad_norm": 11.543569564819336,
      "learning_rate": 9.916916916916918e-06,
      "loss": 0.9445,
      "step": 830
    },
    {
      "epoch": 2.52,
      "grad_norm": 13.50468635559082,
      "learning_rate": 9.915915915915917e-06,
      "loss": 0.9331,
      "step": 840
    },
    {
      "epoch": 2.55,
      "grad_norm": 27.5030574798584,
      "learning_rate": 9.914914914914915e-06,
      "loss": 1.0073,
      "step": 850
    },
    {
      "epoch": 2.58,
      "grad_norm": 17.51815414428711,
      "learning_rate": 9.913913913913916e-06,
      "loss": 0.8938,
      "step": 860
    },
    {
      "epoch": 2.61,
      "grad_norm": 14.364579200744629,
      "learning_rate": 9.912912912912914e-06,
      "loss": 0.9455,
      "step": 870
    },
    {
      "epoch": 2.64,
      "grad_norm": 15.077325820922852,
      "learning_rate": 9.911911911911913e-06,
      "loss": 0.9742,
      "step": 880
    },
    {
      "epoch": 2.67,
      "grad_norm": 13.445960998535156,
      "learning_rate": 9.910910910910911e-06,
      "loss": 0.9341,
      "step": 890
    },
    {
      "epoch": 2.7,
      "grad_norm": 12.205794334411621,
      "learning_rate": 9.90990990990991e-06,
      "loss": 0.9009,
      "step": 900
    },
    {
      "epoch": 2.73,
      "grad_norm": 17.600419998168945,
      "learning_rate": 9.90890890890891e-06,
      "loss": 0.9779,
      "step": 910
    },
    {
      "epoch": 2.76,
      "grad_norm": 13.027902603149414,
      "learning_rate": 9.907907907907909e-06,
      "loss": 0.9338,
      "step": 920
    },
    {
      "epoch": 2.79,
      "grad_norm": 14.624979972839355,
      "learning_rate": 9.906906906906907e-06,
      "loss": 0.9067,
      "step": 930
    },
    {
      "epoch": 2.82,
      "grad_norm": 14.892926216125488,
      "learning_rate": 9.905905905905908e-06,
      "loss": 0.8548,
      "step": 940
    },
    {
      "epoch": 2.85,
      "grad_norm": 20.055511474609375,
      "learning_rate": 9.904904904904906e-06,
      "loss": 0.8416,
      "step": 950
    },
    {
      "epoch": 2.88,
      "grad_norm": 15.868443489074707,
      "learning_rate": 9.903903903903905e-06,
      "loss": 0.9291,
      "step": 960
    },
    {
      "epoch": 2.91,
      "grad_norm": 17.575693130493164,
      "learning_rate": 9.902902902902903e-06,
      "loss": 0.9228,
      "step": 970
    },
    {
      "epoch": 2.94,
      "grad_norm": 12.875110626220703,
      "learning_rate": 9.901901901901902e-06,
      "loss": 0.8616,
      "step": 980
    },
    {
      "epoch": 2.97,
      "grad_norm": 39.7818603515625,
      "learning_rate": 9.900900900900902e-06,
      "loss": 0.9219,
      "step": 990
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8946,
      "eval_loss": 0.35464411973953247,
      "eval_runtime": 30.462,
      "eval_samples_per_second": 328.278,
      "eval_steps_per_second": 1.313,
      "step": 999
    },
    {
      "epoch": 3.0,
      "grad_norm": 9.457967758178711,
      "learning_rate": 9.899899899899901e-06,
      "loss": 0.8197,
      "step": 1000
    },
    {
      "epoch": 3.03,
      "grad_norm": 11.913254737854004,
      "learning_rate": 9.8988988988989e-06,
      "loss": 0.8704,
      "step": 1010
    },
    {
      "epoch": 3.06,
      "grad_norm": 46.069637298583984,
      "learning_rate": 9.8978978978979e-06,
      "loss": 0.8304,
      "step": 1020
    },
    {
      "epoch": 3.09,
      "grad_norm": 14.785714149475098,
      "learning_rate": 9.896896896896898e-06,
      "loss": 0.8993,
      "step": 1030
    },
    {
      "epoch": 3.12,
      "grad_norm": 9.831999778747559,
      "learning_rate": 9.895895895895895e-06,
      "loss": 0.8093,
      "step": 1040
    },
    {
      "epoch": 3.15,
      "grad_norm": 10.466429710388184,
      "learning_rate": 9.894894894894896e-06,
      "loss": 0.9325,
      "step": 1050
    },
    {
      "epoch": 3.18,
      "grad_norm": 12.368927955627441,
      "learning_rate": 9.893893893893894e-06,
      "loss": 0.8663,
      "step": 1060
    },
    {
      "epoch": 3.21,
      "grad_norm": 12.843910217285156,
      "learning_rate": 9.892892892892893e-06,
      "loss": 0.7992,
      "step": 1070
    },
    {
      "epoch": 3.24,
      "grad_norm": 10.470433235168457,
      "learning_rate": 9.891891891891893e-06,
      "loss": 0.8336,
      "step": 1080
    },
    {
      "epoch": 3.27,
      "grad_norm": 11.602510452270508,
      "learning_rate": 9.890890890890892e-06,
      "loss": 0.8936,
      "step": 1090
    },
    {
      "epoch": 3.3,
      "grad_norm": 38.957767486572266,
      "learning_rate": 9.88988988988989e-06,
      "loss": 0.8651,
      "step": 1100
    },
    {
      "epoch": 3.33,
      "grad_norm": 9.675531387329102,
      "learning_rate": 9.88888888888889e-06,
      "loss": 0.8244,
      "step": 1110
    },
    {
      "epoch": 3.36,
      "grad_norm": 15.052042961120605,
      "learning_rate": 9.88788788788789e-06,
      "loss": 0.86,
      "step": 1120
    },
    {
      "epoch": 3.39,
      "grad_norm": 13.488245964050293,
      "learning_rate": 9.886886886886888e-06,
      "loss": 0.8673,
      "step": 1130
    },
    {
      "epoch": 3.42,
      "grad_norm": 14.295967102050781,
      "learning_rate": 9.885885885885886e-06,
      "loss": 0.9099,
      "step": 1140
    },
    {
      "epoch": 3.45,
      "grad_norm": 12.72698974609375,
      "learning_rate": 9.884884884884885e-06,
      "loss": 0.8992,
      "step": 1150
    },
    {
      "epoch": 3.48,
      "grad_norm": 19.24166488647461,
      "learning_rate": 9.883883883883885e-06,
      "loss": 0.9266,
      "step": 1160
    },
    {
      "epoch": 3.51,
      "grad_norm": 15.552074432373047,
      "learning_rate": 9.882882882882884e-06,
      "loss": 0.8785,
      "step": 1170
    },
    {
      "epoch": 3.54,
      "grad_norm": 89.45469665527344,
      "learning_rate": 9.881881881881882e-06,
      "loss": 0.8496,
      "step": 1180
    },
    {
      "epoch": 3.57,
      "grad_norm": 36.776451110839844,
      "learning_rate": 9.880880880880883e-06,
      "loss": 0.7665,
      "step": 1190
    },
    {
      "epoch": 3.6,
      "grad_norm": 12.820813179016113,
      "learning_rate": 9.879879879879881e-06,
      "loss": 0.7873,
      "step": 1200
    },
    {
      "epoch": 3.63,
      "grad_norm": 32.781124114990234,
      "learning_rate": 9.87887887887888e-06,
      "loss": 0.8752,
      "step": 1210
    },
    {
      "epoch": 3.66,
      "grad_norm": 11.789835929870605,
      "learning_rate": 9.877877877877879e-06,
      "loss": 0.844,
      "step": 1220
    },
    {
      "epoch": 3.69,
      "grad_norm": 11.543076515197754,
      "learning_rate": 9.876876876876877e-06,
      "loss": 0.8588,
      "step": 1230
    },
    {
      "epoch": 3.72,
      "grad_norm": 12.701692581176758,
      "learning_rate": 9.875875875875877e-06,
      "loss": 0.833,
      "step": 1240
    },
    {
      "epoch": 3.75,
      "grad_norm": 14.432270050048828,
      "learning_rate": 9.874874874874876e-06,
      "loss": 0.821,
      "step": 1250
    },
    {
      "epoch": 3.78,
      "grad_norm": 12.727407455444336,
      "learning_rate": 9.873873873873875e-06,
      "loss": 0.8628,
      "step": 1260
    },
    {
      "epoch": 3.81,
      "grad_norm": 15.466862678527832,
      "learning_rate": 9.872872872872873e-06,
      "loss": 0.7927,
      "step": 1270
    },
    {
      "epoch": 3.84,
      "grad_norm": 10.200054168701172,
      "learning_rate": 9.871871871871873e-06,
      "loss": 0.7891,
      "step": 1280
    },
    {
      "epoch": 3.87,
      "grad_norm": 13.369035720825195,
      "learning_rate": 9.87087087087087e-06,
      "loss": 0.8204,
      "step": 1290
    },
    {
      "epoch": 3.9,
      "grad_norm": 25.642301559448242,
      "learning_rate": 9.86986986986987e-06,
      "loss": 0.8128,
      "step": 1300
    },
    {
      "epoch": 3.93,
      "grad_norm": 8.70489501953125,
      "learning_rate": 9.86886886886887e-06,
      "loss": 0.8449,
      "step": 1310
    },
    {
      "epoch": 3.96,
      "grad_norm": 13.265392303466797,
      "learning_rate": 9.867867867867868e-06,
      "loss": 0.8596,
      "step": 1320
    },
    {
      "epoch": 3.99,
      "grad_norm": 13.94210147857666,
      "learning_rate": 9.866866866866868e-06,
      "loss": 0.8802,
      "step": 1330
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.9027,
      "eval_loss": 0.3215944468975067,
      "eval_runtime": 30.2535,
      "eval_samples_per_second": 330.54,
      "eval_steps_per_second": 1.322,
      "step": 1332
    },
    {
      "epoch": 4.02,
      "grad_norm": 11.947212219238281,
      "learning_rate": 9.865865865865867e-06,
      "loss": 0.863,
      "step": 1340
    },
    {
      "epoch": 4.05,
      "grad_norm": 19.2293758392334,
      "learning_rate": 9.864864864864865e-06,
      "loss": 0.787,
      "step": 1350
    },
    {
      "epoch": 4.08,
      "grad_norm": 9.49066162109375,
      "learning_rate": 9.863863863863866e-06,
      "loss": 0.799,
      "step": 1360
    },
    {
      "epoch": 4.11,
      "grad_norm": 18.955677032470703,
      "learning_rate": 9.862862862862864e-06,
      "loss": 0.7656,
      "step": 1370
    },
    {
      "epoch": 4.14,
      "grad_norm": 13.782297134399414,
      "learning_rate": 9.861861861861863e-06,
      "loss": 0.8222,
      "step": 1380
    },
    {
      "epoch": 4.17,
      "grad_norm": 10.21755599975586,
      "learning_rate": 9.860860860860861e-06,
      "loss": 0.8259,
      "step": 1390
    },
    {
      "epoch": 4.2,
      "grad_norm": 32.93952941894531,
      "learning_rate": 9.85985985985986e-06,
      "loss": 0.7729,
      "step": 1400
    },
    {
      "epoch": 4.23,
      "grad_norm": 12.50378704071045,
      "learning_rate": 9.85885885885886e-06,
      "loss": 0.8319,
      "step": 1410
    },
    {
      "epoch": 4.26,
      "grad_norm": 13.975879669189453,
      "learning_rate": 9.857857857857859e-06,
      "loss": 0.8158,
      "step": 1420
    },
    {
      "epoch": 4.29,
      "grad_norm": 13.557897567749023,
      "learning_rate": 9.856856856856857e-06,
      "loss": 0.8521,
      "step": 1430
    },
    {
      "epoch": 4.32,
      "grad_norm": 16.076936721801758,
      "learning_rate": 9.855855855855858e-06,
      "loss": 0.787,
      "step": 1440
    },
    {
      "epoch": 4.35,
      "grad_norm": 13.304807662963867,
      "learning_rate": 9.854854854854856e-06,
      "loss": 0.7525,
      "step": 1450
    },
    {
      "epoch": 4.38,
      "grad_norm": 9.973689079284668,
      "learning_rate": 9.853853853853855e-06,
      "loss": 0.8127,
      "step": 1460
    },
    {
      "epoch": 4.41,
      "grad_norm": 11.573305130004883,
      "learning_rate": 9.852852852852854e-06,
      "loss": 0.7792,
      "step": 1470
    },
    {
      "epoch": 4.44,
      "grad_norm": 10.37661361694336,
      "learning_rate": 9.851851851851852e-06,
      "loss": 0.7511,
      "step": 1480
    },
    {
      "epoch": 4.47,
      "grad_norm": 14.020105361938477,
      "learning_rate": 9.85085085085085e-06,
      "loss": 0.7556,
      "step": 1490
    },
    {
      "epoch": 4.5,
      "grad_norm": 13.901347160339355,
      "learning_rate": 9.849849849849851e-06,
      "loss": 0.788,
      "step": 1500
    },
    {
      "epoch": 4.53,
      "grad_norm": 14.113587379455566,
      "learning_rate": 9.84884884884885e-06,
      "loss": 0.7675,
      "step": 1510
    },
    {
      "epoch": 4.56,
      "grad_norm": 10.88308048248291,
      "learning_rate": 9.847847847847848e-06,
      "loss": 0.6993,
      "step": 1520
    },
    {
      "epoch": 4.59,
      "grad_norm": 12.067638397216797,
      "learning_rate": 9.846846846846849e-06,
      "loss": 0.8427,
      "step": 1530
    },
    {
      "epoch": 4.62,
      "grad_norm": 22.864439010620117,
      "learning_rate": 9.845845845845845e-06,
      "loss": 0.7892,
      "step": 1540
    },
    {
      "epoch": 4.65,
      "grad_norm": 83.34579467773438,
      "learning_rate": 9.844844844844846e-06,
      "loss": 0.7976,
      "step": 1550
    },
    {
      "epoch": 4.68,
      "grad_norm": 20.661500930786133,
      "learning_rate": 9.843843843843844e-06,
      "loss": 0.7091,
      "step": 1560
    },
    {
      "epoch": 4.71,
      "grad_norm": 16.65585708618164,
      "learning_rate": 9.842842842842843e-06,
      "loss": 0.7744,
      "step": 1570
    },
    {
      "epoch": 4.74,
      "grad_norm": 11.51008129119873,
      "learning_rate": 9.841841841841843e-06,
      "loss": 0.7727,
      "step": 1580
    },
    {
      "epoch": 4.77,
      "grad_norm": 8.962791442871094,
      "learning_rate": 9.840840840840842e-06,
      "loss": 0.8176,
      "step": 1590
    },
    {
      "epoch": 4.8,
      "grad_norm": 12.087862014770508,
      "learning_rate": 9.83983983983984e-06,
      "loss": 0.8206,
      "step": 1600
    },
    {
      "epoch": 4.83,
      "grad_norm": 13.356203079223633,
      "learning_rate": 9.83883883883884e-06,
      "loss": 0.7516,
      "step": 1610
    },
    {
      "epoch": 4.86,
      "grad_norm": 11.307415008544922,
      "learning_rate": 9.83783783783784e-06,
      "loss": 0.8181,
      "step": 1620
    },
    {
      "epoch": 4.89,
      "grad_norm": 13.581600189208984,
      "learning_rate": 9.836836836836838e-06,
      "loss": 0.8334,
      "step": 1630
    },
    {
      "epoch": 4.92,
      "grad_norm": 12.420446395874023,
      "learning_rate": 9.835835835835836e-06,
      "loss": 0.774,
      "step": 1640
    },
    {
      "epoch": 4.95,
      "grad_norm": 20.350292205810547,
      "learning_rate": 9.834834834834835e-06,
      "loss": 0.8007,
      "step": 1650
    },
    {
      "epoch": 4.98,
      "grad_norm": 11.960819244384766,
      "learning_rate": 9.833833833833835e-06,
      "loss": 0.7332,
      "step": 1660
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.9069,
      "eval_loss": 0.3024939298629761,
      "eval_runtime": 30.2033,
      "eval_samples_per_second": 331.09,
      "eval_steps_per_second": 1.324,
      "step": 1665
    },
    {
      "epoch": 5.02,
      "grad_norm": 14.111451148986816,
      "learning_rate": 9.832832832832834e-06,
      "loss": 0.6782,
      "step": 1670
    },
    {
      "epoch": 5.05,
      "grad_norm": 19.54442596435547,
      "learning_rate": 9.831831831831833e-06,
      "loss": 0.6969,
      "step": 1680
    },
    {
      "epoch": 5.08,
      "grad_norm": 16.62184715270996,
      "learning_rate": 9.830830830830833e-06,
      "loss": 0.7776,
      "step": 1690
    },
    {
      "epoch": 5.11,
      "grad_norm": 10.560216903686523,
      "learning_rate": 9.829829829829831e-06,
      "loss": 0.7622,
      "step": 1700
    },
    {
      "epoch": 5.14,
      "grad_norm": 14.3679780960083,
      "learning_rate": 9.82882882882883e-06,
      "loss": 0.7208,
      "step": 1710
    },
    {
      "epoch": 5.17,
      "grad_norm": 15.207982063293457,
      "learning_rate": 9.827827827827829e-06,
      "loss": 0.7025,
      "step": 1720
    },
    {
      "epoch": 5.2,
      "grad_norm": 9.77590560913086,
      "learning_rate": 9.826826826826827e-06,
      "loss": 0.7776,
      "step": 1730
    },
    {
      "epoch": 5.23,
      "grad_norm": 18.650480270385742,
      "learning_rate": 9.825825825825826e-06,
      "loss": 0.8204,
      "step": 1740
    },
    {
      "epoch": 5.26,
      "grad_norm": 20.201152801513672,
      "learning_rate": 9.824824824824826e-06,
      "loss": 0.6941,
      "step": 1750
    },
    {
      "epoch": 5.29,
      "grad_norm": 12.634096145629883,
      "learning_rate": 9.823823823823825e-06,
      "loss": 0.7537,
      "step": 1760
    },
    {
      "epoch": 5.32,
      "grad_norm": 14.861948013305664,
      "learning_rate": 9.822822822822823e-06,
      "loss": 0.7318,
      "step": 1770
    },
    {
      "epoch": 5.35,
      "grad_norm": 17.764514923095703,
      "learning_rate": 9.821821821821824e-06,
      "loss": 0.7009,
      "step": 1780
    },
    {
      "epoch": 5.38,
      "grad_norm": 11.650490760803223,
      "learning_rate": 9.82082082082082e-06,
      "loss": 0.7511,
      "step": 1790
    },
    {
      "epoch": 5.41,
      "grad_norm": 12.79879093170166,
      "learning_rate": 9.81981981981982e-06,
      "loss": 0.777,
      "step": 1800
    },
    {
      "epoch": 5.44,
      "grad_norm": 12.410004615783691,
      "learning_rate": 9.81881881881882e-06,
      "loss": 0.7176,
      "step": 1810
    },
    {
      "epoch": 5.47,
      "grad_norm": 19.437084197998047,
      "learning_rate": 9.817817817817818e-06,
      "loss": 0.8047,
      "step": 1820
    },
    {
      "epoch": 5.5,
      "grad_norm": 17.08942413330078,
      "learning_rate": 9.816816816816818e-06,
      "loss": 0.7177,
      "step": 1830
    },
    {
      "epoch": 5.53,
      "grad_norm": 15.615683555603027,
      "learning_rate": 9.815815815815817e-06,
      "loss": 0.6848,
      "step": 1840
    },
    {
      "epoch": 5.56,
      "grad_norm": 13.829546928405762,
      "learning_rate": 9.814814814814815e-06,
      "loss": 0.7703,
      "step": 1850
    },
    {
      "epoch": 5.59,
      "grad_norm": 10.06242847442627,
      "learning_rate": 9.813813813813816e-06,
      "loss": 0.6813,
      "step": 1860
    },
    {
      "epoch": 5.62,
      "grad_norm": 11.726490020751953,
      "learning_rate": 9.812812812812814e-06,
      "loss": 0.749,
      "step": 1870
    },
    {
      "epoch": 5.65,
      "grad_norm": 9.272887229919434,
      "learning_rate": 9.811811811811813e-06,
      "loss": 0.7128,
      "step": 1880
    },
    {
      "epoch": 5.68,
      "grad_norm": 12.022890090942383,
      "learning_rate": 9.810810810810811e-06,
      "loss": 0.6505,
      "step": 1890
    },
    {
      "epoch": 5.71,
      "grad_norm": 11.16840934753418,
      "learning_rate": 9.80980980980981e-06,
      "loss": 0.7784,
      "step": 1900
    },
    {
      "epoch": 5.74,
      "grad_norm": 16.869796752929688,
      "learning_rate": 9.80880880880881e-06,
      "loss": 0.7459,
      "step": 1910
    },
    {
      "epoch": 5.77,
      "grad_norm": 12.084444999694824,
      "learning_rate": 9.807807807807809e-06,
      "loss": 0.7068,
      "step": 1920
    },
    {
      "epoch": 5.8,
      "grad_norm": 11.130463600158691,
      "learning_rate": 9.806806806806808e-06,
      "loss": 0.7896,
      "step": 1930
    },
    {
      "epoch": 5.83,
      "grad_norm": 20.33501625061035,
      "learning_rate": 9.805805805805808e-06,
      "loss": 0.7212,
      "step": 1940
    },
    {
      "epoch": 5.86,
      "grad_norm": 14.124403953552246,
      "learning_rate": 9.804804804804806e-06,
      "loss": 0.7743,
      "step": 1950
    },
    {
      "epoch": 5.89,
      "grad_norm": 13.767190933227539,
      "learning_rate": 9.803803803803803e-06,
      "loss": 0.7661,
      "step": 1960
    },
    {
      "epoch": 5.92,
      "grad_norm": 14.234041213989258,
      "learning_rate": 9.802802802802804e-06,
      "loss": 0.7038,
      "step": 1970
    },
    {
      "epoch": 5.95,
      "grad_norm": 16.750995635986328,
      "learning_rate": 9.801801801801802e-06,
      "loss": 0.7367,
      "step": 1980
    },
    {
      "epoch": 5.98,
      "grad_norm": 49.59062194824219,
      "learning_rate": 9.8008008008008e-06,
      "loss": 0.6719,
      "step": 1990
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.9098,
      "eval_loss": 0.2940322756767273,
      "eval_runtime": 30.4626,
      "eval_samples_per_second": 328.271,
      "eval_steps_per_second": 1.313,
      "step": 1998
    },
    {
      "epoch": 6.01,
      "grad_norm": 12.728262901306152,
      "learning_rate": 9.799799799799801e-06,
      "loss": 0.9003,
      "step": 2000
    },
    {
      "epoch": 6.04,
      "grad_norm": 13.666450500488281,
      "learning_rate": 9.7987987987988e-06,
      "loss": 0.7714,
      "step": 2010
    },
    {
      "epoch": 6.07,
      "grad_norm": 9.260753631591797,
      "learning_rate": 9.797797797797798e-06,
      "loss": 0.6515,
      "step": 2020
    },
    {
      "epoch": 6.1,
      "grad_norm": 15.711074829101562,
      "learning_rate": 9.796796796796799e-06,
      "loss": 0.6656,
      "step": 2030
    },
    {
      "epoch": 6.13,
      "grad_norm": 39.45460891723633,
      "learning_rate": 9.795795795795795e-06,
      "loss": 0.6897,
      "step": 2040
    },
    {
      "epoch": 6.16,
      "grad_norm": 12.101982116699219,
      "learning_rate": 9.794794794794796e-06,
      "loss": 0.7077,
      "step": 2050
    },
    {
      "epoch": 6.19,
      "grad_norm": 11.528532028198242,
      "learning_rate": 9.793793793793794e-06,
      "loss": 0.7468,
      "step": 2060
    },
    {
      "epoch": 6.22,
      "grad_norm": 11.694806098937988,
      "learning_rate": 9.792792792792793e-06,
      "loss": 0.7502,
      "step": 2070
    },
    {
      "epoch": 6.25,
      "grad_norm": 8.30628490447998,
      "learning_rate": 9.791791791791793e-06,
      "loss": 0.6735,
      "step": 2080
    },
    {
      "epoch": 6.28,
      "grad_norm": 9.299217224121094,
      "learning_rate": 9.790790790790792e-06,
      "loss": 0.704,
      "step": 2090
    },
    {
      "epoch": 6.31,
      "grad_norm": 10.323074340820312,
      "learning_rate": 9.78978978978979e-06,
      "loss": 0.7196,
      "step": 2100
    },
    {
      "epoch": 6.34,
      "grad_norm": 17.049901962280273,
      "learning_rate": 9.78878878878879e-06,
      "loss": 0.71,
      "step": 2110
    },
    {
      "epoch": 6.37,
      "grad_norm": 13.586891174316406,
      "learning_rate": 9.787787787787788e-06,
      "loss": 0.7129,
      "step": 2120
    },
    {
      "epoch": 6.4,
      "grad_norm": 18.899734497070312,
      "learning_rate": 9.786786786786788e-06,
      "loss": 0.7571,
      "step": 2130
    },
    {
      "epoch": 6.43,
      "grad_norm": 10.381853103637695,
      "learning_rate": 9.785785785785787e-06,
      "loss": 0.6343,
      "step": 2140
    },
    {
      "epoch": 6.46,
      "grad_norm": 14.553983688354492,
      "learning_rate": 9.784784784784785e-06,
      "loss": 0.7838,
      "step": 2150
    },
    {
      "epoch": 6.49,
      "grad_norm": 15.92480182647705,
      "learning_rate": 9.783783783783785e-06,
      "loss": 0.7708,
      "step": 2160
    },
    {
      "epoch": 6.52,
      "grad_norm": 11.899249076843262,
      "learning_rate": 9.782782782782784e-06,
      "loss": 0.7326,
      "step": 2170
    },
    {
      "epoch": 6.55,
      "grad_norm": 14.936653137207031,
      "learning_rate": 9.781781781781783e-06,
      "loss": 0.7262,
      "step": 2180
    },
    {
      "epoch": 6.58,
      "grad_norm": 15.01948070526123,
      "learning_rate": 9.780780780780781e-06,
      "loss": 0.6918,
      "step": 2190
    },
    {
      "epoch": 6.61,
      "grad_norm": 12.893462181091309,
      "learning_rate": 9.779779779779781e-06,
      "loss": 0.7247,
      "step": 2200
    },
    {
      "epoch": 6.64,
      "grad_norm": 22.485645294189453,
      "learning_rate": 9.778778778778778e-06,
      "loss": 0.7185,
      "step": 2210
    },
    {
      "epoch": 6.67,
      "grad_norm": 10.820356369018555,
      "learning_rate": 9.777777777777779e-06,
      "loss": 0.7147,
      "step": 2220
    },
    {
      "epoch": 6.7,
      "grad_norm": 11.032286643981934,
      "learning_rate": 9.776776776776777e-06,
      "loss": 0.7425,
      "step": 2230
    },
    {
      "epoch": 6.73,
      "grad_norm": 11.431447982788086,
      "learning_rate": 9.775775775775776e-06,
      "loss": 0.7005,
      "step": 2240
    },
    {
      "epoch": 6.76,
      "grad_norm": 20.35511016845703,
      "learning_rate": 9.774774774774776e-06,
      "loss": 0.5943,
      "step": 2250
    },
    {
      "epoch": 6.79,
      "grad_norm": 13.538642883300781,
      "learning_rate": 9.773773773773775e-06,
      "loss": 0.6561,
      "step": 2260
    },
    {
      "epoch": 6.82,
      "grad_norm": 10.554620742797852,
      "learning_rate": 9.772772772772773e-06,
      "loss": 0.6407,
      "step": 2270
    },
    {
      "epoch": 6.85,
      "grad_norm": 13.718347549438477,
      "learning_rate": 9.771771771771774e-06,
      "loss": 0.7527,
      "step": 2280
    },
    {
      "epoch": 6.88,
      "grad_norm": 12.707255363464355,
      "learning_rate": 9.77077077077077e-06,
      "loss": 0.6908,
      "step": 2290
    },
    {
      "epoch": 6.91,
      "grad_norm": 14.473544120788574,
      "learning_rate": 9.76976976976977e-06,
      "loss": 0.6696,
      "step": 2300
    },
    {
      "epoch": 6.94,
      "grad_norm": 13.07099437713623,
      "learning_rate": 9.76876876876877e-06,
      "loss": 0.7644,
      "step": 2310
    },
    {
      "epoch": 6.97,
      "grad_norm": 41.19286346435547,
      "learning_rate": 9.767767767767768e-06,
      "loss": 0.7138,
      "step": 2320
    },
    {
      "epoch": 7.0,
      "grad_norm": 9.618319511413574,
      "learning_rate": 9.766766766766768e-06,
      "loss": 0.6996,
      "step": 2330
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.9141,
      "eval_loss": 0.27847668528556824,
      "eval_runtime": 30.5293,
      "eval_samples_per_second": 327.554,
      "eval_steps_per_second": 1.31,
      "step": 2331
    },
    {
      "epoch": 7.03,
      "grad_norm": 10.23026180267334,
      "learning_rate": 9.765765765765767e-06,
      "loss": 0.6614,
      "step": 2340
    },
    {
      "epoch": 7.06,
      "grad_norm": 14.503907203674316,
      "learning_rate": 9.764764764764765e-06,
      "loss": 0.7116,
      "step": 2350
    },
    {
      "epoch": 7.09,
      "grad_norm": 11.488687515258789,
      "learning_rate": 9.763763763763766e-06,
      "loss": 0.7243,
      "step": 2360
    },
    {
      "epoch": 7.12,
      "grad_norm": 10.227832794189453,
      "learning_rate": 9.762762762762763e-06,
      "loss": 0.5943,
      "step": 2370
    },
    {
      "epoch": 7.15,
      "grad_norm": 11.515847206115723,
      "learning_rate": 9.761761761761763e-06,
      "loss": 0.6871,
      "step": 2380
    },
    {
      "epoch": 7.18,
      "grad_norm": 17.236072540283203,
      "learning_rate": 9.760760760760762e-06,
      "loss": 0.72,
      "step": 2390
    },
    {
      "epoch": 7.21,
      "grad_norm": 13.397762298583984,
      "learning_rate": 9.75975975975976e-06,
      "loss": 0.7651,
      "step": 2400
    },
    {
      "epoch": 7.24,
      "grad_norm": 12.64128303527832,
      "learning_rate": 9.758758758758759e-06,
      "loss": 0.632,
      "step": 2410
    },
    {
      "epoch": 7.27,
      "grad_norm": 28.24812126159668,
      "learning_rate": 9.757757757757759e-06,
      "loss": 0.6373,
      "step": 2420
    },
    {
      "epoch": 7.3,
      "grad_norm": 12.245796203613281,
      "learning_rate": 9.756756756756758e-06,
      "loss": 0.6566,
      "step": 2430
    },
    {
      "epoch": 7.33,
      "grad_norm": 18.77831268310547,
      "learning_rate": 9.755755755755756e-06,
      "loss": 0.6115,
      "step": 2440
    },
    {
      "epoch": 7.36,
      "grad_norm": 13.072959899902344,
      "learning_rate": 9.754754754754756e-06,
      "loss": 0.6627,
      "step": 2450
    },
    {
      "epoch": 7.39,
      "grad_norm": 29.834144592285156,
      "learning_rate": 9.753753753753753e-06,
      "loss": 0.71,
      "step": 2460
    },
    {
      "epoch": 7.42,
      "grad_norm": 10.122661590576172,
      "learning_rate": 9.752752752752754e-06,
      "loss": 0.6729,
      "step": 2470
    },
    {
      "epoch": 7.45,
      "grad_norm": 11.016374588012695,
      "learning_rate": 9.751751751751752e-06,
      "loss": 0.7137,
      "step": 2480
    },
    {
      "epoch": 7.48,
      "grad_norm": 11.94205093383789,
      "learning_rate": 9.750750750750751e-06,
      "loss": 0.6615,
      "step": 2490
    },
    {
      "epoch": 7.51,
      "grad_norm": 27.569272994995117,
      "learning_rate": 9.749749749749751e-06,
      "loss": 0.6984,
      "step": 2500
    },
    {
      "epoch": 7.54,
      "grad_norm": 10.764726638793945,
      "learning_rate": 9.74874874874875e-06,
      "loss": 0.697,
      "step": 2510
    },
    {
      "epoch": 7.57,
      "grad_norm": 9.96368408203125,
      "learning_rate": 9.747747747747748e-06,
      "loss": 0.72,
      "step": 2520
    },
    {
      "epoch": 7.6,
      "grad_norm": 8.127985000610352,
      "learning_rate": 9.746746746746749e-06,
      "loss": 0.634,
      "step": 2530
    },
    {
      "epoch": 7.63,
      "grad_norm": 16.233617782592773,
      "learning_rate": 9.745745745745746e-06,
      "loss": 0.7103,
      "step": 2540
    },
    {
      "epoch": 7.66,
      "grad_norm": 11.445056915283203,
      "learning_rate": 9.744744744744746e-06,
      "loss": 0.6502,
      "step": 2550
    },
    {
      "epoch": 7.69,
      "grad_norm": 15.745110511779785,
      "learning_rate": 9.743743743743744e-06,
      "loss": 0.6258,
      "step": 2560
    },
    {
      "epoch": 7.72,
      "grad_norm": 11.728997230529785,
      "learning_rate": 9.742742742742743e-06,
      "loss": 0.7321,
      "step": 2570
    },
    {
      "epoch": 7.75,
      "grad_norm": 12.35846996307373,
      "learning_rate": 9.741741741741743e-06,
      "loss": 0.7286,
      "step": 2580
    },
    {
      "epoch": 7.78,
      "grad_norm": 11.954543113708496,
      "learning_rate": 9.740740740740742e-06,
      "loss": 0.6831,
      "step": 2590
    },
    {
      "epoch": 7.81,
      "grad_norm": 11.742406845092773,
      "learning_rate": 9.73973973973974e-06,
      "loss": 0.7106,
      "step": 2600
    },
    {
      "epoch": 7.84,
      "grad_norm": 9.084891319274902,
      "learning_rate": 9.73873873873874e-06,
      "loss": 0.6903,
      "step": 2610
    },
    {
      "epoch": 7.87,
      "grad_norm": 12.923911094665527,
      "learning_rate": 9.737737737737738e-06,
      "loss": 0.6308,
      "step": 2620
    },
    {
      "epoch": 7.9,
      "grad_norm": 9.698966979980469,
      "learning_rate": 9.736736736736738e-06,
      "loss": 0.693,
      "step": 2630
    },
    {
      "epoch": 7.93,
      "grad_norm": 11.315530776977539,
      "learning_rate": 9.735735735735737e-06,
      "loss": 0.6412,
      "step": 2640
    },
    {
      "epoch": 7.96,
      "grad_norm": 17.644512176513672,
      "learning_rate": 9.734734734734735e-06,
      "loss": 0.6525,
      "step": 2650
    },
    {
      "epoch": 7.99,
      "grad_norm": 32.29540252685547,
      "learning_rate": 9.733733733733734e-06,
      "loss": 0.7023,
      "step": 2660
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.9165,
      "eval_loss": 0.27294430136680603,
      "eval_runtime": 30.9418,
      "eval_samples_per_second": 323.187,
      "eval_steps_per_second": 1.293,
      "step": 2664
    },
    {
      "epoch": 8.02,
      "grad_norm": 11.768852233886719,
      "learning_rate": 9.732732732732734e-06,
      "loss": 0.5853,
      "step": 2670
    },
    {
      "epoch": 8.05,
      "grad_norm": 9.919044494628906,
      "learning_rate": 9.731731731731733e-06,
      "loss": 0.7103,
      "step": 2680
    },
    {
      "epoch": 8.08,
      "grad_norm": 10.119452476501465,
      "learning_rate": 9.730730730730731e-06,
      "loss": 0.6712,
      "step": 2690
    },
    {
      "epoch": 8.11,
      "grad_norm": 11.397403717041016,
      "learning_rate": 9.729729729729732e-06,
      "loss": 0.6385,
      "step": 2700
    },
    {
      "epoch": 8.14,
      "grad_norm": 16.2565975189209,
      "learning_rate": 9.728728728728728e-06,
      "loss": 0.6542,
      "step": 2710
    },
    {
      "epoch": 8.17,
      "grad_norm": 10.146173477172852,
      "learning_rate": 9.727727727727729e-06,
      "loss": 0.645,
      "step": 2720
    },
    {
      "epoch": 8.2,
      "grad_norm": 10.870752334594727,
      "learning_rate": 9.726726726726727e-06,
      "loss": 0.6539,
      "step": 2730
    },
    {
      "epoch": 8.23,
      "grad_norm": 11.892374992370605,
      "learning_rate": 9.725725725725726e-06,
      "loss": 0.6483,
      "step": 2740
    },
    {
      "epoch": 8.26,
      "grad_norm": 12.019987106323242,
      "learning_rate": 9.724724724724726e-06,
      "loss": 0.6794,
      "step": 2750
    },
    {
      "epoch": 8.29,
      "grad_norm": 10.963202476501465,
      "learning_rate": 9.723723723723725e-06,
      "loss": 0.6416,
      "step": 2760
    },
    {
      "epoch": 8.32,
      "grad_norm": 14.188977241516113,
      "learning_rate": 9.722722722722723e-06,
      "loss": 0.6006,
      "step": 2770
    },
    {
      "epoch": 8.35,
      "grad_norm": 11.994919776916504,
      "learning_rate": 9.721721721721724e-06,
      "loss": 0.6577,
      "step": 2780
    },
    {
      "epoch": 8.38,
      "grad_norm": 12.545328140258789,
      "learning_rate": 9.72072072072072e-06,
      "loss": 0.6337,
      "step": 2790
    },
    {
      "epoch": 8.41,
      "grad_norm": 16.774003982543945,
      "learning_rate": 9.719719719719721e-06,
      "loss": 0.6151,
      "step": 2800
    },
    {
      "epoch": 8.44,
      "grad_norm": 22.333213806152344,
      "learning_rate": 9.71871871871872e-06,
      "loss": 0.6167,
      "step": 2810
    },
    {
      "epoch": 8.47,
      "grad_norm": 13.755775451660156,
      "learning_rate": 9.717717717717718e-06,
      "loss": 0.6507,
      "step": 2820
    },
    {
      "epoch": 8.5,
      "grad_norm": 16.17780113220215,
      "learning_rate": 9.716716716716718e-06,
      "loss": 0.6653,
      "step": 2830
    },
    {
      "epoch": 8.53,
      "grad_norm": 11.606565475463867,
      "learning_rate": 9.715715715715717e-06,
      "loss": 0.637,
      "step": 2840
    },
    {
      "epoch": 8.56,
      "grad_norm": 21.008371353149414,
      "learning_rate": 9.714714714714716e-06,
      "loss": 0.6559,
      "step": 2850
    },
    {
      "epoch": 8.59,
      "grad_norm": 9.835844039916992,
      "learning_rate": 9.713713713713714e-06,
      "loss": 0.6633,
      "step": 2860
    },
    {
      "epoch": 8.62,
      "grad_norm": 12.223276138305664,
      "learning_rate": 9.712712712712713e-06,
      "loss": 0.6244,
      "step": 2870
    },
    {
      "epoch": 8.65,
      "grad_norm": 16.365243911743164,
      "learning_rate": 9.711711711711711e-06,
      "loss": 0.6177,
      "step": 2880
    },
    {
      "epoch": 8.68,
      "grad_norm": 21.39632797241211,
      "learning_rate": 9.710710710710712e-06,
      "loss": 0.6716,
      "step": 2890
    },
    {
      "epoch": 8.71,
      "grad_norm": 13.667511940002441,
      "learning_rate": 9.70970970970971e-06,
      "loss": 0.6434,
      "step": 2900
    },
    {
      "epoch": 8.74,
      "grad_norm": 11.143122673034668,
      "learning_rate": 9.708708708708709e-06,
      "loss": 0.6564,
      "step": 2910
    },
    {
      "epoch": 8.77,
      "grad_norm": 10.231430053710938,
      "learning_rate": 9.707707707707709e-06,
      "loss": 0.6212,
      "step": 2920
    },
    {
      "epoch": 8.8,
      "grad_norm": 11.799844741821289,
      "learning_rate": 9.706706706706708e-06,
      "loss": 0.5904,
      "step": 2930
    },
    {
      "epoch": 8.83,
      "grad_norm": 28.14351463317871,
      "learning_rate": 9.705705705705706e-06,
      "loss": 0.66,
      "step": 2940
    },
    {
      "epoch": 8.86,
      "grad_norm": 14.178098678588867,
      "learning_rate": 9.704704704704707e-06,
      "loss": 0.5597,
      "step": 2950
    },
    {
      "epoch": 8.89,
      "grad_norm": 12.042082786560059,
      "learning_rate": 9.703703703703703e-06,
      "loss": 0.6368,
      "step": 2960
    },
    {
      "epoch": 8.92,
      "grad_norm": 11.72601318359375,
      "learning_rate": 9.702702702702704e-06,
      "loss": 0.6435,
      "step": 2970
    },
    {
      "epoch": 8.95,
      "grad_norm": 16.047924041748047,
      "learning_rate": 9.701701701701702e-06,
      "loss": 0.6367,
      "step": 2980
    },
    {
      "epoch": 8.98,
      "grad_norm": 11.935233116149902,
      "learning_rate": 9.700700700700701e-06,
      "loss": 0.6376,
      "step": 2990
    },
    {
      "epoch": 9.0,
      "eval_accuracy": 0.9194,
      "eval_loss": 0.265708327293396,
      "eval_runtime": 30.5217,
      "eval_samples_per_second": 327.636,
      "eval_steps_per_second": 1.311,
      "step": 2997
    },
    {
      "epoch": 9.01,
      "grad_norm": 42.095699310302734,
      "learning_rate": 9.699699699699701e-06,
      "loss": 0.8317,
      "step": 3000
    },
    {
      "epoch": 9.04,
      "grad_norm": 13.209282875061035,
      "learning_rate": 9.6986986986987e-06,
      "loss": 0.6914,
      "step": 3010
    },
    {
      "epoch": 9.07,
      "grad_norm": 9.086323738098145,
      "learning_rate": 9.697697697697698e-06,
      "loss": 0.6362,
      "step": 3020
    },
    {
      "epoch": 9.1,
      "grad_norm": 9.95993423461914,
      "learning_rate": 9.696696696696699e-06,
      "loss": 0.6453,
      "step": 3030
    },
    {
      "epoch": 9.13,
      "grad_norm": 13.244100570678711,
      "learning_rate": 9.695695695695696e-06,
      "loss": 0.5516,
      "step": 3040
    },
    {
      "epoch": 9.16,
      "grad_norm": 11.685038566589355,
      "learning_rate": 9.694694694694696e-06,
      "loss": 0.6275,
      "step": 3050
    },
    {
      "epoch": 9.19,
      "grad_norm": 12.687440872192383,
      "learning_rate": 9.693693693693694e-06,
      "loss": 0.7,
      "step": 3060
    },
    {
      "epoch": 9.22,
      "grad_norm": 14.26815128326416,
      "learning_rate": 9.692692692692693e-06,
      "loss": 0.6488,
      "step": 3070
    },
    {
      "epoch": 9.25,
      "grad_norm": 12.68408203125,
      "learning_rate": 9.691691691691693e-06,
      "loss": 0.6669,
      "step": 3080
    },
    {
      "epoch": 9.28,
      "grad_norm": 19.84491729736328,
      "learning_rate": 9.690690690690692e-06,
      "loss": 0.5962,
      "step": 3090
    },
    {
      "epoch": 9.31,
      "grad_norm": 14.905065536499023,
      "learning_rate": 9.68968968968969e-06,
      "loss": 0.6467,
      "step": 3100
    },
    {
      "epoch": 9.34,
      "grad_norm": 18.167434692382812,
      "learning_rate": 9.68868868868869e-06,
      "loss": 0.6241,
      "step": 3110
    },
    {
      "epoch": 9.37,
      "grad_norm": 12.980717658996582,
      "learning_rate": 9.687687687687688e-06,
      "loss": 0.7141,
      "step": 3120
    },
    {
      "epoch": 9.4,
      "grad_norm": 13.490983009338379,
      "learning_rate": 9.686686686686686e-06,
      "loss": 0.6145,
      "step": 3130
    },
    {
      "epoch": 9.43,
      "grad_norm": 10.717024803161621,
      "learning_rate": 9.685685685685687e-06,
      "loss": 0.6809,
      "step": 3140
    },
    {
      "epoch": 9.46,
      "grad_norm": 11.203835487365723,
      "learning_rate": 9.684684684684685e-06,
      "loss": 0.6164,
      "step": 3150
    },
    {
      "epoch": 9.49,
      "grad_norm": 9.084880828857422,
      "learning_rate": 9.683683683683684e-06,
      "loss": 0.5897,
      "step": 3160
    },
    {
      "epoch": 9.52,
      "grad_norm": 15.867284774780273,
      "learning_rate": 9.682682682682684e-06,
      "loss": 0.5605,
      "step": 3170
    },
    {
      "epoch": 9.55,
      "grad_norm": 14.339218139648438,
      "learning_rate": 9.681681681681683e-06,
      "loss": 0.7164,
      "step": 3180
    },
    {
      "epoch": 9.58,
      "grad_norm": 11.700525283813477,
      "learning_rate": 9.680680680680681e-06,
      "loss": 0.6495,
      "step": 3190
    },
    {
      "epoch": 9.61,
      "grad_norm": 8.924003601074219,
      "learning_rate": 9.67967967967968e-06,
      "loss": 0.5767,
      "step": 3200
    },
    {
      "epoch": 9.64,
      "grad_norm": 12.401798248291016,
      "learning_rate": 9.678678678678679e-06,
      "loss": 0.5759,
      "step": 3210
    },
    {
      "epoch": 9.67,
      "grad_norm": 11.777472496032715,
      "learning_rate": 9.677677677677679e-06,
      "loss": 0.6899,
      "step": 3220
    },
    {
      "epoch": 9.7,
      "grad_norm": 11.41810417175293,
      "learning_rate": 9.676676676676677e-06,
      "loss": 0.6634,
      "step": 3230
    },
    {
      "epoch": 9.73,
      "grad_norm": 12.742891311645508,
      "learning_rate": 9.675675675675676e-06,
      "loss": 0.6457,
      "step": 3240
    },
    {
      "epoch": 9.76,
      "grad_norm": 14.691203117370605,
      "learning_rate": 9.674674674674676e-06,
      "loss": 0.6526,
      "step": 3250
    },
    {
      "epoch": 9.79,
      "grad_norm": 13.14875602722168,
      "learning_rate": 9.673673673673675e-06,
      "loss": 0.5675,
      "step": 3260
    },
    {
      "epoch": 9.82,
      "grad_norm": 14.89482593536377,
      "learning_rate": 9.672672672672673e-06,
      "loss": 0.6553,
      "step": 3270
    },
    {
      "epoch": 9.85,
      "grad_norm": 7.1069416999816895,
      "learning_rate": 9.671671671671674e-06,
      "loss": 0.6353,
      "step": 3280
    },
    {
      "epoch": 9.88,
      "grad_norm": 20.37803077697754,
      "learning_rate": 9.67067067067067e-06,
      "loss": 0.6192,
      "step": 3290
    },
    {
      "epoch": 9.91,
      "grad_norm": 11.729090690612793,
      "learning_rate": 9.669669669669671e-06,
      "loss": 0.5869,
      "step": 3300
    },
    {
      "epoch": 9.94,
      "grad_norm": 23.864059448242188,
      "learning_rate": 9.66866866866867e-06,
      "loss": 0.6176,
      "step": 3310
    },
    {
      "epoch": 9.97,
      "grad_norm": 13.164339065551758,
      "learning_rate": 9.667667667667668e-06,
      "loss": 0.6299,
      "step": 3320
    },
    {
      "epoch": 10.0,
      "grad_norm": 4.83162260055542,
      "learning_rate": 9.666666666666667e-06,
      "loss": 0.571,
      "step": 3330
    },
    {
      "epoch": 10.0,
      "eval_accuracy": 0.9209,
      "eval_loss": 0.26171091198921204,
      "eval_runtime": 30.6144,
      "eval_samples_per_second": 326.644,
      "eval_steps_per_second": 1.307,
      "step": 3330
    },
    {
      "epoch": 10.03,
      "grad_norm": 8.903618812561035,
      "learning_rate": 9.665665665665667e-06,
      "loss": 0.5832,
      "step": 3340
    },
    {
      "epoch": 10.06,
      "grad_norm": 13.1465425491333,
      "learning_rate": 9.664664664664666e-06,
      "loss": 0.6152,
      "step": 3350
    },
    {
      "epoch": 10.09,
      "grad_norm": 8.559700965881348,
      "learning_rate": 9.663663663663664e-06,
      "loss": 0.613,
      "step": 3360
    },
    {
      "epoch": 10.12,
      "grad_norm": 12.810148239135742,
      "learning_rate": 9.662662662662663e-06,
      "loss": 0.5844,
      "step": 3370
    },
    {
      "epoch": 10.15,
      "grad_norm": 26.87183380126953,
      "learning_rate": 9.661661661661661e-06,
      "loss": 0.6117,
      "step": 3380
    },
    {
      "epoch": 10.18,
      "grad_norm": 9.096632957458496,
      "learning_rate": 9.660660660660662e-06,
      "loss": 0.6312,
      "step": 3390
    },
    {
      "epoch": 10.21,
      "grad_norm": 11.064271926879883,
      "learning_rate": 9.65965965965966e-06,
      "loss": 0.6052,
      "step": 3400
    },
    {
      "epoch": 10.24,
      "grad_norm": 10.786611557006836,
      "learning_rate": 9.658658658658659e-06,
      "loss": 0.6101,
      "step": 3410
    },
    {
      "epoch": 10.27,
      "grad_norm": 14.586657524108887,
      "learning_rate": 9.65765765765766e-06,
      "loss": 0.5827,
      "step": 3420
    },
    {
      "epoch": 10.3,
      "grad_norm": 9.581392288208008,
      "learning_rate": 9.656656656656658e-06,
      "loss": 0.6229,
      "step": 3430
    },
    {
      "epoch": 10.33,
      "grad_norm": 15.541650772094727,
      "learning_rate": 9.655655655655656e-06,
      "loss": 0.6739,
      "step": 3440
    },
    {
      "epoch": 10.36,
      "grad_norm": 11.475388526916504,
      "learning_rate": 9.654654654654655e-06,
      "loss": 0.6073,
      "step": 3450
    },
    {
      "epoch": 10.39,
      "grad_norm": 10.81979751586914,
      "learning_rate": 9.653653653653654e-06,
      "loss": 0.5649,
      "step": 3460
    },
    {
      "epoch": 10.42,
      "grad_norm": 17.0631160736084,
      "learning_rate": 9.652652652652654e-06,
      "loss": 0.6303,
      "step": 3470
    },
    {
      "epoch": 10.45,
      "grad_norm": 12.085174560546875,
      "learning_rate": 9.651651651651652e-06,
      "loss": 0.6294,
      "step": 3480
    },
    {
      "epoch": 10.48,
      "grad_norm": 9.632865905761719,
      "learning_rate": 9.650650650650651e-06,
      "loss": 0.6468,
      "step": 3490
    },
    {
      "epoch": 10.51,
      "grad_norm": 14.820611000061035,
      "learning_rate": 9.649649649649651e-06,
      "loss": 0.6449,
      "step": 3500
    },
    {
      "epoch": 10.54,
      "grad_norm": 12.00233268737793,
      "learning_rate": 9.64864864864865e-06,
      "loss": 0.5553,
      "step": 3510
    },
    {
      "epoch": 10.57,
      "grad_norm": 13.7506742477417,
      "learning_rate": 9.647647647647648e-06,
      "loss": 0.576,
      "step": 3520
    },
    {
      "epoch": 10.6,
      "grad_norm": 10.14376449584961,
      "learning_rate": 9.646646646646649e-06,
      "loss": 0.6363,
      "step": 3530
    },
    {
      "epoch": 10.63,
      "grad_norm": 15.22951602935791,
      "learning_rate": 9.645645645645646e-06,
      "loss": 0.6088,
      "step": 3540
    },
    {
      "epoch": 10.66,
      "grad_norm": 13.552799224853516,
      "learning_rate": 9.644644644644644e-06,
      "loss": 0.6076,
      "step": 3550
    },
    {
      "epoch": 10.69,
      "grad_norm": 11.422019958496094,
      "learning_rate": 9.643643643643645e-06,
      "loss": 0.6158,
      "step": 3560
    },
    {
      "epoch": 10.72,
      "grad_norm": 14.275238990783691,
      "learning_rate": 9.642642642642643e-06,
      "loss": 0.5884,
      "step": 3570
    },
    {
      "epoch": 10.75,
      "grad_norm": 16.786712646484375,
      "learning_rate": 9.641641641641642e-06,
      "loss": 0.5642,
      "step": 3580
    },
    {
      "epoch": 10.78,
      "grad_norm": 11.727968215942383,
      "learning_rate": 9.640640640640642e-06,
      "loss": 0.5777,
      "step": 3590
    },
    {
      "epoch": 10.81,
      "grad_norm": 14.0637788772583,
      "learning_rate": 9.63963963963964e-06,
      "loss": 0.5905,
      "step": 3600
    },
    {
      "epoch": 10.84,
      "grad_norm": 11.379030227661133,
      "learning_rate": 9.63863863863864e-06,
      "loss": 0.5427,
      "step": 3610
    },
    {
      "epoch": 10.87,
      "grad_norm": 15.644015312194824,
      "learning_rate": 9.637637637637638e-06,
      "loss": 0.6216,
      "step": 3620
    },
    {
      "epoch": 10.9,
      "grad_norm": 13.289280891418457,
      "learning_rate": 9.636636636636636e-06,
      "loss": 0.6093,
      "step": 3630
    },
    {
      "epoch": 10.93,
      "grad_norm": 23.49856948852539,
      "learning_rate": 9.635635635635637e-06,
      "loss": 0.5797,
      "step": 3640
    },
    {
      "epoch": 10.96,
      "grad_norm": 10.170689582824707,
      "learning_rate": 9.634634634634635e-06,
      "loss": 0.598,
      "step": 3650
    },
    {
      "epoch": 10.99,
      "grad_norm": 32.763023376464844,
      "learning_rate": 9.633633633633634e-06,
      "loss": 0.6006,
      "step": 3660
    },
    {
      "epoch": 11.0,
      "eval_accuracy": 0.9215,
      "eval_loss": 0.26355910301208496,
      "eval_runtime": 30.4205,
      "eval_samples_per_second": 328.726,
      "eval_steps_per_second": 1.315,
      "step": 3663
    },
    {
      "epoch": 11.02,
      "grad_norm": 11.804524421691895,
      "learning_rate": 9.632632632632634e-06,
      "loss": 0.7672,
      "step": 3670
    },
    {
      "epoch": 11.05,
      "grad_norm": 12.28615665435791,
      "learning_rate": 9.631631631631633e-06,
      "loss": 0.5228,
      "step": 3680
    },
    {
      "epoch": 11.08,
      "grad_norm": 11.602808952331543,
      "learning_rate": 9.630630630630631e-06,
      "loss": 0.5835,
      "step": 3690
    },
    {
      "epoch": 11.11,
      "grad_norm": 11.79984188079834,
      "learning_rate": 9.62962962962963e-06,
      "loss": 0.5576,
      "step": 3700
    },
    {
      "epoch": 11.14,
      "grad_norm": 28.47237777709961,
      "learning_rate": 9.628628628628629e-06,
      "loss": 0.6201,
      "step": 3710
    },
    {
      "epoch": 11.17,
      "grad_norm": 29.43509864807129,
      "learning_rate": 9.627627627627629e-06,
      "loss": 0.5559,
      "step": 3720
    },
    {
      "epoch": 11.2,
      "grad_norm": 15.982295036315918,
      "learning_rate": 9.626626626626627e-06,
      "loss": 0.6419,
      "step": 3730
    },
    {
      "epoch": 11.23,
      "grad_norm": 10.097960472106934,
      "learning_rate": 9.625625625625626e-06,
      "loss": 0.5709,
      "step": 3740
    },
    {
      "epoch": 11.26,
      "grad_norm": 12.428840637207031,
      "learning_rate": 9.624624624624626e-06,
      "loss": 0.5881,
      "step": 3750
    },
    {
      "epoch": 11.29,
      "grad_norm": 7.855467319488525,
      "learning_rate": 9.623623623623625e-06,
      "loss": 0.5935,
      "step": 3760
    },
    {
      "epoch": 11.32,
      "grad_norm": 13.401034355163574,
      "learning_rate": 9.622622622622624e-06,
      "loss": 0.6379,
      "step": 3770
    },
    {
      "epoch": 11.35,
      "grad_norm": 11.108011245727539,
      "learning_rate": 9.621621621621622e-06,
      "loss": 0.6269,
      "step": 3780
    },
    {
      "epoch": 11.38,
      "grad_norm": 11.773721694946289,
      "learning_rate": 9.62062062062062e-06,
      "loss": 0.607,
      "step": 3790
    },
    {
      "epoch": 11.41,
      "grad_norm": 17.96162223815918,
      "learning_rate": 9.61961961961962e-06,
      "loss": 0.6359,
      "step": 3800
    },
    {
      "epoch": 11.44,
      "grad_norm": 13.181288719177246,
      "learning_rate": 9.61861861861862e-06,
      "loss": 0.5791,
      "step": 3810
    },
    {
      "epoch": 11.47,
      "grad_norm": 10.062200546264648,
      "learning_rate": 9.617617617617618e-06,
      "loss": 0.5776,
      "step": 3820
    },
    {
      "epoch": 11.5,
      "grad_norm": 11.909144401550293,
      "learning_rate": 9.616616616616617e-06,
      "loss": 0.5902,
      "step": 3830
    },
    {
      "epoch": 11.53,
      "grad_norm": 11.405366897583008,
      "learning_rate": 9.615615615615617e-06,
      "loss": 0.6111,
      "step": 3840
    },
    {
      "epoch": 11.56,
      "grad_norm": 9.566732406616211,
      "learning_rate": 9.614614614614616e-06,
      "loss": 0.6061,
      "step": 3850
    },
    {
      "epoch": 11.59,
      "grad_norm": 12.742683410644531,
      "learning_rate": 9.613613613613614e-06,
      "loss": 0.6257,
      "step": 3860
    },
    {
      "epoch": 11.62,
      "grad_norm": 9.52748966217041,
      "learning_rate": 9.612612612612613e-06,
      "loss": 0.6245,
      "step": 3870
    },
    {
      "epoch": 11.65,
      "grad_norm": 9.65878963470459,
      "learning_rate": 9.611611611611611e-06,
      "loss": 0.6549,
      "step": 3880
    },
    {
      "epoch": 11.68,
      "grad_norm": 9.731181144714355,
      "learning_rate": 9.610610610610612e-06,
      "loss": 0.5947,
      "step": 3890
    },
    {
      "epoch": 11.71,
      "grad_norm": 15.657157897949219,
      "learning_rate": 9.60960960960961e-06,
      "loss": 0.6265,
      "step": 3900
    },
    {
      "epoch": 11.74,
      "grad_norm": 11.384998321533203,
      "learning_rate": 9.608608608608609e-06,
      "loss": 0.5735,
      "step": 3910
    },
    {
      "epoch": 11.77,
      "grad_norm": 17.66204261779785,
      "learning_rate": 9.60760760760761e-06,
      "loss": 0.5889,
      "step": 3920
    },
    {
      "epoch": 11.8,
      "grad_norm": 14.75931453704834,
      "learning_rate": 9.606606606606608e-06,
      "loss": 0.6254,
      "step": 3930
    },
    {
      "epoch": 11.83,
      "grad_norm": 11.870536804199219,
      "learning_rate": 9.605605605605606e-06,
      "loss": 0.5641,
      "step": 3940
    },
    {
      "epoch": 11.86,
      "grad_norm": 13.277334213256836,
      "learning_rate": 9.604604604604605e-06,
      "loss": 0.6206,
      "step": 3950
    },
    {
      "epoch": 11.89,
      "grad_norm": 9.664814949035645,
      "learning_rate": 9.603603603603604e-06,
      "loss": 0.5821,
      "step": 3960
    },
    {
      "epoch": 11.92,
      "grad_norm": 9.741429328918457,
      "learning_rate": 9.602602602602604e-06,
      "loss": 0.5563,
      "step": 3970
    },
    {
      "epoch": 11.95,
      "grad_norm": 8.5345458984375,
      "learning_rate": 9.601601601601602e-06,
      "loss": 0.6133,
      "step": 3980
    },
    {
      "epoch": 11.98,
      "grad_norm": 13.11860179901123,
      "learning_rate": 9.600600600600601e-06,
      "loss": 0.5471,
      "step": 3990
    },
    {
      "epoch": 12.0,
      "eval_accuracy": 0.9239,
      "eval_loss": 0.2516760230064392,
      "eval_runtime": 30.4731,
      "eval_samples_per_second": 328.159,
      "eval_steps_per_second": 1.313,
      "step": 3996
    },
    {
      "epoch": 12.01,
      "grad_norm": 14.547802925109863,
      "learning_rate": 9.5995995995996e-06,
      "loss": 0.955,
      "step": 4000
    },
    {
      "epoch": 12.04,
      "grad_norm": 14.620524406433105,
      "learning_rate": 9.5985985985986e-06,
      "loss": 0.5806,
      "step": 4010
    },
    {
      "epoch": 12.07,
      "grad_norm": 11.669858932495117,
      "learning_rate": 9.597597597597599e-06,
      "loss": 0.5529,
      "step": 4020
    },
    {
      "epoch": 12.1,
      "grad_norm": 7.106392860412598,
      "learning_rate": 9.596596596596597e-06,
      "loss": 0.5495,
      "step": 4030
    },
    {
      "epoch": 12.13,
      "grad_norm": 9.383758544921875,
      "learning_rate": 9.595595595595596e-06,
      "loss": 0.5426,
      "step": 4040
    },
    {
      "epoch": 12.16,
      "grad_norm": 11.75998306274414,
      "learning_rate": 9.594594594594594e-06,
      "loss": 0.5571,
      "step": 4050
    },
    {
      "epoch": 12.19,
      "grad_norm": 8.375412940979004,
      "learning_rate": 9.593593593593595e-06,
      "loss": 0.5685,
      "step": 4060
    },
    {
      "epoch": 12.22,
      "grad_norm": 12.944570541381836,
      "learning_rate": 9.592592592592593e-06,
      "loss": 0.5316,
      "step": 4070
    },
    {
      "epoch": 12.25,
      "grad_norm": 17.294076919555664,
      "learning_rate": 9.591591591591592e-06,
      "loss": 0.6194,
      "step": 4080
    },
    {
      "epoch": 12.28,
      "grad_norm": 10.051231384277344,
      "learning_rate": 9.590590590590592e-06,
      "loss": 0.5876,
      "step": 4090
    },
    {
      "epoch": 12.31,
      "grad_norm": 312.2090148925781,
      "learning_rate": 9.58958958958959e-06,
      "loss": 0.5972,
      "step": 4100
    },
    {
      "epoch": 12.34,
      "grad_norm": 13.954002380371094,
      "learning_rate": 9.58858858858859e-06,
      "loss": 0.562,
      "step": 4110
    },
    {
      "epoch": 12.37,
      "grad_norm": 11.686988830566406,
      "learning_rate": 9.587587587587588e-06,
      "loss": 0.5695,
      "step": 4120
    },
    {
      "epoch": 12.4,
      "grad_norm": 12.769394874572754,
      "learning_rate": 9.586586586586586e-06,
      "loss": 0.6047,
      "step": 4130
    },
    {
      "epoch": 12.43,
      "grad_norm": 8.87507438659668,
      "learning_rate": 9.585585585585587e-06,
      "loss": 0.6268,
      "step": 4140
    },
    {
      "epoch": 12.46,
      "grad_norm": 15.076183319091797,
      "learning_rate": 9.584584584584585e-06,
      "loss": 0.5486,
      "step": 4150
    },
    {
      "epoch": 12.49,
      "grad_norm": 10.28525447845459,
      "learning_rate": 9.583583583583584e-06,
      "loss": 0.6153,
      "step": 4160
    },
    {
      "epoch": 12.52,
      "grad_norm": 10.834978103637695,
      "learning_rate": 9.582582582582584e-06,
      "loss": 0.5824,
      "step": 4170
    },
    {
      "epoch": 12.55,
      "grad_norm": 26.89397621154785,
      "learning_rate": 9.581581581581583e-06,
      "loss": 0.5857,
      "step": 4180
    },
    {
      "epoch": 12.58,
      "grad_norm": 10.6329984664917,
      "learning_rate": 9.580580580580581e-06,
      "loss": 0.5734,
      "step": 4190
    },
    {
      "epoch": 12.61,
      "grad_norm": 11.606546401977539,
      "learning_rate": 9.57957957957958e-06,
      "loss": 0.5593,
      "step": 4200
    },
    {
      "epoch": 12.64,
      "grad_norm": 10.23750114440918,
      "learning_rate": 9.578578578578579e-06,
      "loss": 0.5584,
      "step": 4210
    },
    {
      "epoch": 12.67,
      "grad_norm": 13.51978588104248,
      "learning_rate": 9.577577577577579e-06,
      "loss": 0.5167,
      "step": 4220
    },
    {
      "epoch": 12.7,
      "grad_norm": 11.312134742736816,
      "learning_rate": 9.576576576576578e-06,
      "loss": 0.5347,
      "step": 4230
    },
    {
      "epoch": 12.73,
      "grad_norm": 10.276420593261719,
      "learning_rate": 9.575575575575576e-06,
      "loss": 0.5815,
      "step": 4240
    },
    {
      "epoch": 12.76,
      "grad_norm": 9.689416885375977,
      "learning_rate": 9.574574574574575e-06,
      "loss": 0.5319,
      "step": 4250
    },
    {
      "epoch": 12.79,
      "grad_norm": 9.421140670776367,
      "learning_rate": 9.573573573573575e-06,
      "loss": 0.5544,
      "step": 4260
    },
    {
      "epoch": 12.82,
      "grad_norm": 9.061555862426758,
      "learning_rate": 9.572572572572574e-06,
      "loss": 0.6004,
      "step": 4270
    },
    {
      "epoch": 12.85,
      "grad_norm": 9.207202911376953,
      "learning_rate": 9.571571571571572e-06,
      "loss": 0.5692,
      "step": 4280
    },
    {
      "epoch": 12.88,
      "grad_norm": 21.15509796142578,
      "learning_rate": 9.57057057057057e-06,
      "loss": 0.5143,
      "step": 4290
    },
    {
      "epoch": 12.91,
      "grad_norm": 12.711333274841309,
      "learning_rate": 9.56956956956957e-06,
      "loss": 0.6018,
      "step": 4300
    },
    {
      "epoch": 12.94,
      "grad_norm": 8.004502296447754,
      "learning_rate": 9.56856856856857e-06,
      "loss": 0.5522,
      "step": 4310
    },
    {
      "epoch": 12.97,
      "grad_norm": 17.612730026245117,
      "learning_rate": 9.567567567567568e-06,
      "loss": 0.6324,
      "step": 4320
    },
    {
      "epoch": 13.0,
      "eval_accuracy": 0.9226,
      "eval_loss": 0.2519969344139099,
      "eval_runtime": 30.5594,
      "eval_samples_per_second": 327.231,
      "eval_steps_per_second": 1.309,
      "step": 4329
    },
    {
      "epoch": 13.0,
      "grad_norm": 10.345263481140137,
      "learning_rate": 9.566566566566567e-06,
      "loss": 0.5334,
      "step": 4330
    },
    {
      "epoch": 13.03,
      "grad_norm": 13.402093887329102,
      "learning_rate": 9.565565565565567e-06,
      "loss": 0.5308,
      "step": 4340
    },
    {
      "epoch": 13.06,
      "grad_norm": 18.5904598236084,
      "learning_rate": 9.564564564564566e-06,
      "loss": 0.5594,
      "step": 4350
    },
    {
      "epoch": 13.09,
      "grad_norm": 12.926309585571289,
      "learning_rate": 9.563563563563564e-06,
      "loss": 0.5892,
      "step": 4360
    },
    {
      "epoch": 13.12,
      "grad_norm": 9.018284797668457,
      "learning_rate": 9.562562562562563e-06,
      "loss": 0.6252,
      "step": 4370
    },
    {
      "epoch": 13.15,
      "grad_norm": 12.864182472229004,
      "learning_rate": 9.561561561561562e-06,
      "loss": 0.5282,
      "step": 4380
    },
    {
      "epoch": 13.18,
      "grad_norm": 18.937196731567383,
      "learning_rate": 9.560560560560562e-06,
      "loss": 0.5713,
      "step": 4390
    },
    {
      "epoch": 13.21,
      "grad_norm": 10.198113441467285,
      "learning_rate": 9.55955955955956e-06,
      "loss": 0.5767,
      "step": 4400
    },
    {
      "epoch": 13.24,
      "grad_norm": 8.118650436401367,
      "learning_rate": 9.558558558558559e-06,
      "loss": 0.5426,
      "step": 4410
    },
    {
      "epoch": 13.27,
      "grad_norm": 19.721181869506836,
      "learning_rate": 9.55755755755756e-06,
      "loss": 0.5961,
      "step": 4420
    },
    {
      "epoch": 13.3,
      "grad_norm": 16.725208282470703,
      "learning_rate": 9.556556556556558e-06,
      "loss": 0.6371,
      "step": 4430
    },
    {
      "epoch": 13.33,
      "grad_norm": 10.268779754638672,
      "learning_rate": 9.555555555555556e-06,
      "loss": 0.5646,
      "step": 4440
    },
    {
      "epoch": 13.36,
      "grad_norm": 185.5620574951172,
      "learning_rate": 9.554554554554555e-06,
      "loss": 0.6055,
      "step": 4450
    },
    {
      "epoch": 13.39,
      "grad_norm": 16.159719467163086,
      "learning_rate": 9.553553553553554e-06,
      "loss": 0.582,
      "step": 4460
    },
    {
      "epoch": 13.42,
      "grad_norm": 13.551849365234375,
      "learning_rate": 9.552552552552552e-06,
      "loss": 0.5657,
      "step": 4470
    },
    {
      "epoch": 13.45,
      "grad_norm": 13.877328872680664,
      "learning_rate": 9.551551551551553e-06,
      "loss": 0.6025,
      "step": 4480
    },
    {
      "epoch": 13.48,
      "grad_norm": 29.763002395629883,
      "learning_rate": 9.550550550550551e-06,
      "loss": 0.5481,
      "step": 4490
    },
    {
      "epoch": 13.51,
      "grad_norm": 13.889739990234375,
      "learning_rate": 9.54954954954955e-06,
      "loss": 0.5733,
      "step": 4500
    },
    {
      "epoch": 13.54,
      "grad_norm": 69.9765853881836,
      "learning_rate": 9.54854854854855e-06,
      "loss": 0.6338,
      "step": 4510
    },
    {
      "epoch": 13.57,
      "grad_norm": 19.75606918334961,
      "learning_rate": 9.547547547547549e-06,
      "loss": 0.5628,
      "step": 4520
    },
    {
      "epoch": 13.6,
      "grad_norm": 15.370271682739258,
      "learning_rate": 9.546546546546547e-06,
      "loss": 0.596,
      "step": 4530
    },
    {
      "epoch": 13.63,
      "grad_norm": 10.773337364196777,
      "learning_rate": 9.545545545545546e-06,
      "loss": 0.6277,
      "step": 4540
    },
    {
      "epoch": 13.66,
      "grad_norm": 7.928040504455566,
      "learning_rate": 9.544544544544544e-06,
      "loss": 0.5446,
      "step": 4550
    },
    {
      "epoch": 13.69,
      "grad_norm": 15.335708618164062,
      "learning_rate": 9.543543543543545e-06,
      "loss": 0.587,
      "step": 4560
    },
    {
      "epoch": 13.72,
      "grad_norm": 12.43051815032959,
      "learning_rate": 9.542542542542543e-06,
      "loss": 0.5645,
      "step": 4570
    },
    {
      "epoch": 13.75,
      "grad_norm": 8.507277488708496,
      "learning_rate": 9.541541541541542e-06,
      "loss": 0.5238,
      "step": 4580
    },
    {
      "epoch": 13.78,
      "grad_norm": 12.679448127746582,
      "learning_rate": 9.540540540540542e-06,
      "loss": 0.5287,
      "step": 4590
    },
    {
      "epoch": 13.81,
      "grad_norm": 12.938888549804688,
      "learning_rate": 9.53953953953954e-06,
      "loss": 0.5759,
      "step": 4600
    },
    {
      "epoch": 13.84,
      "grad_norm": 11.686616897583008,
      "learning_rate": 9.53853853853854e-06,
      "loss": 0.556,
      "step": 4610
    },
    {
      "epoch": 13.87,
      "grad_norm": 10.387247085571289,
      "learning_rate": 9.537537537537538e-06,
      "loss": 0.5433,
      "step": 4620
    },
    {
      "epoch": 13.9,
      "grad_norm": 82.08344268798828,
      "learning_rate": 9.536536536536537e-06,
      "loss": 0.6063,
      "step": 4630
    },
    {
      "epoch": 13.93,
      "grad_norm": 10.989513397216797,
      "learning_rate": 9.535535535535537e-06,
      "loss": 0.5787,
      "step": 4640
    },
    {
      "epoch": 13.96,
      "grad_norm": 14.753369331359863,
      "learning_rate": 9.534534534534535e-06,
      "loss": 0.5232,
      "step": 4650
    },
    {
      "epoch": 13.99,
      "grad_norm": 7.610456943511963,
      "learning_rate": 9.533533533533534e-06,
      "loss": 0.56,
      "step": 4660
    },
    {
      "epoch": 14.0,
      "eval_accuracy": 0.924,
      "eval_loss": 0.25301432609558105,
      "eval_runtime": 30.0786,
      "eval_samples_per_second": 332.463,
      "eval_steps_per_second": 1.33,
      "step": 4662
    },
    {
      "epoch": 14.02,
      "grad_norm": 10.896953582763672,
      "learning_rate": 9.532532532532534e-06,
      "loss": 0.5122,
      "step": 4670
    },
    {
      "epoch": 14.05,
      "grad_norm": 10.77942180633545,
      "learning_rate": 9.531531531531533e-06,
      "loss": 0.5062,
      "step": 4680
    },
    {
      "epoch": 14.08,
      "grad_norm": 8.38948917388916,
      "learning_rate": 9.530530530530532e-06,
      "loss": 0.4863,
      "step": 4690
    },
    {
      "epoch": 14.11,
      "grad_norm": 11.789532661437988,
      "learning_rate": 9.52952952952953e-06,
      "loss": 0.5271,
      "step": 4700
    },
    {
      "epoch": 14.14,
      "grad_norm": 12.290909767150879,
      "learning_rate": 9.528528528528529e-06,
      "loss": 0.5155,
      "step": 4710
    },
    {
      "epoch": 14.17,
      "grad_norm": 8.020136833190918,
      "learning_rate": 9.527527527527527e-06,
      "loss": 0.5747,
      "step": 4720
    },
    {
      "epoch": 14.2,
      "grad_norm": 29.248809814453125,
      "learning_rate": 9.526526526526528e-06,
      "loss": 0.6196,
      "step": 4730
    },
    {
      "epoch": 14.23,
      "grad_norm": 19.838760375976562,
      "learning_rate": 9.525525525525526e-06,
      "loss": 0.5818,
      "step": 4740
    },
    {
      "epoch": 14.26,
      "grad_norm": 22.35586929321289,
      "learning_rate": 9.524524524524525e-06,
      "loss": 0.5739,
      "step": 4750
    },
    {
      "epoch": 14.29,
      "grad_norm": 10.237540245056152,
      "learning_rate": 9.523523523523525e-06,
      "loss": 0.5978,
      "step": 4760
    },
    {
      "epoch": 14.32,
      "grad_norm": 8.147768020629883,
      "learning_rate": 9.522522522522524e-06,
      "loss": 0.5385,
      "step": 4770
    },
    {
      "epoch": 14.35,
      "grad_norm": 10.893712997436523,
      "learning_rate": 9.521521521521522e-06,
      "loss": 0.5704,
      "step": 4780
    },
    {
      "epoch": 14.38,
      "grad_norm": 17.572235107421875,
      "learning_rate": 9.520520520520521e-06,
      "loss": 0.5272,
      "step": 4790
    },
    {
      "epoch": 14.41,
      "grad_norm": 12.654411315917969,
      "learning_rate": 9.51951951951952e-06,
      "loss": 0.5491,
      "step": 4800
    },
    {
      "epoch": 14.44,
      "grad_norm": 14.064764976501465,
      "learning_rate": 9.51851851851852e-06,
      "loss": 0.5954,
      "step": 4810
    },
    {
      "epoch": 14.47,
      "grad_norm": 15.357030868530273,
      "learning_rate": 9.517517517517518e-06,
      "loss": 0.566,
      "step": 4820
    },
    {
      "epoch": 14.5,
      "grad_norm": 12.153959274291992,
      "learning_rate": 9.516516516516517e-06,
      "loss": 0.6124,
      "step": 4830
    },
    {
      "epoch": 14.53,
      "grad_norm": 9.057621002197266,
      "learning_rate": 9.515515515515517e-06,
      "loss": 0.5356,
      "step": 4840
    },
    {
      "epoch": 14.56,
      "grad_norm": 10.922284126281738,
      "learning_rate": 9.514514514514516e-06,
      "loss": 0.573,
      "step": 4850
    },
    {
      "epoch": 14.59,
      "grad_norm": 13.01318359375,
      "learning_rate": 9.513513513513514e-06,
      "loss": 0.5135,
      "step": 4860
    },
    {
      "epoch": 14.62,
      "grad_norm": 10.830948829650879,
      "learning_rate": 9.512512512512513e-06,
      "loss": 0.5725,
      "step": 4870
    },
    {
      "epoch": 14.65,
      "grad_norm": 14.062458992004395,
      "learning_rate": 9.511511511511512e-06,
      "loss": 0.5882,
      "step": 4880
    },
    {
      "epoch": 14.68,
      "grad_norm": 30.20458221435547,
      "learning_rate": 9.510510510510512e-06,
      "loss": 0.5735,
      "step": 4890
    },
    {
      "epoch": 14.71,
      "grad_norm": 8.41458511352539,
      "learning_rate": 9.50950950950951e-06,
      "loss": 0.5331,
      "step": 4900
    },
    {
      "epoch": 14.74,
      "grad_norm": 7.415578365325928,
      "learning_rate": 9.508508508508509e-06,
      "loss": 0.5463,
      "step": 4910
    },
    {
      "epoch": 14.77,
      "grad_norm": 16.323087692260742,
      "learning_rate": 9.507507507507508e-06,
      "loss": 0.5347,
      "step": 4920
    },
    {
      "epoch": 14.8,
      "grad_norm": 19.459257125854492,
      "learning_rate": 9.506506506506508e-06,
      "loss": 0.5883,
      "step": 4930
    },
    {
      "epoch": 14.83,
      "grad_norm": 10.431855201721191,
      "learning_rate": 9.505505505505507e-06,
      "loss": 0.5907,
      "step": 4940
    },
    {
      "epoch": 14.86,
      "grad_norm": 10.564922332763672,
      "learning_rate": 9.504504504504505e-06,
      "loss": 0.5597,
      "step": 4950
    },
    {
      "epoch": 14.89,
      "grad_norm": 9.982166290283203,
      "learning_rate": 9.503503503503504e-06,
      "loss": 0.5107,
      "step": 4960
    },
    {
      "epoch": 14.92,
      "grad_norm": 16.064804077148438,
      "learning_rate": 9.502502502502502e-06,
      "loss": 0.5819,
      "step": 4970
    },
    {
      "epoch": 14.95,
      "grad_norm": 12.23503589630127,
      "learning_rate": 9.501501501501503e-06,
      "loss": 0.6121,
      "step": 4980
    },
    {
      "epoch": 14.98,
      "grad_norm": 12.281106948852539,
      "learning_rate": 9.500500500500501e-06,
      "loss": 0.5821,
      "step": 4990
    },
    {
      "epoch": 15.0,
      "eval_accuracy": 0.9217,
      "eval_loss": 0.25685247778892517,
      "eval_runtime": 30.4112,
      "eval_samples_per_second": 328.827,
      "eval_steps_per_second": 1.315,
      "step": 4995
    },
    {
      "epoch": 15.02,
      "grad_norm": 13.23077392578125,
      "learning_rate": 9.4994994994995e-06,
      "loss": 0.4503,
      "step": 5000
    },
    {
      "epoch": 15.05,
      "grad_norm": 12.722877502441406,
      "learning_rate": 9.4984984984985e-06,
      "loss": 0.5069,
      "step": 5010
    },
    {
      "epoch": 15.08,
      "grad_norm": 15.306122779846191,
      "learning_rate": 9.497497497497499e-06,
      "loss": 0.5421,
      "step": 5020
    },
    {
      "epoch": 15.11,
      "grad_norm": 10.477108001708984,
      "learning_rate": 9.496496496496497e-06,
      "loss": 0.5861,
      "step": 5030
    },
    {
      "epoch": 15.14,
      "grad_norm": 11.663418769836426,
      "learning_rate": 9.495495495495496e-06,
      "loss": 0.4983,
      "step": 5040
    },
    {
      "epoch": 15.17,
      "grad_norm": 15.94667911529541,
      "learning_rate": 9.494494494494494e-06,
      "loss": 0.5003,
      "step": 5050
    },
    {
      "epoch": 15.2,
      "grad_norm": 10.300756454467773,
      "learning_rate": 9.493493493493495e-06,
      "loss": 0.5396,
      "step": 5060
    },
    {
      "epoch": 15.23,
      "grad_norm": 9.597443580627441,
      "learning_rate": 9.492492492492493e-06,
      "loss": 0.5719,
      "step": 5070
    },
    {
      "epoch": 15.26,
      "grad_norm": 11.399624824523926,
      "learning_rate": 9.491491491491492e-06,
      "loss": 0.5314,
      "step": 5080
    },
    {
      "epoch": 15.29,
      "grad_norm": 10.787353515625,
      "learning_rate": 9.490490490490492e-06,
      "loss": 0.5105,
      "step": 5090
    },
    {
      "epoch": 15.32,
      "grad_norm": 9.648843765258789,
      "learning_rate": 9.489489489489491e-06,
      "loss": 0.5592,
      "step": 5100
    },
    {
      "epoch": 15.35,
      "grad_norm": 11.683751106262207,
      "learning_rate": 9.48848848848849e-06,
      "loss": 0.4935,
      "step": 5110
    },
    {
      "epoch": 15.38,
      "grad_norm": 9.463213920593262,
      "learning_rate": 9.487487487487488e-06,
      "loss": 0.5321,
      "step": 5120
    },
    {
      "epoch": 15.41,
      "grad_norm": 15.175768852233887,
      "learning_rate": 9.486486486486487e-06,
      "loss": 0.5266,
      "step": 5130
    },
    {
      "epoch": 15.44,
      "grad_norm": 12.34326171875,
      "learning_rate": 9.485485485485487e-06,
      "loss": 0.522,
      "step": 5140
    },
    {
      "epoch": 15.47,
      "grad_norm": 9.944204330444336,
      "learning_rate": 9.484484484484486e-06,
      "loss": 0.5229,
      "step": 5150
    },
    {
      "epoch": 15.5,
      "grad_norm": 11.730018615722656,
      "learning_rate": 9.483483483483484e-06,
      "loss": 0.5248,
      "step": 5160
    },
    {
      "epoch": 15.53,
      "grad_norm": 11.413137435913086,
      "learning_rate": 9.482482482482483e-06,
      "loss": 0.5654,
      "step": 5170
    },
    {
      "epoch": 15.56,
      "grad_norm": 14.541297912597656,
      "learning_rate": 9.481481481481483e-06,
      "loss": 0.5079,
      "step": 5180
    },
    {
      "epoch": 15.59,
      "grad_norm": 16.22747230529785,
      "learning_rate": 9.480480480480482e-06,
      "loss": 0.5292,
      "step": 5190
    },
    {
      "epoch": 15.62,
      "grad_norm": 10.435576438903809,
      "learning_rate": 9.47947947947948e-06,
      "loss": 0.488,
      "step": 5200
    },
    {
      "epoch": 15.65,
      "grad_norm": 9.990010261535645,
      "learning_rate": 9.478478478478479e-06,
      "loss": 0.4933,
      "step": 5210
    },
    {
      "epoch": 15.68,
      "grad_norm": 13.72536563873291,
      "learning_rate": 9.477477477477477e-06,
      "loss": 0.5082,
      "step": 5220
    },
    {
      "epoch": 15.71,
      "grad_norm": 15.257143020629883,
      "learning_rate": 9.476476476476478e-06,
      "loss": 0.5054,
      "step": 5230
    },
    {
      "epoch": 15.74,
      "grad_norm": 13.4158935546875,
      "learning_rate": 9.475475475475476e-06,
      "loss": 0.5062,
      "step": 5240
    },
    {
      "epoch": 15.77,
      "grad_norm": 9.417861938476562,
      "learning_rate": 9.474474474474475e-06,
      "loss": 0.5095,
      "step": 5250
    },
    {
      "epoch": 15.8,
      "grad_norm": 15.251216888427734,
      "learning_rate": 9.473473473473475e-06,
      "loss": 0.601,
      "step": 5260
    },
    {
      "epoch": 15.83,
      "grad_norm": 9.081033706665039,
      "learning_rate": 9.472472472472474e-06,
      "loss": 0.4405,
      "step": 5270
    },
    {
      "epoch": 15.86,
      "grad_norm": 9.81359577178955,
      "learning_rate": 9.471471471471472e-06,
      "loss": 0.5421,
      "step": 5280
    },
    {
      "epoch": 15.89,
      "grad_norm": 9.995782852172852,
      "learning_rate": 9.470470470470471e-06,
      "loss": 0.5325,
      "step": 5290
    },
    {
      "epoch": 15.92,
      "grad_norm": 10.108242988586426,
      "learning_rate": 9.46946946946947e-06,
      "loss": 0.5547,
      "step": 5300
    },
    {
      "epoch": 15.95,
      "grad_norm": 16.46483039855957,
      "learning_rate": 9.46846846846847e-06,
      "loss": 0.5079,
      "step": 5310
    },
    {
      "epoch": 15.98,
      "grad_norm": 9.164888381958008,
      "learning_rate": 9.467467467467468e-06,
      "loss": 0.5203,
      "step": 5320
    },
    {
      "epoch": 16.0,
      "eval_accuracy": 0.9278,
      "eval_loss": 0.24757839739322662,
      "eval_runtime": 29.8672,
      "eval_samples_per_second": 334.816,
      "eval_steps_per_second": 1.339,
      "step": 5328
    },
    {
      "epoch": 16.01,
      "grad_norm": 10.40545654296875,
      "learning_rate": 9.466466466466467e-06,
      "loss": 0.6069,
      "step": 5330
    },
    {
      "epoch": 16.04,
      "grad_norm": 11.244277954101562,
      "learning_rate": 9.465465465465467e-06,
      "loss": 0.5398,
      "step": 5340
    },
    {
      "epoch": 16.07,
      "grad_norm": 9.848921775817871,
      "learning_rate": 9.464464464464466e-06,
      "loss": 0.522,
      "step": 5350
    },
    {
      "epoch": 16.1,
      "grad_norm": 10.520909309387207,
      "learning_rate": 9.463463463463464e-06,
      "loss": 0.5363,
      "step": 5360
    },
    {
      "epoch": 16.13,
      "grad_norm": 15.065001487731934,
      "learning_rate": 9.462462462462463e-06,
      "loss": 0.6021,
      "step": 5370
    },
    {
      "epoch": 16.16,
      "grad_norm": 11.10487174987793,
      "learning_rate": 9.461461461461462e-06,
      "loss": 0.5189,
      "step": 5380
    },
    {
      "epoch": 16.19,
      "grad_norm": 10.829561233520508,
      "learning_rate": 9.46046046046046e-06,
      "loss": 0.5338,
      "step": 5390
    },
    {
      "epoch": 16.22,
      "grad_norm": 11.856501579284668,
      "learning_rate": 9.45945945945946e-06,
      "loss": 0.503,
      "step": 5400
    },
    {
      "epoch": 16.25,
      "grad_norm": 17.04743003845215,
      "learning_rate": 9.458458458458459e-06,
      "loss": 0.5408,
      "step": 5410
    },
    {
      "epoch": 16.28,
      "grad_norm": 11.075129508972168,
      "learning_rate": 9.457457457457458e-06,
      "loss": 0.506,
      "step": 5420
    },
    {
      "epoch": 16.31,
      "grad_norm": 9.650065422058105,
      "learning_rate": 9.456456456456458e-06,
      "loss": 0.5234,
      "step": 5430
    },
    {
      "epoch": 16.34,
      "grad_norm": 13.577381134033203,
      "learning_rate": 9.455455455455457e-06,
      "loss": 0.5978,
      "step": 5440
    },
    {
      "epoch": 16.37,
      "grad_norm": 21.916799545288086,
      "learning_rate": 9.454454454454455e-06,
      "loss": 0.4931,
      "step": 5450
    },
    {
      "epoch": 16.4,
      "grad_norm": 12.489849090576172,
      "learning_rate": 9.453453453453454e-06,
      "loss": 0.5898,
      "step": 5460
    },
    {
      "epoch": 16.43,
      "grad_norm": 13.279026985168457,
      "learning_rate": 9.452452452452452e-06,
      "loss": 0.5541,
      "step": 5470
    },
    {
      "epoch": 16.46,
      "grad_norm": 13.039314270019531,
      "learning_rate": 9.451451451451453e-06,
      "loss": 0.5193,
      "step": 5480
    },
    {
      "epoch": 16.49,
      "grad_norm": 7.700623989105225,
      "learning_rate": 9.450450450450451e-06,
      "loss": 0.5478,
      "step": 5490
    },
    {
      "epoch": 16.52,
      "grad_norm": 9.966477394104004,
      "learning_rate": 9.44944944944945e-06,
      "loss": 0.5737,
      "step": 5500
    },
    {
      "epoch": 16.55,
      "grad_norm": 10.083463668823242,
      "learning_rate": 9.44844844844845e-06,
      "loss": 0.5538,
      "step": 5510
    },
    {
      "epoch": 16.58,
      "grad_norm": 20.965614318847656,
      "learning_rate": 9.447447447447449e-06,
      "loss": 0.523,
      "step": 5520
    },
    {
      "epoch": 16.61,
      "grad_norm": 24.368310928344727,
      "learning_rate": 9.446446446446447e-06,
      "loss": 0.5488,
      "step": 5530
    },
    {
      "epoch": 16.64,
      "grad_norm": 11.053398132324219,
      "learning_rate": 9.445445445445446e-06,
      "loss": 0.5208,
      "step": 5540
    },
    {
      "epoch": 16.67,
      "grad_norm": 12.406140327453613,
      "learning_rate": 9.444444444444445e-06,
      "loss": 0.5106,
      "step": 5550
    },
    {
      "epoch": 16.7,
      "grad_norm": 9.5691556930542,
      "learning_rate": 9.443443443443445e-06,
      "loss": 0.5151,
      "step": 5560
    },
    {
      "epoch": 16.73,
      "grad_norm": 8.728398323059082,
      "learning_rate": 9.442442442442443e-06,
      "loss": 0.4747,
      "step": 5570
    },
    {
      "epoch": 16.76,
      "grad_norm": 8.877335548400879,
      "learning_rate": 9.441441441441442e-06,
      "loss": 0.5235,
      "step": 5580
    },
    {
      "epoch": 16.79,
      "grad_norm": 12.02859878540039,
      "learning_rate": 9.440440440440442e-06,
      "loss": 0.5394,
      "step": 5590
    },
    {
      "epoch": 16.82,
      "grad_norm": 11.32858657836914,
      "learning_rate": 9.439439439439441e-06,
      "loss": 0.5502,
      "step": 5600
    },
    {
      "epoch": 16.85,
      "grad_norm": 11.6917085647583,
      "learning_rate": 9.43843843843844e-06,
      "loss": 0.5332,
      "step": 5610
    },
    {
      "epoch": 16.88,
      "grad_norm": 21.33887481689453,
      "learning_rate": 9.437437437437438e-06,
      "loss": 0.55,
      "step": 5620
    },
    {
      "epoch": 16.91,
      "grad_norm": 10.514164924621582,
      "learning_rate": 9.436436436436437e-06,
      "loss": 0.5489,
      "step": 5630
    },
    {
      "epoch": 16.94,
      "grad_norm": 12.23349666595459,
      "learning_rate": 9.435435435435435e-06,
      "loss": 0.4969,
      "step": 5640
    },
    {
      "epoch": 16.97,
      "grad_norm": 10.175886154174805,
      "learning_rate": 9.434434434434436e-06,
      "loss": 0.5815,
      "step": 5650
    },
    {
      "epoch": 17.0,
      "grad_norm": 23.207658767700195,
      "learning_rate": 9.433433433433434e-06,
      "loss": 0.5387,
      "step": 5660
    },
    {
      "epoch": 17.0,
      "eval_accuracy": 0.9264,
      "eval_loss": 0.25176796317100525,
      "eval_runtime": 30.4959,
      "eval_samples_per_second": 327.913,
      "eval_steps_per_second": 1.312,
      "step": 5661
    },
    {
      "epoch": 17.03,
      "grad_norm": 13.02208137512207,
      "learning_rate": 9.432432432432433e-06,
      "loss": 0.6268,
      "step": 5670
    },
    {
      "epoch": 17.06,
      "grad_norm": 12.880228996276855,
      "learning_rate": 9.431431431431433e-06,
      "loss": 0.5462,
      "step": 5680
    },
    {
      "epoch": 17.09,
      "grad_norm": 12.443921089172363,
      "learning_rate": 9.430430430430432e-06,
      "loss": 0.5521,
      "step": 5690
    },
    {
      "epoch": 17.12,
      "grad_norm": 7.886068820953369,
      "learning_rate": 9.42942942942943e-06,
      "loss": 0.5117,
      "step": 5700
    },
    {
      "epoch": 17.15,
      "grad_norm": 15.501462936401367,
      "learning_rate": 9.428428428428429e-06,
      "loss": 0.5302,
      "step": 5710
    },
    {
      "epoch": 17.18,
      "grad_norm": 12.38558578491211,
      "learning_rate": 9.427427427427427e-06,
      "loss": 0.5136,
      "step": 5720
    },
    {
      "epoch": 17.21,
      "grad_norm": 12.14132308959961,
      "learning_rate": 9.426426426426428e-06,
      "loss": 0.4861,
      "step": 5730
    },
    {
      "epoch": 17.24,
      "grad_norm": 10.382802963256836,
      "learning_rate": 9.425425425425426e-06,
      "loss": 0.5109,
      "step": 5740
    },
    {
      "epoch": 17.27,
      "grad_norm": 15.48659610748291,
      "learning_rate": 9.424424424424425e-06,
      "loss": 0.4857,
      "step": 5750
    },
    {
      "epoch": 17.3,
      "grad_norm": 8.58863639831543,
      "learning_rate": 9.423423423423425e-06,
      "loss": 0.5337,
      "step": 5760
    },
    {
      "epoch": 17.33,
      "grad_norm": 17.04405403137207,
      "learning_rate": 9.422422422422424e-06,
      "loss": 0.5847,
      "step": 5770
    },
    {
      "epoch": 17.36,
      "grad_norm": 12.838520050048828,
      "learning_rate": 9.421421421421422e-06,
      "loss": 0.5376,
      "step": 5780
    },
    {
      "epoch": 17.39,
      "grad_norm": 7.937318325042725,
      "learning_rate": 9.420420420420421e-06,
      "loss": 0.5146,
      "step": 5790
    },
    {
      "epoch": 17.42,
      "grad_norm": 9.863258361816406,
      "learning_rate": 9.41941941941942e-06,
      "loss": 0.5409,
      "step": 5800
    },
    {
      "epoch": 17.45,
      "grad_norm": 14.783023834228516,
      "learning_rate": 9.41841841841842e-06,
      "loss": 0.5522,
      "step": 5810
    },
    {
      "epoch": 17.48,
      "grad_norm": 9.500492095947266,
      "learning_rate": 9.417417417417418e-06,
      "loss": 0.4801,
      "step": 5820
    },
    {
      "epoch": 17.51,
      "grad_norm": 9.685933113098145,
      "learning_rate": 9.416416416416417e-06,
      "loss": 0.53,
      "step": 5830
    },
    {
      "epoch": 17.54,
      "grad_norm": 20.050275802612305,
      "learning_rate": 9.415415415415416e-06,
      "loss": 0.5096,
      "step": 5840
    },
    {
      "epoch": 17.57,
      "grad_norm": 10.203237533569336,
      "learning_rate": 9.414414414414416e-06,
      "loss": 0.5229,
      "step": 5850
    },
    {
      "epoch": 17.6,
      "grad_norm": 8.337947845458984,
      "learning_rate": 9.413413413413413e-06,
      "loss": 0.4883,
      "step": 5860
    },
    {
      "epoch": 17.63,
      "grad_norm": 10.492974281311035,
      "learning_rate": 9.412412412412413e-06,
      "loss": 0.552,
      "step": 5870
    },
    {
      "epoch": 17.66,
      "grad_norm": 19.00461196899414,
      "learning_rate": 9.411411411411412e-06,
      "loss": 0.5478,
      "step": 5880
    },
    {
      "epoch": 17.69,
      "grad_norm": 8.795330047607422,
      "learning_rate": 9.41041041041041e-06,
      "loss": 0.4974,
      "step": 5890
    },
    {
      "epoch": 17.72,
      "grad_norm": 14.406013488769531,
      "learning_rate": 9.40940940940941e-06,
      "loss": 0.5234,
      "step": 5900
    },
    {
      "epoch": 17.75,
      "grad_norm": 18.222362518310547,
      "learning_rate": 9.40840840840841e-06,
      "loss": 0.523,
      "step": 5910
    },
    {
      "epoch": 17.78,
      "grad_norm": 26.603675842285156,
      "learning_rate": 9.407407407407408e-06,
      "loss": 0.4944,
      "step": 5920
    },
    {
      "epoch": 17.81,
      "grad_norm": 10.870711326599121,
      "learning_rate": 9.406406406406408e-06,
      "loss": 0.579,
      "step": 5930
    },
    {
      "epoch": 17.84,
      "grad_norm": 15.319726943969727,
      "learning_rate": 9.405405405405407e-06,
      "loss": 0.4967,
      "step": 5940
    },
    {
      "epoch": 17.87,
      "grad_norm": 12.175629615783691,
      "learning_rate": 9.404404404404405e-06,
      "loss": 0.4831,
      "step": 5950
    },
    {
      "epoch": 17.9,
      "grad_norm": 12.597267150878906,
      "learning_rate": 9.403403403403404e-06,
      "loss": 0.4885,
      "step": 5960
    },
    {
      "epoch": 17.93,
      "grad_norm": 18.843530654907227,
      "learning_rate": 9.402402402402402e-06,
      "loss": 0.5816,
      "step": 5970
    },
    {
      "epoch": 17.96,
      "grad_norm": 27.022314071655273,
      "learning_rate": 9.401401401401403e-06,
      "loss": 0.4817,
      "step": 5980
    },
    {
      "epoch": 17.99,
      "grad_norm": 53.82378387451172,
      "learning_rate": 9.400400400400401e-06,
      "loss": 0.4921,
      "step": 5990
    },
    {
      "epoch": 18.0,
      "eval_accuracy": 0.9282,
      "eval_loss": 0.2474725842475891,
      "eval_runtime": 30.3563,
      "eval_samples_per_second": 329.421,
      "eval_steps_per_second": 1.318,
      "step": 5994
    },
    {
      "epoch": 18.02,
      "grad_norm": 12.790140151977539,
      "learning_rate": 9.3993993993994e-06,
      "loss": 0.4826,
      "step": 6000
    },
    {
      "epoch": 18.05,
      "grad_norm": 11.542041778564453,
      "learning_rate": 9.3983983983984e-06,
      "loss": 0.492,
      "step": 6010
    },
    {
      "epoch": 18.08,
      "grad_norm": 27.646615982055664,
      "learning_rate": 9.397397397397399e-06,
      "loss": 0.5432,
      "step": 6020
    },
    {
      "epoch": 18.11,
      "grad_norm": 9.859136581420898,
      "learning_rate": 9.396396396396397e-06,
      "loss": 0.4926,
      "step": 6030
    },
    {
      "epoch": 18.14,
      "grad_norm": 12.287638664245605,
      "learning_rate": 9.395395395395396e-06,
      "loss": 0.5048,
      "step": 6040
    },
    {
      "epoch": 18.17,
      "grad_norm": 12.397451400756836,
      "learning_rate": 9.394394394394395e-06,
      "loss": 0.4915,
      "step": 6050
    },
    {
      "epoch": 18.2,
      "grad_norm": 10.373201370239258,
      "learning_rate": 9.393393393393393e-06,
      "loss": 0.4925,
      "step": 6060
    },
    {
      "epoch": 18.23,
      "grad_norm": 12.417383193969727,
      "learning_rate": 9.392392392392394e-06,
      "loss": 0.5435,
      "step": 6070
    },
    {
      "epoch": 18.26,
      "grad_norm": 7.52992057800293,
      "learning_rate": 9.391391391391392e-06,
      "loss": 0.5586,
      "step": 6080
    },
    {
      "epoch": 18.29,
      "grad_norm": 11.180517196655273,
      "learning_rate": 9.39039039039039e-06,
      "loss": 0.5261,
      "step": 6090
    },
    {
      "epoch": 18.32,
      "grad_norm": 14.029952049255371,
      "learning_rate": 9.389389389389391e-06,
      "loss": 0.5059,
      "step": 6100
    },
    {
      "epoch": 18.35,
      "grad_norm": 13.72758960723877,
      "learning_rate": 9.388388388388388e-06,
      "loss": 0.5025,
      "step": 6110
    },
    {
      "epoch": 18.38,
      "grad_norm": 9.351816177368164,
      "learning_rate": 9.387387387387388e-06,
      "loss": 0.4725,
      "step": 6120
    },
    {
      "epoch": 18.41,
      "grad_norm": 12.995992660522461,
      "learning_rate": 9.386386386386387e-06,
      "loss": 0.5565,
      "step": 6130
    },
    {
      "epoch": 18.44,
      "grad_norm": 9.428690910339355,
      "learning_rate": 9.385385385385385e-06,
      "loss": 0.4801,
      "step": 6140
    },
    {
      "epoch": 18.47,
      "grad_norm": 15.22610092163086,
      "learning_rate": 9.384384384384386e-06,
      "loss": 0.4922,
      "step": 6150
    },
    {
      "epoch": 18.5,
      "grad_norm": 9.8953857421875,
      "learning_rate": 9.383383383383384e-06,
      "loss": 0.5103,
      "step": 6160
    },
    {
      "epoch": 18.53,
      "grad_norm": 15.755030632019043,
      "learning_rate": 9.382382382382383e-06,
      "loss": 0.5289,
      "step": 6170
    },
    {
      "epoch": 18.56,
      "grad_norm": 8.752655029296875,
      "learning_rate": 9.381381381381383e-06,
      "loss": 0.5092,
      "step": 6180
    },
    {
      "epoch": 18.59,
      "grad_norm": 57.676734924316406,
      "learning_rate": 9.380380380380382e-06,
      "loss": 0.5321,
      "step": 6190
    },
    {
      "epoch": 18.62,
      "grad_norm": 12.530813217163086,
      "learning_rate": 9.37937937937938e-06,
      "loss": 0.4906,
      "step": 6200
    },
    {
      "epoch": 18.65,
      "grad_norm": 10.715028762817383,
      "learning_rate": 9.378378378378379e-06,
      "loss": 0.4887,
      "step": 6210
    },
    {
      "epoch": 18.68,
      "grad_norm": 14.905635833740234,
      "learning_rate": 9.377377377377378e-06,
      "loss": 0.5427,
      "step": 6220
    },
    {
      "epoch": 18.71,
      "grad_norm": 12.105393409729004,
      "learning_rate": 9.376376376376378e-06,
      "loss": 0.4693,
      "step": 6230
    },
    {
      "epoch": 18.74,
      "grad_norm": 14.1998929977417,
      "learning_rate": 9.375375375375376e-06,
      "loss": 0.4998,
      "step": 6240
    },
    {
      "epoch": 18.77,
      "grad_norm": 11.141003608703613,
      "learning_rate": 9.374374374374375e-06,
      "loss": 0.5454,
      "step": 6250
    },
    {
      "epoch": 18.8,
      "grad_norm": 13.6138916015625,
      "learning_rate": 9.373373373373375e-06,
      "loss": 0.5684,
      "step": 6260
    },
    {
      "epoch": 18.83,
      "grad_norm": 9.682354927062988,
      "learning_rate": 9.372372372372374e-06,
      "loss": 0.5221,
      "step": 6270
    },
    {
      "epoch": 18.86,
      "grad_norm": 10.344449043273926,
      "learning_rate": 9.371371371371372e-06,
      "loss": 0.4974,
      "step": 6280
    },
    {
      "epoch": 18.89,
      "grad_norm": 10.217352867126465,
      "learning_rate": 9.370370370370371e-06,
      "loss": 0.429,
      "step": 6290
    },
    {
      "epoch": 18.92,
      "grad_norm": 15.496273040771484,
      "learning_rate": 9.36936936936937e-06,
      "loss": 0.4606,
      "step": 6300
    },
    {
      "epoch": 18.95,
      "grad_norm": 9.484021186828613,
      "learning_rate": 9.368368368368368e-06,
      "loss": 0.5258,
      "step": 6310
    },
    {
      "epoch": 18.98,
      "grad_norm": 11.742782592773438,
      "learning_rate": 9.367367367367369e-06,
      "loss": 0.413,
      "step": 6320
    },
    {
      "epoch": 19.0,
      "eval_accuracy": 0.9271,
      "eval_loss": 0.2516428232192993,
      "eval_runtime": 30.2548,
      "eval_samples_per_second": 330.526,
      "eval_steps_per_second": 1.322,
      "step": 6327
    },
    {
      "epoch": 19.01,
      "grad_norm": 12.082165718078613,
      "learning_rate": 9.366366366366367e-06,
      "loss": 0.5247,
      "step": 6330
    },
    {
      "epoch": 19.04,
      "grad_norm": 32.512901306152344,
      "learning_rate": 9.365365365365366e-06,
      "loss": 0.5459,
      "step": 6340
    },
    {
      "epoch": 19.07,
      "grad_norm": 19.557708740234375,
      "learning_rate": 9.364364364364366e-06,
      "loss": 0.4909,
      "step": 6350
    },
    {
      "epoch": 19.1,
      "grad_norm": 14.893004417419434,
      "learning_rate": 9.363363363363363e-06,
      "loss": 0.4836,
      "step": 6360
    },
    {
      "epoch": 19.13,
      "grad_norm": 16.80282974243164,
      "learning_rate": 9.362362362362363e-06,
      "loss": 0.5214,
      "step": 6370
    },
    {
      "epoch": 19.16,
      "grad_norm": 9.32387638092041,
      "learning_rate": 9.361361361361362e-06,
      "loss": 0.4313,
      "step": 6380
    },
    {
      "epoch": 19.19,
      "grad_norm": 10.158374786376953,
      "learning_rate": 9.36036036036036e-06,
      "loss": 0.5065,
      "step": 6390
    },
    {
      "epoch": 19.22,
      "grad_norm": 15.345830917358398,
      "learning_rate": 9.35935935935936e-06,
      "loss": 0.5007,
      "step": 6400
    },
    {
      "epoch": 19.25,
      "grad_norm": 21.002599716186523,
      "learning_rate": 9.35835835835836e-06,
      "loss": 0.5401,
      "step": 6410
    },
    {
      "epoch": 19.28,
      "grad_norm": 40.8951416015625,
      "learning_rate": 9.357357357357358e-06,
      "loss": 0.4987,
      "step": 6420
    },
    {
      "epoch": 19.31,
      "grad_norm": 6.890993595123291,
      "learning_rate": 9.356356356356358e-06,
      "loss": 0.5339,
      "step": 6430
    },
    {
      "epoch": 19.34,
      "grad_norm": 10.289228439331055,
      "learning_rate": 9.355355355355357e-06,
      "loss": 0.4809,
      "step": 6440
    },
    {
      "epoch": 19.37,
      "grad_norm": 13.26909351348877,
      "learning_rate": 9.354354354354355e-06,
      "loss": 0.5463,
      "step": 6450
    },
    {
      "epoch": 19.4,
      "grad_norm": 14.539989471435547,
      "learning_rate": 9.353353353353354e-06,
      "loss": 0.5098,
      "step": 6460
    },
    {
      "epoch": 19.43,
      "grad_norm": 10.325398445129395,
      "learning_rate": 9.352352352352353e-06,
      "loss": 0.4576,
      "step": 6470
    },
    {
      "epoch": 19.46,
      "grad_norm": 6.7183518409729,
      "learning_rate": 9.351351351351353e-06,
      "loss": 0.4559,
      "step": 6480
    },
    {
      "epoch": 19.49,
      "grad_norm": 16.46378517150879,
      "learning_rate": 9.350350350350351e-06,
      "loss": 0.5428,
      "step": 6490
    },
    {
      "epoch": 19.52,
      "grad_norm": 93.92578887939453,
      "learning_rate": 9.34934934934935e-06,
      "loss": 0.484,
      "step": 6500
    },
    {
      "epoch": 19.55,
      "grad_norm": 8.785821914672852,
      "learning_rate": 9.34834834834835e-06,
      "loss": 0.5488,
      "step": 6510
    },
    {
      "epoch": 19.58,
      "grad_norm": 11.448771476745605,
      "learning_rate": 9.347347347347349e-06,
      "loss": 0.4804,
      "step": 6520
    },
    {
      "epoch": 19.61,
      "grad_norm": 8.699326515197754,
      "learning_rate": 9.346346346346346e-06,
      "loss": 0.4397,
      "step": 6530
    },
    {
      "epoch": 19.64,
      "grad_norm": 109.07538604736328,
      "learning_rate": 9.345345345345346e-06,
      "loss": 0.5142,
      "step": 6540
    },
    {
      "epoch": 19.67,
      "grad_norm": 25.838956832885742,
      "learning_rate": 9.344344344344345e-06,
      "loss": 0.4915,
      "step": 6550
    },
    {
      "epoch": 19.7,
      "grad_norm": 14.016207695007324,
      "learning_rate": 9.343343343343343e-06,
      "loss": 0.502,
      "step": 6560
    },
    {
      "epoch": 19.73,
      "grad_norm": 10.681465148925781,
      "learning_rate": 9.342342342342344e-06,
      "loss": 0.4819,
      "step": 6570
    },
    {
      "epoch": 19.76,
      "grad_norm": 14.305381774902344,
      "learning_rate": 9.341341341341342e-06,
      "loss": 0.4675,
      "step": 6580
    },
    {
      "epoch": 19.79,
      "grad_norm": 8.296285629272461,
      "learning_rate": 9.34034034034034e-06,
      "loss": 0.4814,
      "step": 6590
    },
    {
      "epoch": 19.82,
      "grad_norm": 28.747377395629883,
      "learning_rate": 9.339339339339341e-06,
      "loss": 0.5212,
      "step": 6600
    },
    {
      "epoch": 19.85,
      "grad_norm": 9.720871925354004,
      "learning_rate": 9.338338338338338e-06,
      "loss": 0.5133,
      "step": 6610
    },
    {
      "epoch": 19.88,
      "grad_norm": 12.930212020874023,
      "learning_rate": 9.337337337337338e-06,
      "loss": 0.5087,
      "step": 6620
    },
    {
      "epoch": 19.91,
      "grad_norm": 12.70535945892334,
      "learning_rate": 9.336336336336337e-06,
      "loss": 0.497,
      "step": 6630
    },
    {
      "epoch": 19.94,
      "grad_norm": 22.562301635742188,
      "learning_rate": 9.335335335335335e-06,
      "loss": 0.4719,
      "step": 6640
    },
    {
      "epoch": 19.97,
      "grad_norm": 10.32685375213623,
      "learning_rate": 9.334334334334336e-06,
      "loss": 0.4814,
      "step": 6650
    },
    {
      "epoch": 20.0,
      "grad_norm": 26.123905181884766,
      "learning_rate": 9.333333333333334e-06,
      "loss": 0.4612,
      "step": 6660
    },
    {
      "epoch": 20.0,
      "eval_accuracy": 0.9242,
      "eval_loss": 0.2537807822227478,
      "eval_runtime": 30.8556,
      "eval_samples_per_second": 324.09,
      "eval_steps_per_second": 1.296,
      "step": 6660
    },
    {
      "epoch": 20.03,
      "grad_norm": 9.324190139770508,
      "learning_rate": 9.332332332332333e-06,
      "loss": 0.4782,
      "step": 6670
    },
    {
      "epoch": 20.06,
      "grad_norm": 8.56516170501709,
      "learning_rate": 9.331331331331333e-06,
      "loss": 0.4571,
      "step": 6680
    },
    {
      "epoch": 20.09,
      "grad_norm": 10.29270076751709,
      "learning_rate": 9.330330330330332e-06,
      "loss": 0.5438,
      "step": 6690
    },
    {
      "epoch": 20.12,
      "grad_norm": 8.01328182220459,
      "learning_rate": 9.32932932932933e-06,
      "loss": 0.4622,
      "step": 6700
    },
    {
      "epoch": 20.15,
      "grad_norm": 12.914417266845703,
      "learning_rate": 9.328328328328329e-06,
      "loss": 0.403,
      "step": 6710
    },
    {
      "epoch": 20.18,
      "grad_norm": 13.325395584106445,
      "learning_rate": 9.327327327327328e-06,
      "loss": 0.4534,
      "step": 6720
    },
    {
      "epoch": 20.21,
      "grad_norm": 11.514555931091309,
      "learning_rate": 9.326326326326328e-06,
      "loss": 0.4949,
      "step": 6730
    },
    {
      "epoch": 20.24,
      "grad_norm": 8.554108619689941,
      "learning_rate": 9.325325325325326e-06,
      "loss": 0.4979,
      "step": 6740
    },
    {
      "epoch": 20.27,
      "grad_norm": 6.8894362449646,
      "learning_rate": 9.324324324324325e-06,
      "loss": 0.5234,
      "step": 6750
    },
    {
      "epoch": 20.3,
      "grad_norm": 36.356204986572266,
      "learning_rate": 9.323323323323324e-06,
      "loss": 0.4435,
      "step": 6760
    },
    {
      "epoch": 20.33,
      "grad_norm": 10.051351547241211,
      "learning_rate": 9.322322322322324e-06,
      "loss": 0.4812,
      "step": 6770
    },
    {
      "epoch": 20.36,
      "grad_norm": 7.804532527923584,
      "learning_rate": 9.321321321321321e-06,
      "loss": 0.5634,
      "step": 6780
    },
    {
      "epoch": 20.39,
      "grad_norm": 11.409031867980957,
      "learning_rate": 9.320320320320321e-06,
      "loss": 0.4961,
      "step": 6790
    },
    {
      "epoch": 20.42,
      "grad_norm": 12.909208297729492,
      "learning_rate": 9.31931931931932e-06,
      "loss": 0.5435,
      "step": 6800
    },
    {
      "epoch": 20.45,
      "grad_norm": 9.927260398864746,
      "learning_rate": 9.318318318318318e-06,
      "loss": 0.5077,
      "step": 6810
    },
    {
      "epoch": 20.48,
      "grad_norm": 16.0677490234375,
      "learning_rate": 9.317317317317319e-06,
      "loss": 0.5106,
      "step": 6820
    },
    {
      "epoch": 20.51,
      "grad_norm": 10.34533405303955,
      "learning_rate": 9.316316316316317e-06,
      "loss": 0.4648,
      "step": 6830
    },
    {
      "epoch": 20.54,
      "grad_norm": 8.697833061218262,
      "learning_rate": 9.315315315315316e-06,
      "loss": 0.4776,
      "step": 6840
    },
    {
      "epoch": 20.57,
      "grad_norm": 11.206692695617676,
      "learning_rate": 9.314314314314316e-06,
      "loss": 0.4511,
      "step": 6850
    },
    {
      "epoch": 20.6,
      "grad_norm": 18.051631927490234,
      "learning_rate": 9.313313313313313e-06,
      "loss": 0.5444,
      "step": 6860
    },
    {
      "epoch": 20.63,
      "grad_norm": 11.191703796386719,
      "learning_rate": 9.312312312312313e-06,
      "loss": 0.3936,
      "step": 6870
    },
    {
      "epoch": 20.66,
      "grad_norm": 17.013591766357422,
      "learning_rate": 9.311311311311312e-06,
      "loss": 0.4638,
      "step": 6880
    },
    {
      "epoch": 20.69,
      "grad_norm": 9.712099075317383,
      "learning_rate": 9.31031031031031e-06,
      "loss": 0.5224,
      "step": 6890
    },
    {
      "epoch": 20.72,
      "grad_norm": 8.011286735534668,
      "learning_rate": 9.30930930930931e-06,
      "loss": 0.4889,
      "step": 6900
    },
    {
      "epoch": 20.75,
      "grad_norm": 11.292800903320312,
      "learning_rate": 9.30830830830831e-06,
      "loss": 0.4924,
      "step": 6910
    },
    {
      "epoch": 20.78,
      "grad_norm": 12.472173690795898,
      "learning_rate": 9.307307307307308e-06,
      "loss": 0.4625,
      "step": 6920
    },
    {
      "epoch": 20.81,
      "grad_norm": 8.170212745666504,
      "learning_rate": 9.306306306306308e-06,
      "loss": 0.4415,
      "step": 6930
    },
    {
      "epoch": 20.84,
      "grad_norm": 20.078588485717773,
      "learning_rate": 9.305305305305305e-06,
      "loss": 0.4981,
      "step": 6940
    },
    {
      "epoch": 20.87,
      "grad_norm": 11.406576156616211,
      "learning_rate": 9.304304304304305e-06,
      "loss": 0.5696,
      "step": 6950
    },
    {
      "epoch": 20.9,
      "grad_norm": 10.49374008178711,
      "learning_rate": 9.303303303303304e-06,
      "loss": 0.5004,
      "step": 6960
    },
    {
      "epoch": 20.93,
      "grad_norm": 9.33711051940918,
      "learning_rate": 9.302302302302303e-06,
      "loss": 0.5085,
      "step": 6970
    },
    {
      "epoch": 20.96,
      "grad_norm": 13.773951530456543,
      "learning_rate": 9.301301301301301e-06,
      "loss": 0.5632,
      "step": 6980
    },
    {
      "epoch": 20.99,
      "grad_norm": 9.44569206237793,
      "learning_rate": 9.300300300300302e-06,
      "loss": 0.4903,
      "step": 6990
    },
    {
      "epoch": 21.0,
      "eval_accuracy": 0.9262,
      "eval_loss": 0.2555881142616272,
      "eval_runtime": 30.7701,
      "eval_samples_per_second": 324.99,
      "eval_steps_per_second": 1.3,
      "step": 6993
    },
    {
      "epoch": 21.02,
      "grad_norm": 17.463153839111328,
      "learning_rate": 9.2992992992993e-06,
      "loss": 0.5627,
      "step": 7000
    },
    {
      "epoch": 21.05,
      "grad_norm": 13.008602142333984,
      "learning_rate": 9.298298298298299e-06,
      "loss": 0.5099,
      "step": 7010
    },
    {
      "epoch": 21.08,
      "grad_norm": 9.015759468078613,
      "learning_rate": 9.297297297297299e-06,
      "loss": 0.4618,
      "step": 7020
    },
    {
      "epoch": 21.11,
      "grad_norm": 12.224336624145508,
      "learning_rate": 9.296296296296296e-06,
      "loss": 0.4893,
      "step": 7030
    },
    {
      "epoch": 21.14,
      "grad_norm": 14.461615562438965,
      "learning_rate": 9.295295295295296e-06,
      "loss": 0.4554,
      "step": 7040
    },
    {
      "epoch": 21.17,
      "grad_norm": 14.201723098754883,
      "learning_rate": 9.294294294294295e-06,
      "loss": 0.4771,
      "step": 7050
    },
    {
      "epoch": 21.2,
      "grad_norm": 7.477570533752441,
      "learning_rate": 9.293293293293293e-06,
      "loss": 0.4875,
      "step": 7060
    },
    {
      "epoch": 21.23,
      "grad_norm": 9.28654956817627,
      "learning_rate": 9.292292292292294e-06,
      "loss": 0.4739,
      "step": 7070
    },
    {
      "epoch": 21.26,
      "grad_norm": 9.934181213378906,
      "learning_rate": 9.291291291291292e-06,
      "loss": 0.5152,
      "step": 7080
    },
    {
      "epoch": 21.29,
      "grad_norm": 18.241933822631836,
      "learning_rate": 9.29029029029029e-06,
      "loss": 0.535,
      "step": 7090
    },
    {
      "epoch": 21.32,
      "grad_norm": 8.119966506958008,
      "learning_rate": 9.289289289289291e-06,
      "loss": 0.4178,
      "step": 7100
    },
    {
      "epoch": 21.35,
      "grad_norm": 20.503637313842773,
      "learning_rate": 9.288288288288288e-06,
      "loss": 0.5279,
      "step": 7110
    },
    {
      "epoch": 21.38,
      "grad_norm": 13.679118156433105,
      "learning_rate": 9.287287287287288e-06,
      "loss": 0.4412,
      "step": 7120
    },
    {
      "epoch": 21.41,
      "grad_norm": 10.256949424743652,
      "learning_rate": 9.286286286286287e-06,
      "loss": 0.5258,
      "step": 7130
    },
    {
      "epoch": 21.44,
      "grad_norm": 14.46385383605957,
      "learning_rate": 9.285285285285286e-06,
      "loss": 0.5184,
      "step": 7140
    },
    {
      "epoch": 21.47,
      "grad_norm": 8.546163558959961,
      "learning_rate": 9.284284284284286e-06,
      "loss": 0.4498,
      "step": 7150
    },
    {
      "epoch": 21.5,
      "grad_norm": 13.359642028808594,
      "learning_rate": 9.283283283283284e-06,
      "loss": 0.4551,
      "step": 7160
    },
    {
      "epoch": 21.53,
      "grad_norm": 8.469135284423828,
      "learning_rate": 9.282282282282283e-06,
      "loss": 0.5594,
      "step": 7170
    },
    {
      "epoch": 21.56,
      "grad_norm": 11.59361457824707,
      "learning_rate": 9.281281281281283e-06,
      "loss": 0.4551,
      "step": 7180
    },
    {
      "epoch": 21.59,
      "grad_norm": 10.340875625610352,
      "learning_rate": 9.28028028028028e-06,
      "loss": 0.5166,
      "step": 7190
    },
    {
      "epoch": 21.62,
      "grad_norm": 14.073348045349121,
      "learning_rate": 9.27927927927928e-06,
      "loss": 0.4156,
      "step": 7200
    },
    {
      "epoch": 21.65,
      "grad_norm": 17.552919387817383,
      "learning_rate": 9.278278278278279e-06,
      "loss": 0.5288,
      "step": 7210
    },
    {
      "epoch": 21.68,
      "grad_norm": 8.071002006530762,
      "learning_rate": 9.277277277277278e-06,
      "loss": 0.4867,
      "step": 7220
    },
    {
      "epoch": 21.71,
      "grad_norm": 10.877429008483887,
      "learning_rate": 9.276276276276276e-06,
      "loss": 0.4382,
      "step": 7230
    },
    {
      "epoch": 21.74,
      "grad_norm": 16.8603458404541,
      "learning_rate": 9.275275275275277e-06,
      "loss": 0.4981,
      "step": 7240
    },
    {
      "epoch": 21.77,
      "grad_norm": 16.255592346191406,
      "learning_rate": 9.274274274274275e-06,
      "loss": 0.4993,
      "step": 7250
    },
    {
      "epoch": 21.8,
      "grad_norm": 12.949196815490723,
      "learning_rate": 9.273273273273274e-06,
      "loss": 0.5542,
      "step": 7260
    },
    {
      "epoch": 21.83,
      "grad_norm": 12.743133544921875,
      "learning_rate": 9.272272272272274e-06,
      "loss": 0.5043,
      "step": 7270
    },
    {
      "epoch": 21.86,
      "grad_norm": 13.237237930297852,
      "learning_rate": 9.271271271271271e-06,
      "loss": 0.505,
      "step": 7280
    },
    {
      "epoch": 21.89,
      "grad_norm": 9.304459571838379,
      "learning_rate": 9.270270270270271e-06,
      "loss": 0.4848,
      "step": 7290
    },
    {
      "epoch": 21.92,
      "grad_norm": 12.196815490722656,
      "learning_rate": 9.26926926926927e-06,
      "loss": 0.4643,
      "step": 7300
    },
    {
      "epoch": 21.95,
      "grad_norm": 6.767624378204346,
      "learning_rate": 9.268268268268268e-06,
      "loss": 0.541,
      "step": 7310
    },
    {
      "epoch": 21.98,
      "grad_norm": 12.607462882995605,
      "learning_rate": 9.267267267267269e-06,
      "loss": 0.4953,
      "step": 7320
    },
    {
      "epoch": 22.0,
      "eval_accuracy": 0.9271,
      "eval_loss": 0.2500782608985901,
      "eval_runtime": 30.4081,
      "eval_samples_per_second": 328.859,
      "eval_steps_per_second": 1.315,
      "step": 7326
    },
    {
      "epoch": 22.01,
      "grad_norm": 7.150274276733398,
      "learning_rate": 9.266266266266267e-06,
      "loss": 0.5962,
      "step": 7330
    },
    {
      "epoch": 22.04,
      "grad_norm": 13.817320823669434,
      "learning_rate": 9.265265265265266e-06,
      "loss": 0.4671,
      "step": 7340
    },
    {
      "epoch": 22.07,
      "grad_norm": 9.749372482299805,
      "learning_rate": 9.264264264264266e-06,
      "loss": 0.4525,
      "step": 7350
    },
    {
      "epoch": 22.1,
      "grad_norm": 17.435985565185547,
      "learning_rate": 9.263263263263263e-06,
      "loss": 0.4505,
      "step": 7360
    },
    {
      "epoch": 22.13,
      "grad_norm": 25.06301498413086,
      "learning_rate": 9.262262262262263e-06,
      "loss": 0.4748,
      "step": 7370
    },
    {
      "epoch": 22.16,
      "grad_norm": 7.567307949066162,
      "learning_rate": 9.261261261261262e-06,
      "loss": 0.4632,
      "step": 7380
    },
    {
      "epoch": 22.19,
      "grad_norm": 10.676952362060547,
      "learning_rate": 9.26026026026026e-06,
      "loss": 0.466,
      "step": 7390
    },
    {
      "epoch": 22.22,
      "grad_norm": 9.568814277648926,
      "learning_rate": 9.25925925925926e-06,
      "loss": 0.4918,
      "step": 7400
    },
    {
      "epoch": 22.25,
      "grad_norm": 11.789005279541016,
      "learning_rate": 9.25825825825826e-06,
      "loss": 0.5194,
      "step": 7410
    },
    {
      "epoch": 22.28,
      "grad_norm": 10.343085289001465,
      "learning_rate": 9.257257257257258e-06,
      "loss": 0.4521,
      "step": 7420
    },
    {
      "epoch": 22.31,
      "grad_norm": 8.934516906738281,
      "learning_rate": 9.256256256256257e-06,
      "loss": 0.4829,
      "step": 7430
    },
    {
      "epoch": 22.34,
      "grad_norm": 8.359495162963867,
      "learning_rate": 9.255255255255255e-06,
      "loss": 0.4337,
      "step": 7440
    },
    {
      "epoch": 22.37,
      "grad_norm": 10.556571006774902,
      "learning_rate": 9.254254254254254e-06,
      "loss": 0.483,
      "step": 7450
    },
    {
      "epoch": 22.4,
      "grad_norm": 14.519379615783691,
      "learning_rate": 9.253253253253254e-06,
      "loss": 0.4496,
      "step": 7460
    },
    {
      "epoch": 22.43,
      "grad_norm": 18.63585662841797,
      "learning_rate": 9.252252252252253e-06,
      "loss": 0.5033,
      "step": 7470
    },
    {
      "epoch": 22.46,
      "grad_norm": 10.354231834411621,
      "learning_rate": 9.251251251251251e-06,
      "loss": 0.5256,
      "step": 7480
    },
    {
      "epoch": 22.49,
      "grad_norm": 10.034125328063965,
      "learning_rate": 9.250250250250252e-06,
      "loss": 0.4682,
      "step": 7490
    },
    {
      "epoch": 22.52,
      "grad_norm": 10.081664085388184,
      "learning_rate": 9.24924924924925e-06,
      "loss": 0.4549,
      "step": 7500
    },
    {
      "epoch": 22.55,
      "grad_norm": 12.067422866821289,
      "learning_rate": 9.248248248248249e-06,
      "loss": 0.5043,
      "step": 7510
    },
    {
      "epoch": 22.58,
      "grad_norm": 8.950556755065918,
      "learning_rate": 9.247247247247249e-06,
      "loss": 0.4407,
      "step": 7520
    },
    {
      "epoch": 22.61,
      "grad_norm": 9.73210334777832,
      "learning_rate": 9.246246246246246e-06,
      "loss": 0.4622,
      "step": 7530
    },
    {
      "epoch": 22.64,
      "grad_norm": 9.844476699829102,
      "learning_rate": 9.245245245245246e-06,
      "loss": 0.4713,
      "step": 7540
    },
    {
      "epoch": 22.67,
      "grad_norm": 10.3902587890625,
      "learning_rate": 9.244244244244245e-06,
      "loss": 0.4744,
      "step": 7550
    },
    {
      "epoch": 22.7,
      "grad_norm": 9.012811660766602,
      "learning_rate": 9.243243243243243e-06,
      "loss": 0.5062,
      "step": 7560
    },
    {
      "epoch": 22.73,
      "grad_norm": 7.681649684906006,
      "learning_rate": 9.242242242242244e-06,
      "loss": 0.4653,
      "step": 7570
    },
    {
      "epoch": 22.76,
      "grad_norm": 7.8287506103515625,
      "learning_rate": 9.241241241241242e-06,
      "loss": 0.5323,
      "step": 7580
    },
    {
      "epoch": 22.79,
      "grad_norm": 20.302045822143555,
      "learning_rate": 9.240240240240241e-06,
      "loss": 0.4863,
      "step": 7590
    },
    {
      "epoch": 22.82,
      "grad_norm": 8.254586219787598,
      "learning_rate": 9.239239239239241e-06,
      "loss": 0.5281,
      "step": 7600
    },
    {
      "epoch": 22.85,
      "grad_norm": 18.055051803588867,
      "learning_rate": 9.238238238238238e-06,
      "loss": 0.4896,
      "step": 7610
    },
    {
      "epoch": 22.88,
      "grad_norm": 12.08217716217041,
      "learning_rate": 9.237237237237238e-06,
      "loss": 0.5,
      "step": 7620
    },
    {
      "epoch": 22.91,
      "grad_norm": 8.12684440612793,
      "learning_rate": 9.236236236236237e-06,
      "loss": 0.449,
      "step": 7630
    },
    {
      "epoch": 22.94,
      "grad_norm": 10.148077011108398,
      "learning_rate": 9.235235235235236e-06,
      "loss": 0.5071,
      "step": 7640
    },
    {
      "epoch": 22.97,
      "grad_norm": 8.919116973876953,
      "learning_rate": 9.234234234234236e-06,
      "loss": 0.4922,
      "step": 7650
    },
    {
      "epoch": 23.0,
      "eval_accuracy": 0.9277,
      "eval_loss": 0.24855679273605347,
      "eval_runtime": 30.7483,
      "eval_samples_per_second": 325.221,
      "eval_steps_per_second": 1.301,
      "step": 7659
    },
    {
      "epoch": 23.0,
      "grad_norm": 7.698756217956543,
      "learning_rate": 9.233233233233234e-06,
      "loss": 0.4575,
      "step": 7660
    },
    {
      "epoch": 23.03,
      "grad_norm": 12.64100170135498,
      "learning_rate": 9.232232232232233e-06,
      "loss": 0.4228,
      "step": 7670
    },
    {
      "epoch": 23.06,
      "grad_norm": 14.073904991149902,
      "learning_rate": 9.231231231231232e-06,
      "loss": 0.5228,
      "step": 7680
    },
    {
      "epoch": 23.09,
      "grad_norm": 13.450045585632324,
      "learning_rate": 9.23023023023023e-06,
      "loss": 0.4917,
      "step": 7690
    },
    {
      "epoch": 23.12,
      "grad_norm": 9.442963600158691,
      "learning_rate": 9.229229229229229e-06,
      "loss": 0.4431,
      "step": 7700
    },
    {
      "epoch": 23.15,
      "grad_norm": 12.14179515838623,
      "learning_rate": 9.228228228228229e-06,
      "loss": 0.4929,
      "step": 7710
    },
    {
      "epoch": 23.18,
      "grad_norm": 15.105382919311523,
      "learning_rate": 9.227227227227228e-06,
      "loss": 0.4966,
      "step": 7720
    },
    {
      "epoch": 23.21,
      "grad_norm": 15.966352462768555,
      "learning_rate": 9.226226226226226e-06,
      "loss": 0.4768,
      "step": 7730
    },
    {
      "epoch": 23.24,
      "grad_norm": 17.863000869750977,
      "learning_rate": 9.225225225225227e-06,
      "loss": 0.4558,
      "step": 7740
    },
    {
      "epoch": 23.27,
      "grad_norm": 12.311907768249512,
      "learning_rate": 9.224224224224225e-06,
      "loss": 0.4995,
      "step": 7750
    },
    {
      "epoch": 23.3,
      "grad_norm": 17.10275650024414,
      "learning_rate": 9.223223223223224e-06,
      "loss": 0.4514,
      "step": 7760
    },
    {
      "epoch": 23.33,
      "grad_norm": 17.76338005065918,
      "learning_rate": 9.222222222222224e-06,
      "loss": 0.4349,
      "step": 7770
    },
    {
      "epoch": 23.36,
      "grad_norm": 16.42094612121582,
      "learning_rate": 9.221221221221221e-06,
      "loss": 0.4812,
      "step": 7780
    },
    {
      "epoch": 23.39,
      "grad_norm": 14.530420303344727,
      "learning_rate": 9.220220220220221e-06,
      "loss": 0.4498,
      "step": 7790
    },
    {
      "epoch": 23.42,
      "grad_norm": 14.333878517150879,
      "learning_rate": 9.21921921921922e-06,
      "loss": 0.4657,
      "step": 7800
    },
    {
      "epoch": 23.45,
      "grad_norm": 29.57023048400879,
      "learning_rate": 9.218218218218218e-06,
      "loss": 0.46,
      "step": 7810
    },
    {
      "epoch": 23.48,
      "grad_norm": 12.47533130645752,
      "learning_rate": 9.217217217217219e-06,
      "loss": 0.4458,
      "step": 7820
    },
    {
      "epoch": 23.51,
      "grad_norm": 17.979331970214844,
      "learning_rate": 9.216216216216217e-06,
      "loss": 0.4812,
      "step": 7830
    },
    {
      "epoch": 23.54,
      "grad_norm": 33.53706741333008,
      "learning_rate": 9.215215215215216e-06,
      "loss": 0.4412,
      "step": 7840
    },
    {
      "epoch": 23.57,
      "grad_norm": 9.694183349609375,
      "learning_rate": 9.214214214214216e-06,
      "loss": 0.5278,
      "step": 7850
    },
    {
      "epoch": 23.6,
      "grad_norm": 10.997142791748047,
      "learning_rate": 9.213213213213213e-06,
      "loss": 0.3873,
      "step": 7860
    },
    {
      "epoch": 23.63,
      "grad_norm": 14.332605361938477,
      "learning_rate": 9.212212212212213e-06,
      "loss": 0.4381,
      "step": 7870
    },
    {
      "epoch": 23.66,
      "grad_norm": 21.167810440063477,
      "learning_rate": 9.211211211211212e-06,
      "loss": 0.4479,
      "step": 7880
    },
    {
      "epoch": 23.69,
      "grad_norm": 11.316388130187988,
      "learning_rate": 9.21021021021021e-06,
      "loss": 0.4813,
      "step": 7890
    },
    {
      "epoch": 23.72,
      "grad_norm": 10.724557876586914,
      "learning_rate": 9.20920920920921e-06,
      "loss": 0.4825,
      "step": 7900
    },
    {
      "epoch": 23.75,
      "grad_norm": 8.695270538330078,
      "learning_rate": 9.20820820820821e-06,
      "loss": 0.4389,
      "step": 7910
    },
    {
      "epoch": 23.78,
      "grad_norm": 8.966339111328125,
      "learning_rate": 9.207207207207208e-06,
      "loss": 0.4852,
      "step": 7920
    },
    {
      "epoch": 23.81,
      "grad_norm": 9.002543449401855,
      "learning_rate": 9.206206206206207e-06,
      "loss": 0.4548,
      "step": 7930
    },
    {
      "epoch": 23.84,
      "grad_norm": 7.919975280761719,
      "learning_rate": 9.205205205205205e-06,
      "loss": 0.5072,
      "step": 7940
    },
    {
      "epoch": 23.87,
      "grad_norm": 10.650671005249023,
      "learning_rate": 9.204204204204204e-06,
      "loss": 0.4615,
      "step": 7950
    },
    {
      "epoch": 23.9,
      "grad_norm": 17.29688262939453,
      "learning_rate": 9.203203203203204e-06,
      "loss": 0.3928,
      "step": 7960
    },
    {
      "epoch": 23.93,
      "grad_norm": 9.760119438171387,
      "learning_rate": 9.202202202202203e-06,
      "loss": 0.5134,
      "step": 7970
    },
    {
      "epoch": 23.96,
      "grad_norm": 11.86888313293457,
      "learning_rate": 9.201201201201201e-06,
      "loss": 0.4764,
      "step": 7980
    },
    {
      "epoch": 23.99,
      "grad_norm": 47.17730712890625,
      "learning_rate": 9.200200200200202e-06,
      "loss": 0.4603,
      "step": 7990
    },
    {
      "epoch": 24.0,
      "eval_accuracy": 0.9234,
      "eval_loss": 0.25501587986946106,
      "eval_runtime": 30.8318,
      "eval_samples_per_second": 324.341,
      "eval_steps_per_second": 1.297,
      "step": 7992
    },
    {
      "epoch": 24.02,
      "grad_norm": 10.768325805664062,
      "learning_rate": 9.1991991991992e-06,
      "loss": 0.4188,
      "step": 8000
    },
    {
      "epoch": 24.05,
      "grad_norm": 14.675819396972656,
      "learning_rate": 9.198198198198199e-06,
      "loss": 0.4144,
      "step": 8010
    },
    {
      "epoch": 24.08,
      "grad_norm": 8.436361312866211,
      "learning_rate": 9.197197197197199e-06,
      "loss": 0.4573,
      "step": 8020
    },
    {
      "epoch": 24.11,
      "grad_norm": 10.41925048828125,
      "learning_rate": 9.196196196196196e-06,
      "loss": 0.4338,
      "step": 8030
    },
    {
      "epoch": 24.14,
      "grad_norm": 15.963869094848633,
      "learning_rate": 9.195195195195196e-06,
      "loss": 0.4744,
      "step": 8040
    },
    {
      "epoch": 24.17,
      "grad_norm": 14.007654190063477,
      "learning_rate": 9.194194194194195e-06,
      "loss": 0.4865,
      "step": 8050
    },
    {
      "epoch": 24.2,
      "grad_norm": 11.407424926757812,
      "learning_rate": 9.193193193193194e-06,
      "loss": 0.4371,
      "step": 8060
    },
    {
      "epoch": 24.23,
      "grad_norm": 41.07179641723633,
      "learning_rate": 9.192192192192194e-06,
      "loss": 0.4536,
      "step": 8070
    },
    {
      "epoch": 24.26,
      "grad_norm": 8.165257453918457,
      "learning_rate": 9.191191191191192e-06,
      "loss": 0.4284,
      "step": 8080
    },
    {
      "epoch": 24.29,
      "grad_norm": 9.9638671875,
      "learning_rate": 9.190190190190191e-06,
      "loss": 0.4581,
      "step": 8090
    },
    {
      "epoch": 24.32,
      "grad_norm": 15.189397811889648,
      "learning_rate": 9.189189189189191e-06,
      "loss": 0.4933,
      "step": 8100
    },
    {
      "epoch": 24.35,
      "grad_norm": 12.482817649841309,
      "learning_rate": 9.188188188188188e-06,
      "loss": 0.4576,
      "step": 8110
    },
    {
      "epoch": 24.38,
      "grad_norm": 14.716872215270996,
      "learning_rate": 9.187187187187187e-06,
      "loss": 0.4888,
      "step": 8120
    },
    {
      "epoch": 24.41,
      "grad_norm": 17.546016693115234,
      "learning_rate": 9.186186186186187e-06,
      "loss": 0.4797,
      "step": 8130
    },
    {
      "epoch": 24.44,
      "grad_norm": 12.303563117980957,
      "learning_rate": 9.185185185185186e-06,
      "loss": 0.5145,
      "step": 8140
    },
    {
      "epoch": 24.47,
      "grad_norm": 8.901636123657227,
      "learning_rate": 9.184184184184184e-06,
      "loss": 0.4398,
      "step": 8150
    },
    {
      "epoch": 24.5,
      "grad_norm": 13.526920318603516,
      "learning_rate": 9.183183183183185e-06,
      "loss": 0.4182,
      "step": 8160
    },
    {
      "epoch": 24.53,
      "grad_norm": 11.501439094543457,
      "learning_rate": 9.182182182182183e-06,
      "loss": 0.4582,
      "step": 8170
    },
    {
      "epoch": 24.56,
      "grad_norm": 10.772521018981934,
      "learning_rate": 9.181181181181182e-06,
      "loss": 0.4541,
      "step": 8180
    },
    {
      "epoch": 24.59,
      "grad_norm": 11.049300193786621,
      "learning_rate": 9.18018018018018e-06,
      "loss": 0.4674,
      "step": 8190
    },
    {
      "epoch": 24.62,
      "grad_norm": 19.82034683227539,
      "learning_rate": 9.179179179179179e-06,
      "loss": 0.4205,
      "step": 8200
    },
    {
      "epoch": 24.65,
      "grad_norm": 8.384696006774902,
      "learning_rate": 9.17817817817818e-06,
      "loss": 0.4723,
      "step": 8210
    },
    {
      "epoch": 24.68,
      "grad_norm": 11.282562255859375,
      "learning_rate": 9.177177177177178e-06,
      "loss": 0.4711,
      "step": 8220
    },
    {
      "epoch": 24.71,
      "grad_norm": 13.955788612365723,
      "learning_rate": 9.176176176176176e-06,
      "loss": 0.4702,
      "step": 8230
    },
    {
      "epoch": 24.74,
      "grad_norm": 5.679299831390381,
      "learning_rate": 9.175175175175177e-06,
      "loss": 0.4309,
      "step": 8240
    },
    {
      "epoch": 24.77,
      "grad_norm": 10.345995903015137,
      "learning_rate": 9.174174174174175e-06,
      "loss": 0.4701,
      "step": 8250
    },
    {
      "epoch": 24.8,
      "grad_norm": 20.478515625,
      "learning_rate": 9.173173173173174e-06,
      "loss": 0.4685,
      "step": 8260
    },
    {
      "epoch": 24.83,
      "grad_norm": 14.467691421508789,
      "learning_rate": 9.172172172172172e-06,
      "loss": 0.5148,
      "step": 8270
    },
    {
      "epoch": 24.86,
      "grad_norm": 8.202590942382812,
      "learning_rate": 9.171171171171171e-06,
      "loss": 0.3606,
      "step": 8280
    },
    {
      "epoch": 24.89,
      "grad_norm": 8.882107734680176,
      "learning_rate": 9.170170170170171e-06,
      "loss": 0.4808,
      "step": 8290
    },
    {
      "epoch": 24.92,
      "grad_norm": 14.242910385131836,
      "learning_rate": 9.16916916916917e-06,
      "loss": 0.4177,
      "step": 8300
    },
    {
      "epoch": 24.95,
      "grad_norm": 9.910883903503418,
      "learning_rate": 9.168168168168169e-06,
      "loss": 0.4853,
      "step": 8310
    },
    {
      "epoch": 24.98,
      "grad_norm": 10.825551986694336,
      "learning_rate": 9.167167167167169e-06,
      "loss": 0.4405,
      "step": 8320
    },
    {
      "epoch": 25.0,
      "eval_accuracy": 0.9285,
      "eval_loss": 0.2476295828819275,
      "eval_runtime": 30.5772,
      "eval_samples_per_second": 327.041,
      "eval_steps_per_second": 1.308,
      "step": 8325
    },
    {
      "epoch": 25.02,
      "grad_norm": 16.443073272705078,
      "learning_rate": 9.166166166166167e-06,
      "loss": 0.4394,
      "step": 8330
    },
    {
      "epoch": 25.05,
      "grad_norm": 13.089273452758789,
      "learning_rate": 9.165165165165166e-06,
      "loss": 0.4354,
      "step": 8340
    },
    {
      "epoch": 25.08,
      "grad_norm": 17.05743408203125,
      "learning_rate": 9.164164164164165e-06,
      "loss": 0.4766,
      "step": 8350
    },
    {
      "epoch": 25.11,
      "grad_norm": 17.6065673828125,
      "learning_rate": 9.163163163163163e-06,
      "loss": 0.4629,
      "step": 8360
    },
    {
      "epoch": 25.14,
      "grad_norm": 12.650850296020508,
      "learning_rate": 9.162162162162162e-06,
      "loss": 0.472,
      "step": 8370
    },
    {
      "epoch": 25.17,
      "grad_norm": 33.77656936645508,
      "learning_rate": 9.161161161161162e-06,
      "loss": 0.4871,
      "step": 8380
    },
    {
      "epoch": 25.2,
      "grad_norm": 10.210184097290039,
      "learning_rate": 9.16016016016016e-06,
      "loss": 0.4343,
      "step": 8390
    },
    {
      "epoch": 25.23,
      "grad_norm": 13.365535736083984,
      "learning_rate": 9.15915915915916e-06,
      "loss": 0.3632,
      "step": 8400
    },
    {
      "epoch": 25.26,
      "grad_norm": 8.703694343566895,
      "learning_rate": 9.15815815815816e-06,
      "loss": 0.4402,
      "step": 8410
    },
    {
      "epoch": 25.29,
      "grad_norm": 12.968696594238281,
      "learning_rate": 9.157157157157158e-06,
      "loss": 0.4308,
      "step": 8420
    },
    {
      "epoch": 25.32,
      "grad_norm": 13.030708312988281,
      "learning_rate": 9.156156156156157e-06,
      "loss": 0.4876,
      "step": 8430
    },
    {
      "epoch": 25.35,
      "grad_norm": 11.491731643676758,
      "learning_rate": 9.155155155155155e-06,
      "loss": 0.4824,
      "step": 8440
    },
    {
      "epoch": 25.38,
      "grad_norm": 15.222594261169434,
      "learning_rate": 9.154154154154154e-06,
      "loss": 0.4647,
      "step": 8450
    },
    {
      "epoch": 25.41,
      "grad_norm": 10.160212516784668,
      "learning_rate": 9.153153153153154e-06,
      "loss": 0.5069,
      "step": 8460
    },
    {
      "epoch": 25.44,
      "grad_norm": 9.530731201171875,
      "learning_rate": 9.152152152152153e-06,
      "loss": 0.4706,
      "step": 8470
    },
    {
      "epoch": 25.47,
      "grad_norm": 9.125948905944824,
      "learning_rate": 9.151151151151151e-06,
      "loss": 0.4115,
      "step": 8480
    },
    {
      "epoch": 25.5,
      "grad_norm": 15.237028121948242,
      "learning_rate": 9.150150150150152e-06,
      "loss": 0.503,
      "step": 8490
    },
    {
      "epoch": 25.53,
      "grad_norm": 10.005708694458008,
      "learning_rate": 9.14914914914915e-06,
      "loss": 0.4717,
      "step": 8500
    },
    {
      "epoch": 25.56,
      "grad_norm": 14.051669120788574,
      "learning_rate": 9.148148148148149e-06,
      "loss": 0.492,
      "step": 8510
    },
    {
      "epoch": 25.59,
      "grad_norm": 13.112764358520508,
      "learning_rate": 9.147147147147147e-06,
      "loss": 0.4358,
      "step": 8520
    },
    {
      "epoch": 25.62,
      "grad_norm": 26.472837448120117,
      "learning_rate": 9.146146146146146e-06,
      "loss": 0.498,
      "step": 8530
    },
    {
      "epoch": 25.65,
      "grad_norm": 10.999613761901855,
      "learning_rate": 9.145145145145146e-06,
      "loss": 0.4739,
      "step": 8540
    },
    {
      "epoch": 25.68,
      "grad_norm": 7.995015621185303,
      "learning_rate": 9.144144144144145e-06,
      "loss": 0.5002,
      "step": 8550
    },
    {
      "epoch": 25.71,
      "grad_norm": 10.026015281677246,
      "learning_rate": 9.143143143143144e-06,
      "loss": 0.4528,
      "step": 8560
    },
    {
      "epoch": 25.74,
      "grad_norm": 10.452445030212402,
      "learning_rate": 9.142142142142144e-06,
      "loss": 0.4851,
      "step": 8570
    },
    {
      "epoch": 25.77,
      "grad_norm": 10.0867280960083,
      "learning_rate": 9.141141141141142e-06,
      "loss": 0.4482,
      "step": 8580
    },
    {
      "epoch": 25.8,
      "grad_norm": 12.90353012084961,
      "learning_rate": 9.140140140140141e-06,
      "loss": 0.4755,
      "step": 8590
    },
    {
      "epoch": 25.83,
      "grad_norm": 7.924851417541504,
      "learning_rate": 9.13913913913914e-06,
      "loss": 0.465,
      "step": 8600
    },
    {
      "epoch": 25.86,
      "grad_norm": 11.196653366088867,
      "learning_rate": 9.138138138138138e-06,
      "loss": 0.4396,
      "step": 8610
    },
    {
      "epoch": 25.89,
      "grad_norm": 7.144512176513672,
      "learning_rate": 9.137137137137137e-06,
      "loss": 0.4915,
      "step": 8620
    },
    {
      "epoch": 25.92,
      "grad_norm": 8.691585540771484,
      "learning_rate": 9.136136136136137e-06,
      "loss": 0.4835,
      "step": 8630
    },
    {
      "epoch": 25.95,
      "grad_norm": 9.7068452835083,
      "learning_rate": 9.135135135135136e-06,
      "loss": 0.4593,
      "step": 8640
    },
    {
      "epoch": 25.98,
      "grad_norm": 11.20572566986084,
      "learning_rate": 9.134134134134134e-06,
      "loss": 0.4867,
      "step": 8650
    },
    {
      "epoch": 26.0,
      "eval_accuracy": 0.9295,
      "eval_loss": 0.24820935726165771,
      "eval_runtime": 30.6834,
      "eval_samples_per_second": 325.909,
      "eval_steps_per_second": 1.304,
      "step": 8658
    },
    {
      "epoch": 26.01,
      "grad_norm": 10.692183494567871,
      "learning_rate": 9.133133133133135e-06,
      "loss": 0.4562,
      "step": 8660
    },
    {
      "epoch": 26.04,
      "grad_norm": 11.422379493713379,
      "learning_rate": 9.132132132132133e-06,
      "loss": 0.4686,
      "step": 8670
    },
    {
      "epoch": 26.07,
      "grad_norm": 21.960498809814453,
      "learning_rate": 9.131131131131132e-06,
      "loss": 0.4379,
      "step": 8680
    },
    {
      "epoch": 26.1,
      "grad_norm": 6.990413188934326,
      "learning_rate": 9.13013013013013e-06,
      "loss": 0.4402,
      "step": 8690
    },
    {
      "epoch": 26.13,
      "grad_norm": 11.235393524169922,
      "learning_rate": 9.129129129129129e-06,
      "loss": 0.4269,
      "step": 8700
    },
    {
      "epoch": 26.16,
      "grad_norm": 11.997522354125977,
      "learning_rate": 9.12812812812813e-06,
      "loss": 0.4442,
      "step": 8710
    },
    {
      "epoch": 26.19,
      "grad_norm": 10.500715255737305,
      "learning_rate": 9.127127127127128e-06,
      "loss": 0.4521,
      "step": 8720
    },
    {
      "epoch": 26.22,
      "grad_norm": 9.616368293762207,
      "learning_rate": 9.126126126126126e-06,
      "loss": 0.4274,
      "step": 8730
    },
    {
      "epoch": 26.25,
      "grad_norm": 11.904566764831543,
      "learning_rate": 9.125125125125127e-06,
      "loss": 0.4542,
      "step": 8740
    },
    {
      "epoch": 26.28,
      "grad_norm": 11.470881462097168,
      "learning_rate": 9.124124124124125e-06,
      "loss": 0.4126,
      "step": 8750
    },
    {
      "epoch": 26.31,
      "grad_norm": 10.796327590942383,
      "learning_rate": 9.123123123123124e-06,
      "loss": 0.5106,
      "step": 8760
    },
    {
      "epoch": 26.34,
      "grad_norm": 9.003201484680176,
      "learning_rate": 9.122122122122123e-06,
      "loss": 0.4288,
      "step": 8770
    },
    {
      "epoch": 26.37,
      "grad_norm": 12.855923652648926,
      "learning_rate": 9.121121121121121e-06,
      "loss": 0.5086,
      "step": 8780
    },
    {
      "epoch": 26.4,
      "grad_norm": 10.519691467285156,
      "learning_rate": 9.120120120120121e-06,
      "loss": 0.4857,
      "step": 8790
    },
    {
      "epoch": 26.43,
      "grad_norm": 10.088730812072754,
      "learning_rate": 9.11911911911912e-06,
      "loss": 0.5249,
      "step": 8800
    },
    {
      "epoch": 26.46,
      "grad_norm": 7.654215335845947,
      "learning_rate": 9.118118118118119e-06,
      "loss": 0.4909,
      "step": 8810
    },
    {
      "epoch": 26.49,
      "grad_norm": 10.246333122253418,
      "learning_rate": 9.117117117117117e-06,
      "loss": 0.5,
      "step": 8820
    },
    {
      "epoch": 26.52,
      "grad_norm": 9.856045722961426,
      "learning_rate": 9.116116116116117e-06,
      "loss": 0.4561,
      "step": 8830
    },
    {
      "epoch": 26.55,
      "grad_norm": 8.209404945373535,
      "learning_rate": 9.115115115115116e-06,
      "loss": 0.4903,
      "step": 8840
    },
    {
      "epoch": 26.58,
      "grad_norm": 9.775983810424805,
      "learning_rate": 9.114114114114115e-06,
      "loss": 0.4763,
      "step": 8850
    },
    {
      "epoch": 26.61,
      "grad_norm": 16.25369644165039,
      "learning_rate": 9.113113113113113e-06,
      "loss": 0.4846,
      "step": 8860
    },
    {
      "epoch": 26.64,
      "grad_norm": 10.342110633850098,
      "learning_rate": 9.112112112112112e-06,
      "loss": 0.4912,
      "step": 8870
    },
    {
      "epoch": 26.67,
      "grad_norm": 12.154577255249023,
      "learning_rate": 9.111111111111112e-06,
      "loss": 0.4941,
      "step": 8880
    },
    {
      "epoch": 26.7,
      "grad_norm": 8.336454391479492,
      "learning_rate": 9.11011011011011e-06,
      "loss": 0.4475,
      "step": 8890
    },
    {
      "epoch": 26.73,
      "grad_norm": 7.27547025680542,
      "learning_rate": 9.10910910910911e-06,
      "loss": 0.4518,
      "step": 8900
    },
    {
      "epoch": 26.76,
      "grad_norm": 8.158557891845703,
      "learning_rate": 9.10810810810811e-06,
      "loss": 0.3841,
      "step": 8910
    },
    {
      "epoch": 26.79,
      "grad_norm": 13.1236572265625,
      "learning_rate": 9.107107107107108e-06,
      "loss": 0.4412,
      "step": 8920
    },
    {
      "epoch": 26.82,
      "grad_norm": 9.89743423461914,
      "learning_rate": 9.106106106106107e-06,
      "loss": 0.4979,
      "step": 8930
    },
    {
      "epoch": 26.85,
      "grad_norm": 8.974017143249512,
      "learning_rate": 9.105105105105105e-06,
      "loss": 0.4194,
      "step": 8940
    },
    {
      "epoch": 26.88,
      "grad_norm": 12.197188377380371,
      "learning_rate": 9.104104104104104e-06,
      "loss": 0.4394,
      "step": 8950
    },
    {
      "epoch": 26.91,
      "grad_norm": 17.870830535888672,
      "learning_rate": 9.103103103103104e-06,
      "loss": 0.4176,
      "step": 8960
    },
    {
      "epoch": 26.94,
      "grad_norm": 11.099676132202148,
      "learning_rate": 9.102102102102103e-06,
      "loss": 0.4828,
      "step": 8970
    },
    {
      "epoch": 26.97,
      "grad_norm": 14.031774520874023,
      "learning_rate": 9.101101101101101e-06,
      "loss": 0.4873,
      "step": 8980
    },
    {
      "epoch": 27.0,
      "grad_norm": 7.13455867767334,
      "learning_rate": 9.100100100100102e-06,
      "loss": 0.4414,
      "step": 8990
    },
    {
      "epoch": 27.0,
      "eval_accuracy": 0.9267,
      "eval_loss": 0.25395047664642334,
      "eval_runtime": 30.0761,
      "eval_samples_per_second": 332.49,
      "eval_steps_per_second": 1.33,
      "step": 8991
    },
    {
      "epoch": 27.03,
      "grad_norm": 8.033126831054688,
      "learning_rate": 9.0990990990991e-06,
      "loss": 0.3802,
      "step": 9000
    },
    {
      "epoch": 27.06,
      "grad_norm": 8.065628051757812,
      "learning_rate": 9.098098098098099e-06,
      "loss": 0.4256,
      "step": 9010
    },
    {
      "epoch": 27.09,
      "grad_norm": 14.366639137268066,
      "learning_rate": 9.097097097097098e-06,
      "loss": 0.4556,
      "step": 9020
    },
    {
      "epoch": 27.12,
      "grad_norm": 10.824115753173828,
      "learning_rate": 9.096096096096096e-06,
      "loss": 0.4359,
      "step": 9030
    },
    {
      "epoch": 27.15,
      "grad_norm": 9.604802131652832,
      "learning_rate": 9.095095095095095e-06,
      "loss": 0.4823,
      "step": 9040
    },
    {
      "epoch": 27.18,
      "grad_norm": 29.343629837036133,
      "learning_rate": 9.094094094094095e-06,
      "loss": 0.4764,
      "step": 9050
    },
    {
      "epoch": 27.21,
      "grad_norm": 7.678133487701416,
      "learning_rate": 9.093093093093094e-06,
      "loss": 0.4019,
      "step": 9060
    },
    {
      "epoch": 27.24,
      "grad_norm": 12.999184608459473,
      "learning_rate": 9.092092092092092e-06,
      "loss": 0.4373,
      "step": 9070
    },
    {
      "epoch": 27.27,
      "grad_norm": 27.29155921936035,
      "learning_rate": 9.091091091091093e-06,
      "loss": 0.4027,
      "step": 9080
    },
    {
      "epoch": 27.3,
      "grad_norm": 9.286530494689941,
      "learning_rate": 9.090090090090091e-06,
      "loss": 0.441,
      "step": 9090
    },
    {
      "epoch": 27.33,
      "grad_norm": 8.044090270996094,
      "learning_rate": 9.08908908908909e-06,
      "loss": 0.3782,
      "step": 9100
    },
    {
      "epoch": 27.36,
      "grad_norm": 9.737653732299805,
      "learning_rate": 9.088088088088088e-06,
      "loss": 0.4717,
      "step": 9110
    },
    {
      "epoch": 27.39,
      "grad_norm": 11.821213722229004,
      "learning_rate": 9.087087087087087e-06,
      "loss": 0.4395,
      "step": 9120
    },
    {
      "epoch": 27.42,
      "grad_norm": 11.546897888183594,
      "learning_rate": 9.086086086086087e-06,
      "loss": 0.4485,
      "step": 9130
    },
    {
      "epoch": 27.45,
      "grad_norm": 11.010965347290039,
      "learning_rate": 9.085085085085086e-06,
      "loss": 0.459,
      "step": 9140
    },
    {
      "epoch": 27.48,
      "grad_norm": 9.444851875305176,
      "learning_rate": 9.084084084084084e-06,
      "loss": 0.4984,
      "step": 9150
    },
    {
      "epoch": 27.51,
      "grad_norm": 19.694604873657227,
      "learning_rate": 9.083083083083085e-06,
      "loss": 0.4741,
      "step": 9160
    },
    {
      "epoch": 27.54,
      "grad_norm": 7.38561487197876,
      "learning_rate": 9.082082082082083e-06,
      "loss": 0.435,
      "step": 9170
    },
    {
      "epoch": 27.57,
      "grad_norm": 11.60218620300293,
      "learning_rate": 9.081081081081082e-06,
      "loss": 0.4238,
      "step": 9180
    },
    {
      "epoch": 27.6,
      "grad_norm": 13.115228652954102,
      "learning_rate": 9.08008008008008e-06,
      "loss": 0.4175,
      "step": 9190
    },
    {
      "epoch": 27.63,
      "grad_norm": 22.14258575439453,
      "learning_rate": 9.079079079079079e-06,
      "loss": 0.4655,
      "step": 9200
    },
    {
      "epoch": 27.66,
      "grad_norm": 9.186527252197266,
      "learning_rate": 9.07807807807808e-06,
      "loss": 0.4148,
      "step": 9210
    },
    {
      "epoch": 27.69,
      "grad_norm": 8.470562934875488,
      "learning_rate": 9.077077077077078e-06,
      "loss": 0.4463,
      "step": 9220
    },
    {
      "epoch": 27.72,
      "grad_norm": 8.844724655151367,
      "learning_rate": 9.076076076076077e-06,
      "loss": 0.5129,
      "step": 9230
    },
    {
      "epoch": 27.75,
      "grad_norm": 14.874906539916992,
      "learning_rate": 9.075075075075077e-06,
      "loss": 0.508,
      "step": 9240
    },
    {
      "epoch": 27.78,
      "grad_norm": 9.115788459777832,
      "learning_rate": 9.074074074074075e-06,
      "loss": 0.4789,
      "step": 9250
    },
    {
      "epoch": 27.81,
      "grad_norm": 11.254831314086914,
      "learning_rate": 9.073073073073074e-06,
      "loss": 0.4359,
      "step": 9260
    },
    {
      "epoch": 27.84,
      "grad_norm": 9.67915153503418,
      "learning_rate": 9.072072072072073e-06,
      "loss": 0.4465,
      "step": 9270
    },
    {
      "epoch": 27.87,
      "grad_norm": 7.342519283294678,
      "learning_rate": 9.071071071071071e-06,
      "loss": 0.4468,
      "step": 9280
    },
    {
      "epoch": 27.9,
      "grad_norm": 9.372382164001465,
      "learning_rate": 9.07007007007007e-06,
      "loss": 0.4373,
      "step": 9290
    },
    {
      "epoch": 27.93,
      "grad_norm": 28.929168701171875,
      "learning_rate": 9.06906906906907e-06,
      "loss": 0.4206,
      "step": 9300
    },
    {
      "epoch": 27.96,
      "grad_norm": 88.10252380371094,
      "learning_rate": 9.068068068068069e-06,
      "loss": 0.4712,
      "step": 9310
    },
    {
      "epoch": 27.99,
      "grad_norm": 20.861848831176758,
      "learning_rate": 9.067067067067067e-06,
      "loss": 0.4574,
      "step": 9320
    },
    {
      "epoch": 28.0,
      "eval_accuracy": 0.9287,
      "eval_loss": 0.24936652183532715,
      "eval_runtime": 30.6074,
      "eval_samples_per_second": 326.718,
      "eval_steps_per_second": 1.307,
      "step": 9324
    },
    {
      "epoch": 28.02,
      "grad_norm": 18.218908309936523,
      "learning_rate": 9.066066066066068e-06,
      "loss": 0.4536,
      "step": 9330
    },
    {
      "epoch": 28.05,
      "grad_norm": 22.617238998413086,
      "learning_rate": 9.065065065065066e-06,
      "loss": 0.4148,
      "step": 9340
    },
    {
      "epoch": 28.08,
      "grad_norm": 13.542783737182617,
      "learning_rate": 9.064064064064065e-06,
      "loss": 0.4575,
      "step": 9350
    },
    {
      "epoch": 28.11,
      "grad_norm": 9.672589302062988,
      "learning_rate": 9.063063063063063e-06,
      "loss": 0.4472,
      "step": 9360
    },
    {
      "epoch": 28.14,
      "grad_norm": 10.828847885131836,
      "learning_rate": 9.062062062062062e-06,
      "loss": 0.4097,
      "step": 9370
    },
    {
      "epoch": 28.17,
      "grad_norm": 8.991840362548828,
      "learning_rate": 9.061061061061062e-06,
      "loss": 0.4342,
      "step": 9380
    },
    {
      "epoch": 28.2,
      "grad_norm": 10.406083106994629,
      "learning_rate": 9.06006006006006e-06,
      "loss": 0.4401,
      "step": 9390
    },
    {
      "epoch": 28.23,
      "grad_norm": 9.752996444702148,
      "learning_rate": 9.05905905905906e-06,
      "loss": 0.4453,
      "step": 9400
    },
    {
      "epoch": 28.26,
      "grad_norm": 14.64594841003418,
      "learning_rate": 9.05805805805806e-06,
      "loss": 0.4074,
      "step": 9410
    },
    {
      "epoch": 28.29,
      "grad_norm": 6.0390191078186035,
      "learning_rate": 9.057057057057058e-06,
      "loss": 0.4417,
      "step": 9420
    },
    {
      "epoch": 28.32,
      "grad_norm": 10.605696678161621,
      "learning_rate": 9.056056056056057e-06,
      "loss": 0.4227,
      "step": 9430
    },
    {
      "epoch": 28.35,
      "grad_norm": 9.552401542663574,
      "learning_rate": 9.055055055055055e-06,
      "loss": 0.4654,
      "step": 9440
    },
    {
      "epoch": 28.38,
      "grad_norm": 10.555407524108887,
      "learning_rate": 9.054054054054054e-06,
      "loss": 0.4677,
      "step": 9450
    },
    {
      "epoch": 28.41,
      "grad_norm": 10.286060333251953,
      "learning_rate": 9.053053053053054e-06,
      "loss": 0.3975,
      "step": 9460
    },
    {
      "epoch": 28.44,
      "grad_norm": 13.12886905670166,
      "learning_rate": 9.052052052052053e-06,
      "loss": 0.4472,
      "step": 9470
    },
    {
      "epoch": 28.47,
      "grad_norm": 9.690975189208984,
      "learning_rate": 9.051051051051052e-06,
      "loss": 0.4633,
      "step": 9480
    },
    {
      "epoch": 28.5,
      "grad_norm": 6.2562456130981445,
      "learning_rate": 9.05005005005005e-06,
      "loss": 0.4694,
      "step": 9490
    },
    {
      "epoch": 28.53,
      "grad_norm": 9.984062194824219,
      "learning_rate": 9.04904904904905e-06,
      "loss": 0.4463,
      "step": 9500
    },
    {
      "epoch": 28.56,
      "grad_norm": 11.556092262268066,
      "learning_rate": 9.048048048048049e-06,
      "loss": 0.4532,
      "step": 9510
    },
    {
      "epoch": 28.59,
      "grad_norm": 11.245503425598145,
      "learning_rate": 9.047047047047048e-06,
      "loss": 0.4504,
      "step": 9520
    },
    {
      "epoch": 28.62,
      "grad_norm": 10.833431243896484,
      "learning_rate": 9.046046046046046e-06,
      "loss": 0.5309,
      "step": 9530
    },
    {
      "epoch": 28.65,
      "grad_norm": 15.275016784667969,
      "learning_rate": 9.045045045045045e-06,
      "loss": 0.3924,
      "step": 9540
    },
    {
      "epoch": 28.68,
      "grad_norm": 8.553839683532715,
      "learning_rate": 9.044044044044045e-06,
      "loss": 0.4527,
      "step": 9550
    },
    {
      "epoch": 28.71,
      "grad_norm": 11.880951881408691,
      "learning_rate": 9.043043043043044e-06,
      "loss": 0.5098,
      "step": 9560
    },
    {
      "epoch": 28.74,
      "grad_norm": 12.77589225769043,
      "learning_rate": 9.042042042042042e-06,
      "loss": 0.4748,
      "step": 9570
    },
    {
      "epoch": 28.77,
      "grad_norm": 8.945269584655762,
      "learning_rate": 9.041041041041043e-06,
      "loss": 0.4274,
      "step": 9580
    },
    {
      "epoch": 28.8,
      "grad_norm": 9.76388931274414,
      "learning_rate": 9.040040040040041e-06,
      "loss": 0.5033,
      "step": 9590
    },
    {
      "epoch": 28.83,
      "grad_norm": 10.47138786315918,
      "learning_rate": 9.03903903903904e-06,
      "loss": 0.4589,
      "step": 9600
    },
    {
      "epoch": 28.86,
      "grad_norm": 11.299651145935059,
      "learning_rate": 9.038038038038038e-06,
      "loss": 0.4671,
      "step": 9610
    },
    {
      "epoch": 28.89,
      "grad_norm": 20.282169342041016,
      "learning_rate": 9.037037037037037e-06,
      "loss": 0.4182,
      "step": 9620
    },
    {
      "epoch": 28.92,
      "grad_norm": 13.877592086791992,
      "learning_rate": 9.036036036036037e-06,
      "loss": 0.4441,
      "step": 9630
    },
    {
      "epoch": 28.95,
      "grad_norm": 33.09040069580078,
      "learning_rate": 9.035035035035036e-06,
      "loss": 0.493,
      "step": 9640
    },
    {
      "epoch": 28.98,
      "grad_norm": 12.959691047668457,
      "learning_rate": 9.034034034034034e-06,
      "loss": 0.4109,
      "step": 9650
    },
    {
      "epoch": 29.0,
      "eval_accuracy": 0.928,
      "eval_loss": 0.2532622218132019,
      "eval_runtime": 30.4138,
      "eval_samples_per_second": 328.798,
      "eval_steps_per_second": 1.315,
      "step": 9657
    },
    {
      "epoch": 29.01,
      "grad_norm": 7.716038703918457,
      "learning_rate": 9.033033033033035e-06,
      "loss": 0.4287,
      "step": 9660
    },
    {
      "epoch": 29.04,
      "grad_norm": 14.750630378723145,
      "learning_rate": 9.032032032032033e-06,
      "loss": 0.4264,
      "step": 9670
    },
    {
      "epoch": 29.07,
      "grad_norm": 24.9686222076416,
      "learning_rate": 9.031031031031032e-06,
      "loss": 0.4061,
      "step": 9680
    },
    {
      "epoch": 29.1,
      "grad_norm": 11.106522560119629,
      "learning_rate": 9.03003003003003e-06,
      "loss": 0.4478,
      "step": 9690
    },
    {
      "epoch": 29.13,
      "grad_norm": 12.746867179870605,
      "learning_rate": 9.029029029029029e-06,
      "loss": 0.4483,
      "step": 9700
    },
    {
      "epoch": 29.16,
      "grad_norm": 10.57923698425293,
      "learning_rate": 9.02802802802803e-06,
      "loss": 0.4322,
      "step": 9710
    },
    {
      "epoch": 29.19,
      "grad_norm": 8.244423866271973,
      "learning_rate": 9.027027027027028e-06,
      "loss": 0.4101,
      "step": 9720
    },
    {
      "epoch": 29.22,
      "grad_norm": 13.49233341217041,
      "learning_rate": 9.026026026026027e-06,
      "loss": 0.4631,
      "step": 9730
    },
    {
      "epoch": 29.25,
      "grad_norm": 14.60818099975586,
      "learning_rate": 9.025025025025025e-06,
      "loss": 0.4404,
      "step": 9740
    },
    {
      "epoch": 29.28,
      "grad_norm": 12.197328567504883,
      "learning_rate": 9.024024024024025e-06,
      "loss": 0.4371,
      "step": 9750
    },
    {
      "epoch": 29.31,
      "grad_norm": 10.69261646270752,
      "learning_rate": 9.023023023023024e-06,
      "loss": 0.4042,
      "step": 9760
    },
    {
      "epoch": 29.34,
      "grad_norm": 8.400017738342285,
      "learning_rate": 9.022022022022023e-06,
      "loss": 0.4588,
      "step": 9770
    },
    {
      "epoch": 29.37,
      "grad_norm": 6.550203800201416,
      "learning_rate": 9.021021021021021e-06,
      "loss": 0.3832,
      "step": 9780
    },
    {
      "epoch": 29.4,
      "grad_norm": 10.774068832397461,
      "learning_rate": 9.02002002002002e-06,
      "loss": 0.4582,
      "step": 9790
    },
    {
      "epoch": 29.43,
      "grad_norm": 14.41346263885498,
      "learning_rate": 9.01901901901902e-06,
      "loss": 0.429,
      "step": 9800
    },
    {
      "epoch": 29.46,
      "grad_norm": 11.75338077545166,
      "learning_rate": 9.018018018018019e-06,
      "loss": 0.4302,
      "step": 9810
    },
    {
      "epoch": 29.49,
      "grad_norm": 9.722987174987793,
      "learning_rate": 9.017017017017017e-06,
      "loss": 0.4308,
      "step": 9820
    },
    {
      "epoch": 29.52,
      "grad_norm": 10.287395477294922,
      "learning_rate": 9.016016016016018e-06,
      "loss": 0.4354,
      "step": 9830
    },
    {
      "epoch": 29.55,
      "grad_norm": 24.349010467529297,
      "learning_rate": 9.015015015015016e-06,
      "loss": 0.442,
      "step": 9840
    },
    {
      "epoch": 29.58,
      "grad_norm": 10.922633171081543,
      "learning_rate": 9.014014014014015e-06,
      "loss": 0.4221,
      "step": 9850
    },
    {
      "epoch": 29.61,
      "grad_norm": 11.137175559997559,
      "learning_rate": 9.013013013013013e-06,
      "loss": 0.4693,
      "step": 9860
    },
    {
      "epoch": 29.64,
      "grad_norm": 13.711250305175781,
      "learning_rate": 9.012012012012012e-06,
      "loss": 0.4086,
      "step": 9870
    },
    {
      "epoch": 29.67,
      "grad_norm": 19.143190383911133,
      "learning_rate": 9.011011011011012e-06,
      "loss": 0.4192,
      "step": 9880
    },
    {
      "epoch": 29.7,
      "grad_norm": 12.561365127563477,
      "learning_rate": 9.010010010010011e-06,
      "loss": 0.471,
      "step": 9890
    },
    {
      "epoch": 29.73,
      "grad_norm": 12.069458961486816,
      "learning_rate": 9.00900900900901e-06,
      "loss": 0.4193,
      "step": 9900
    },
    {
      "epoch": 29.76,
      "grad_norm": 11.876630783081055,
      "learning_rate": 9.00800800800801e-06,
      "loss": 0.4067,
      "step": 9910
    },
    {
      "epoch": 29.79,
      "grad_norm": 11.242775917053223,
      "learning_rate": 9.007007007007008e-06,
      "loss": 0.4756,
      "step": 9920
    },
    {
      "epoch": 29.82,
      "grad_norm": 11.685412406921387,
      "learning_rate": 9.006006006006007e-06,
      "loss": 0.4608,
      "step": 9930
    },
    {
      "epoch": 29.85,
      "grad_norm": 8.70016098022461,
      "learning_rate": 9.005005005005006e-06,
      "loss": 0.5007,
      "step": 9940
    },
    {
      "epoch": 29.88,
      "grad_norm": 11.36416244506836,
      "learning_rate": 9.004004004004004e-06,
      "loss": 0.4296,
      "step": 9950
    },
    {
      "epoch": 29.91,
      "grad_norm": 14.544692039489746,
      "learning_rate": 9.003003003003003e-06,
      "loss": 0.4573,
      "step": 9960
    },
    {
      "epoch": 29.94,
      "grad_norm": 5.4239044189453125,
      "learning_rate": 9.002002002002003e-06,
      "loss": 0.4643,
      "step": 9970
    },
    {
      "epoch": 29.97,
      "grad_norm": 6.896058082580566,
      "learning_rate": 9.001001001001002e-06,
      "loss": 0.4539,
      "step": 9980
    },
    {
      "epoch": 30.0,
      "grad_norm": 46.193382263183594,
      "learning_rate": 9e-06,
      "loss": 0.4433,
      "step": 9990
    },
    {
      "epoch": 30.0,
      "eval_accuracy": 0.9258,
      "eval_loss": 0.257062703371048,
      "eval_runtime": 30.4684,
      "eval_samples_per_second": 328.209,
      "eval_steps_per_second": 1.313,
      "step": 9990
    },
    {
      "epoch": 30.03,
      "grad_norm": 11.51973819732666,
      "learning_rate": 8.998998998999e-06,
      "loss": 0.4466,
      "step": 10000
    },
    {
      "epoch": 30.06,
      "grad_norm": 9.196100234985352,
      "learning_rate": 8.997997997997999e-06,
      "loss": 0.4622,
      "step": 10010
    },
    {
      "epoch": 30.09,
      "grad_norm": 6.755195140838623,
      "learning_rate": 8.996996996996998e-06,
      "loss": 0.4014,
      "step": 10020
    },
    {
      "epoch": 30.12,
      "grad_norm": 10.63003158569336,
      "learning_rate": 8.995995995995996e-06,
      "loss": 0.4284,
      "step": 10030
    },
    {
      "epoch": 30.15,
      "grad_norm": 11.699910163879395,
      "learning_rate": 8.994994994994995e-06,
      "loss": 0.4425,
      "step": 10040
    },
    {
      "epoch": 30.18,
      "grad_norm": 10.2941255569458,
      "learning_rate": 8.993993993993995e-06,
      "loss": 0.3961,
      "step": 10050
    },
    {
      "epoch": 30.21,
      "grad_norm": 7.498856067657471,
      "learning_rate": 8.992992992992994e-06,
      "loss": 0.3904,
      "step": 10060
    },
    {
      "epoch": 30.24,
      "grad_norm": 10.58414077758789,
      "learning_rate": 8.991991991991992e-06,
      "loss": 0.401,
      "step": 10070
    },
    {
      "epoch": 30.27,
      "grad_norm": 9.456209182739258,
      "learning_rate": 8.990990990990993e-06,
      "loss": 0.4734,
      "step": 10080
    },
    {
      "epoch": 30.3,
      "grad_norm": 8.097994804382324,
      "learning_rate": 8.989989989989991e-06,
      "loss": 0.381,
      "step": 10090
    },
    {
      "epoch": 30.33,
      "grad_norm": 9.386022567749023,
      "learning_rate": 8.98898898898899e-06,
      "loss": 0.4055,
      "step": 10100
    },
    {
      "epoch": 30.36,
      "grad_norm": 10.85639476776123,
      "learning_rate": 8.987987987987988e-06,
      "loss": 0.3886,
      "step": 10110
    },
    {
      "epoch": 30.39,
      "grad_norm": 25.46282196044922,
      "learning_rate": 8.986986986986987e-06,
      "loss": 0.4192,
      "step": 10120
    },
    {
      "epoch": 30.42,
      "grad_norm": 17.969945907592773,
      "learning_rate": 8.985985985985987e-06,
      "loss": 0.4346,
      "step": 10130
    },
    {
      "epoch": 30.45,
      "grad_norm": 13.142504692077637,
      "learning_rate": 8.984984984984986e-06,
      "loss": 0.4449,
      "step": 10140
    },
    {
      "epoch": 30.48,
      "grad_norm": 10.24559211730957,
      "learning_rate": 8.983983983983985e-06,
      "loss": 0.4229,
      "step": 10150
    },
    {
      "epoch": 30.51,
      "grad_norm": 13.709941864013672,
      "learning_rate": 8.982982982982985e-06,
      "loss": 0.4483,
      "step": 10160
    },
    {
      "epoch": 30.54,
      "grad_norm": 8.802275657653809,
      "learning_rate": 8.981981981981983e-06,
      "loss": 0.4159,
      "step": 10170
    },
    {
      "epoch": 30.57,
      "grad_norm": 13.62450122833252,
      "learning_rate": 8.980980980980982e-06,
      "loss": 0.3761,
      "step": 10180
    },
    {
      "epoch": 30.6,
      "grad_norm": 8.44536018371582,
      "learning_rate": 8.97997997997998e-06,
      "loss": 0.4117,
      "step": 10190
    },
    {
      "epoch": 30.63,
      "grad_norm": 10.537995338439941,
      "learning_rate": 8.97897897897898e-06,
      "loss": 0.4116,
      "step": 10200
    },
    {
      "epoch": 30.66,
      "grad_norm": 8.89731502532959,
      "learning_rate": 8.977977977977978e-06,
      "loss": 0.3843,
      "step": 10210
    },
    {
      "epoch": 30.69,
      "grad_norm": 9.666942596435547,
      "learning_rate": 8.976976976976978e-06,
      "loss": 0.3763,
      "step": 10220
    },
    {
      "epoch": 30.72,
      "grad_norm": 7.942086696624756,
      "learning_rate": 8.975975975975977e-06,
      "loss": 0.4354,
      "step": 10230
    },
    {
      "epoch": 30.75,
      "grad_norm": 11.044801712036133,
      "learning_rate": 8.974974974974975e-06,
      "loss": 0.4023,
      "step": 10240
    },
    {
      "epoch": 30.78,
      "grad_norm": 20.663305282592773,
      "learning_rate": 8.973973973973976e-06,
      "loss": 0.402,
      "step": 10250
    },
    {
      "epoch": 30.81,
      "grad_norm": 8.409327507019043,
      "learning_rate": 8.972972972972974e-06,
      "loss": 0.4478,
      "step": 10260
    },
    {
      "epoch": 30.84,
      "grad_norm": 17.233417510986328,
      "learning_rate": 8.971971971971973e-06,
      "loss": 0.4025,
      "step": 10270
    },
    {
      "epoch": 30.87,
      "grad_norm": 11.380728721618652,
      "learning_rate": 8.970970970970971e-06,
      "loss": 0.3812,
      "step": 10280
    },
    {
      "epoch": 30.9,
      "grad_norm": 10.061477661132812,
      "learning_rate": 8.96996996996997e-06,
      "loss": 0.4453,
      "step": 10290
    },
    {
      "epoch": 30.93,
      "grad_norm": 32.968605041503906,
      "learning_rate": 8.96896896896897e-06,
      "loss": 0.4616,
      "step": 10300
    },
    {
      "epoch": 30.96,
      "grad_norm": 10.942723274230957,
      "learning_rate": 8.967967967967969e-06,
      "loss": 0.4368,
      "step": 10310
    },
    {
      "epoch": 30.99,
      "grad_norm": 9.55047607421875,
      "learning_rate": 8.966966966966967e-06,
      "loss": 0.4034,
      "step": 10320
    },
    {
      "epoch": 31.0,
      "eval_accuracy": 0.9265,
      "eval_loss": 0.25430214405059814,
      "eval_runtime": 30.1984,
      "eval_samples_per_second": 331.144,
      "eval_steps_per_second": 1.325,
      "step": 10323
    },
    {
      "epoch": 31.02,
      "grad_norm": 10.171669006347656,
      "learning_rate": 8.965965965965968e-06,
      "loss": 0.406,
      "step": 10330
    },
    {
      "epoch": 31.05,
      "grad_norm": 18.30038833618164,
      "learning_rate": 8.964964964964966e-06,
      "loss": 0.506,
      "step": 10340
    },
    {
      "epoch": 31.08,
      "grad_norm": 10.151785850524902,
      "learning_rate": 8.963963963963965e-06,
      "loss": 0.4268,
      "step": 10350
    },
    {
      "epoch": 31.11,
      "grad_norm": 11.276358604431152,
      "learning_rate": 8.962962962962963e-06,
      "loss": 0.392,
      "step": 10360
    },
    {
      "epoch": 31.14,
      "grad_norm": 9.019916534423828,
      "learning_rate": 8.961961961961962e-06,
      "loss": 0.3692,
      "step": 10370
    },
    {
      "epoch": 31.17,
      "grad_norm": 13.906360626220703,
      "learning_rate": 8.960960960960962e-06,
      "loss": 0.3863,
      "step": 10380
    },
    {
      "epoch": 31.2,
      "grad_norm": 7.59787130355835,
      "learning_rate": 8.959959959959961e-06,
      "loss": 0.441,
      "step": 10390
    },
    {
      "epoch": 31.23,
      "grad_norm": 7.223018169403076,
      "learning_rate": 8.95895895895896e-06,
      "loss": 0.4255,
      "step": 10400
    },
    {
      "epoch": 31.26,
      "grad_norm": 8.906543731689453,
      "learning_rate": 8.957957957957958e-06,
      "loss": 0.4221,
      "step": 10410
    },
    {
      "epoch": 31.29,
      "grad_norm": 18.36624526977539,
      "learning_rate": 8.956956956956958e-06,
      "loss": 0.397,
      "step": 10420
    },
    {
      "epoch": 31.32,
      "grad_norm": 37.474876403808594,
      "learning_rate": 8.955955955955957e-06,
      "loss": 0.4041,
      "step": 10430
    },
    {
      "epoch": 31.35,
      "grad_norm": 12.822399139404297,
      "learning_rate": 8.954954954954956e-06,
      "loss": 0.4391,
      "step": 10440
    },
    {
      "epoch": 31.38,
      "grad_norm": 11.556109428405762,
      "learning_rate": 8.953953953953954e-06,
      "loss": 0.4278,
      "step": 10450
    },
    {
      "epoch": 31.41,
      "grad_norm": 9.80760669708252,
      "learning_rate": 8.952952952952953e-06,
      "loss": 0.3978,
      "step": 10460
    },
    {
      "epoch": 31.44,
      "grad_norm": 14.095552444458008,
      "learning_rate": 8.951951951951953e-06,
      "loss": 0.39,
      "step": 10470
    },
    {
      "epoch": 31.47,
      "grad_norm": 13.767873764038086,
      "learning_rate": 8.950950950950952e-06,
      "loss": 0.4704,
      "step": 10480
    },
    {
      "epoch": 31.5,
      "grad_norm": 9.213407516479492,
      "learning_rate": 8.94994994994995e-06,
      "loss": 0.4067,
      "step": 10490
    },
    {
      "epoch": 31.53,
      "grad_norm": 8.17852783203125,
      "learning_rate": 8.94894894894895e-06,
      "loss": 0.4305,
      "step": 10500
    },
    {
      "epoch": 31.56,
      "grad_norm": 11.813791275024414,
      "learning_rate": 8.94794794794795e-06,
      "loss": 0.4228,
      "step": 10510
    },
    {
      "epoch": 31.59,
      "grad_norm": 12.407458305358887,
      "learning_rate": 8.946946946946948e-06,
      "loss": 0.4218,
      "step": 10520
    },
    {
      "epoch": 31.62,
      "grad_norm": 11.357696533203125,
      "learning_rate": 8.945945945945946e-06,
      "loss": 0.4716,
      "step": 10530
    },
    {
      "epoch": 31.65,
      "grad_norm": 14.272128105163574,
      "learning_rate": 8.944944944944945e-06,
      "loss": 0.4015,
      "step": 10540
    },
    {
      "epoch": 31.68,
      "grad_norm": 9.179841995239258,
      "learning_rate": 8.943943943943945e-06,
      "loss": 0.4237,
      "step": 10550
    },
    {
      "epoch": 31.71,
      "grad_norm": 11.722990036010742,
      "learning_rate": 8.942942942942944e-06,
      "loss": 0.4525,
      "step": 10560
    },
    {
      "epoch": 31.74,
      "grad_norm": 9.312932968139648,
      "learning_rate": 8.941941941941942e-06,
      "loss": 0.4601,
      "step": 10570
    },
    {
      "epoch": 31.77,
      "grad_norm": 11.05932903289795,
      "learning_rate": 8.940940940940943e-06,
      "loss": 0.4365,
      "step": 10580
    },
    {
      "epoch": 31.8,
      "grad_norm": 12.362451553344727,
      "learning_rate": 8.939939939939941e-06,
      "loss": 0.4413,
      "step": 10590
    },
    {
      "epoch": 31.83,
      "grad_norm": 9.756061553955078,
      "learning_rate": 8.93893893893894e-06,
      "loss": 0.4036,
      "step": 10600
    },
    {
      "epoch": 31.86,
      "grad_norm": 12.465987205505371,
      "learning_rate": 8.937937937937939e-06,
      "loss": 0.4504,
      "step": 10610
    },
    {
      "epoch": 31.89,
      "grad_norm": 10.340137481689453,
      "learning_rate": 8.936936936936937e-06,
      "loss": 0.4414,
      "step": 10620
    },
    {
      "epoch": 31.92,
      "grad_norm": 7.671935558319092,
      "learning_rate": 8.935935935935937e-06,
      "loss": 0.4278,
      "step": 10630
    },
    {
      "epoch": 31.95,
      "grad_norm": 7.171747207641602,
      "learning_rate": 8.934934934934936e-06,
      "loss": 0.4192,
      "step": 10640
    },
    {
      "epoch": 31.98,
      "grad_norm": 14.753591537475586,
      "learning_rate": 8.933933933933935e-06,
      "loss": 0.4203,
      "step": 10650
    },
    {
      "epoch": 32.0,
      "eval_accuracy": 0.9286,
      "eval_loss": 0.2586733400821686,
      "eval_runtime": 30.4616,
      "eval_samples_per_second": 328.282,
      "eval_steps_per_second": 1.313,
      "step": 10656
    },
    {
      "epoch": 32.01,
      "grad_norm": 9.285674095153809,
      "learning_rate": 8.932932932932933e-06,
      "loss": 0.3513,
      "step": 10660
    },
    {
      "epoch": 32.04,
      "grad_norm": 13.126435279846191,
      "learning_rate": 8.931931931931933e-06,
      "loss": 0.4091,
      "step": 10670
    },
    {
      "epoch": 32.07,
      "grad_norm": 14.447212219238281,
      "learning_rate": 8.93093093093093e-06,
      "loss": 0.4669,
      "step": 10680
    },
    {
      "epoch": 32.1,
      "grad_norm": 8.851954460144043,
      "learning_rate": 8.92992992992993e-06,
      "loss": 0.4089,
      "step": 10690
    },
    {
      "epoch": 32.13,
      "grad_norm": 9.75550651550293,
      "learning_rate": 8.92892892892893e-06,
      "loss": 0.4309,
      "step": 10700
    },
    {
      "epoch": 32.16,
      "grad_norm": 8.048290252685547,
      "learning_rate": 8.927927927927928e-06,
      "loss": 0.4512,
      "step": 10710
    },
    {
      "epoch": 32.19,
      "grad_norm": 7.996254920959473,
      "learning_rate": 8.926926926926928e-06,
      "loss": 0.4154,
      "step": 10720
    },
    {
      "epoch": 32.22,
      "grad_norm": 11.290484428405762,
      "learning_rate": 8.925925925925927e-06,
      "loss": 0.3833,
      "step": 10730
    },
    {
      "epoch": 32.25,
      "grad_norm": 12.008049964904785,
      "learning_rate": 8.924924924924925e-06,
      "loss": 0.4744,
      "step": 10740
    },
    {
      "epoch": 32.28,
      "grad_norm": 9.068510055541992,
      "learning_rate": 8.923923923923926e-06,
      "loss": 0.4121,
      "step": 10750
    },
    {
      "epoch": 32.31,
      "grad_norm": 18.756370544433594,
      "learning_rate": 8.922922922922924e-06,
      "loss": 0.3766,
      "step": 10760
    },
    {
      "epoch": 32.34,
      "grad_norm": 12.292619705200195,
      "learning_rate": 8.921921921921923e-06,
      "loss": 0.3634,
      "step": 10770
    },
    {
      "epoch": 32.37,
      "grad_norm": 8.977395057678223,
      "learning_rate": 8.920920920920921e-06,
      "loss": 0.4379,
      "step": 10780
    },
    {
      "epoch": 32.4,
      "grad_norm": 13.57693862915039,
      "learning_rate": 8.91991991991992e-06,
      "loss": 0.4078,
      "step": 10790
    },
    {
      "epoch": 32.43,
      "grad_norm": 11.201655387878418,
      "learning_rate": 8.91891891891892e-06,
      "loss": 0.4462,
      "step": 10800
    },
    {
      "epoch": 32.46,
      "grad_norm": 14.462066650390625,
      "learning_rate": 8.917917917917919e-06,
      "loss": 0.4056,
      "step": 10810
    },
    {
      "epoch": 32.49,
      "grad_norm": 10.209726333618164,
      "learning_rate": 8.916916916916917e-06,
      "loss": 0.3494,
      "step": 10820
    },
    {
      "epoch": 32.52,
      "grad_norm": 12.745504379272461,
      "learning_rate": 8.915915915915918e-06,
      "loss": 0.4028,
      "step": 10830
    },
    {
      "epoch": 32.55,
      "grad_norm": 7.859507083892822,
      "learning_rate": 8.914914914914916e-06,
      "loss": 0.4311,
      "step": 10840
    },
    {
      "epoch": 32.58,
      "grad_norm": 7.138404369354248,
      "learning_rate": 8.913913913913915e-06,
      "loss": 0.4027,
      "step": 10850
    },
    {
      "epoch": 32.61,
      "grad_norm": 6.884012222290039,
      "learning_rate": 8.912912912912914e-06,
      "loss": 0.4139,
      "step": 10860
    },
    {
      "epoch": 32.64,
      "grad_norm": 9.922371864318848,
      "learning_rate": 8.911911911911912e-06,
      "loss": 0.4052,
      "step": 10870
    },
    {
      "epoch": 32.67,
      "grad_norm": 9.349055290222168,
      "learning_rate": 8.91091091091091e-06,
      "loss": 0.3701,
      "step": 10880
    },
    {
      "epoch": 32.7,
      "grad_norm": 11.731735229492188,
      "learning_rate": 8.909909909909911e-06,
      "loss": 0.405,
      "step": 10890
    },
    {
      "epoch": 32.73,
      "grad_norm": 9.601366996765137,
      "learning_rate": 8.90890890890891e-06,
      "loss": 0.4134,
      "step": 10900
    },
    {
      "epoch": 32.76,
      "grad_norm": 9.94797134399414,
      "learning_rate": 8.907907907907908e-06,
      "loss": 0.4073,
      "step": 10910
    },
    {
      "epoch": 32.79,
      "grad_norm": 10.514374732971191,
      "learning_rate": 8.906906906906909e-06,
      "loss": 0.4244,
      "step": 10920
    },
    {
      "epoch": 32.82,
      "grad_norm": 8.28952407836914,
      "learning_rate": 8.905905905905905e-06,
      "loss": 0.396,
      "step": 10930
    },
    {
      "epoch": 32.85,
      "grad_norm": 8.094382286071777,
      "learning_rate": 8.904904904904906e-06,
      "loss": 0.3911,
      "step": 10940
    },
    {
      "epoch": 32.88,
      "grad_norm": 10.968464851379395,
      "learning_rate": 8.903903903903904e-06,
      "loss": 0.4026,
      "step": 10950
    },
    {
      "epoch": 32.91,
      "grad_norm": 9.243600845336914,
      "learning_rate": 8.902902902902903e-06,
      "loss": 0.4211,
      "step": 10960
    },
    {
      "epoch": 32.94,
      "grad_norm": 17.827457427978516,
      "learning_rate": 8.901901901901903e-06,
      "loss": 0.4502,
      "step": 10970
    },
    {
      "epoch": 32.97,
      "grad_norm": 8.173558235168457,
      "learning_rate": 8.900900900900902e-06,
      "loss": 0.3942,
      "step": 10980
    },
    {
      "epoch": 33.0,
      "eval_accuracy": 0.927,
      "eval_loss": 0.2555399537086487,
      "eval_runtime": 30.1335,
      "eval_samples_per_second": 331.857,
      "eval_steps_per_second": 1.327,
      "step": 10989
    },
    {
      "epoch": 33.0,
      "grad_norm": 10.555376052856445,
      "learning_rate": 8.8998998998999e-06,
      "loss": 0.3512,
      "step": 10990
    },
    {
      "epoch": 33.03,
      "grad_norm": 12.584593772888184,
      "learning_rate": 8.8988988988989e-06,
      "loss": 0.4775,
      "step": 11000
    },
    {
      "epoch": 33.06,
      "grad_norm": 19.95534896850586,
      "learning_rate": 8.8978978978979e-06,
      "loss": 0.4426,
      "step": 11010
    },
    {
      "epoch": 33.09,
      "grad_norm": 9.170835494995117,
      "learning_rate": 8.896896896896898e-06,
      "loss": 0.3763,
      "step": 11020
    },
    {
      "epoch": 33.12,
      "grad_norm": 9.26423168182373,
      "learning_rate": 8.895895895895896e-06,
      "loss": 0.4534,
      "step": 11030
    },
    {
      "epoch": 33.15,
      "grad_norm": 43.479000091552734,
      "learning_rate": 8.894894894894895e-06,
      "loss": 0.4026,
      "step": 11040
    },
    {
      "epoch": 33.18,
      "grad_norm": 18.981931686401367,
      "learning_rate": 8.893893893893895e-06,
      "loss": 0.3983,
      "step": 11050
    },
    {
      "epoch": 33.21,
      "grad_norm": 10.87066650390625,
      "learning_rate": 8.892892892892894e-06,
      "loss": 0.5128,
      "step": 11060
    },
    {
      "epoch": 33.24,
      "grad_norm": 15.586153984069824,
      "learning_rate": 8.891891891891893e-06,
      "loss": 0.4168,
      "step": 11070
    },
    {
      "epoch": 33.27,
      "grad_norm": 10.101288795471191,
      "learning_rate": 8.890890890890893e-06,
      "loss": 0.3545,
      "step": 11080
    },
    {
      "epoch": 33.3,
      "grad_norm": 13.756767272949219,
      "learning_rate": 8.889889889889891e-06,
      "loss": 0.4007,
      "step": 11090
    },
    {
      "epoch": 33.33,
      "grad_norm": 9.537035942077637,
      "learning_rate": 8.888888888888888e-06,
      "loss": 0.4091,
      "step": 11100
    },
    {
      "epoch": 33.36,
      "grad_norm": 14.72399616241455,
      "learning_rate": 8.887887887887889e-06,
      "loss": 0.4212,
      "step": 11110
    },
    {
      "epoch": 33.39,
      "grad_norm": 9.135162353515625,
      "learning_rate": 8.886886886886887e-06,
      "loss": 0.3875,
      "step": 11120
    },
    {
      "epoch": 33.42,
      "grad_norm": 68.5506820678711,
      "learning_rate": 8.885885885885886e-06,
      "loss": 0.4631,
      "step": 11130
    },
    {
      "epoch": 33.45,
      "grad_norm": 8.094832420349121,
      "learning_rate": 8.884884884884886e-06,
      "loss": 0.4356,
      "step": 11140
    },
    {
      "epoch": 33.48,
      "grad_norm": 39.31940460205078,
      "learning_rate": 8.883883883883885e-06,
      "loss": 0.3732,
      "step": 11150
    },
    {
      "epoch": 33.51,
      "grad_norm": 8.211509704589844,
      "learning_rate": 8.882882882882883e-06,
      "loss": 0.4348,
      "step": 11160
    },
    {
      "epoch": 33.54,
      "grad_norm": 14.594659805297852,
      "learning_rate": 8.881881881881884e-06,
      "loss": 0.3853,
      "step": 11170
    },
    {
      "epoch": 33.57,
      "grad_norm": 10.51009750366211,
      "learning_rate": 8.88088088088088e-06,
      "loss": 0.4139,
      "step": 11180
    },
    {
      "epoch": 33.6,
      "grad_norm": 8.209303855895996,
      "learning_rate": 8.87987987987988e-06,
      "loss": 0.3847,
      "step": 11190
    },
    {
      "epoch": 33.63,
      "grad_norm": 7.231290817260742,
      "learning_rate": 8.87887887887888e-06,
      "loss": 0.3514,
      "step": 11200
    },
    {
      "epoch": 33.66,
      "grad_norm": 7.874517917633057,
      "learning_rate": 8.877877877877878e-06,
      "loss": 0.4409,
      "step": 11210
    },
    {
      "epoch": 33.69,
      "grad_norm": 12.88748836517334,
      "learning_rate": 8.876876876876878e-06,
      "loss": 0.4702,
      "step": 11220
    },
    {
      "epoch": 33.72,
      "grad_norm": 7.494571208953857,
      "learning_rate": 8.875875875875877e-06,
      "loss": 0.3996,
      "step": 11230
    },
    {
      "epoch": 33.75,
      "grad_norm": 9.620100975036621,
      "learning_rate": 8.874874874874875e-06,
      "loss": 0.4272,
      "step": 11240
    },
    {
      "epoch": 33.78,
      "grad_norm": 15.190299034118652,
      "learning_rate": 8.873873873873876e-06,
      "loss": 0.4536,
      "step": 11250
    },
    {
      "epoch": 33.81,
      "grad_norm": 31.521875381469727,
      "learning_rate": 8.872872872872874e-06,
      "loss": 0.418,
      "step": 11260
    },
    {
      "epoch": 33.84,
      "grad_norm": 10.580955505371094,
      "learning_rate": 8.871871871871873e-06,
      "loss": 0.4782,
      "step": 11270
    },
    {
      "epoch": 33.87,
      "grad_norm": 10.014829635620117,
      "learning_rate": 8.870870870870871e-06,
      "loss": 0.3915,
      "step": 11280
    },
    {
      "epoch": 33.9,
      "grad_norm": 14.541464805603027,
      "learning_rate": 8.86986986986987e-06,
      "loss": 0.385,
      "step": 11290
    },
    {
      "epoch": 33.93,
      "grad_norm": 8.281698226928711,
      "learning_rate": 8.86886886886887e-06,
      "loss": 0.4109,
      "step": 11300
    },
    {
      "epoch": 33.96,
      "grad_norm": 15.637141227722168,
      "learning_rate": 8.867867867867869e-06,
      "loss": 0.4507,
      "step": 11310
    },
    {
      "epoch": 33.99,
      "grad_norm": 12.205958366394043,
      "learning_rate": 8.866866866866868e-06,
      "loss": 0.3991,
      "step": 11320
    },
    {
      "epoch": 34.0,
      "eval_accuracy": 0.9271,
      "eval_loss": 0.25638455152511597,
      "eval_runtime": 30.4793,
      "eval_samples_per_second": 328.091,
      "eval_steps_per_second": 1.312,
      "step": 11322
    },
    {
      "epoch": 34.02,
      "grad_norm": 19.83987045288086,
      "learning_rate": 8.865865865865866e-06,
      "loss": 0.4396,
      "step": 11330
    },
    {
      "epoch": 34.05,
      "grad_norm": 15.217132568359375,
      "learning_rate": 8.864864864864866e-06,
      "loss": 0.4039,
      "step": 11340
    },
    {
      "epoch": 34.08,
      "grad_norm": 11.251469612121582,
      "learning_rate": 8.863863863863863e-06,
      "loss": 0.4203,
      "step": 11350
    },
    {
      "epoch": 34.11,
      "grad_norm": 18.08286476135254,
      "learning_rate": 8.862862862862864e-06,
      "loss": 0.3677,
      "step": 11360
    },
    {
      "epoch": 34.14,
      "grad_norm": 10.37747859954834,
      "learning_rate": 8.861861861861862e-06,
      "loss": 0.3824,
      "step": 11370
    },
    {
      "epoch": 34.17,
      "grad_norm": 8.164061546325684,
      "learning_rate": 8.86086086086086e-06,
      "loss": 0.3733,
      "step": 11380
    },
    {
      "epoch": 34.2,
      "grad_norm": 10.746585845947266,
      "learning_rate": 8.859859859859861e-06,
      "loss": 0.4573,
      "step": 11390
    },
    {
      "epoch": 34.23,
      "grad_norm": 39.38847732543945,
      "learning_rate": 8.85885885885886e-06,
      "loss": 0.439,
      "step": 11400
    },
    {
      "epoch": 34.26,
      "grad_norm": 13.934308052062988,
      "learning_rate": 8.857857857857858e-06,
      "loss": 0.4049,
      "step": 11410
    },
    {
      "epoch": 34.29,
      "grad_norm": 16.80377960205078,
      "learning_rate": 8.856856856856859e-06,
      "loss": 0.4029,
      "step": 11420
    },
    {
      "epoch": 34.32,
      "grad_norm": 17.65918731689453,
      "learning_rate": 8.855855855855855e-06,
      "loss": 0.3925,
      "step": 11430
    },
    {
      "epoch": 34.35,
      "grad_norm": 8.79592227935791,
      "learning_rate": 8.854854854854856e-06,
      "loss": 0.3893,
      "step": 11440
    },
    {
      "epoch": 34.38,
      "grad_norm": 9.0423002243042,
      "learning_rate": 8.853853853853854e-06,
      "loss": 0.4194,
      "step": 11450
    },
    {
      "epoch": 34.41,
      "grad_norm": 9.881715774536133,
      "learning_rate": 8.852852852852853e-06,
      "loss": 0.4095,
      "step": 11460
    },
    {
      "epoch": 34.44,
      "grad_norm": 10.891952514648438,
      "learning_rate": 8.851851851851853e-06,
      "loss": 0.4182,
      "step": 11470
    },
    {
      "epoch": 34.47,
      "grad_norm": 5.583596706390381,
      "learning_rate": 8.850850850850852e-06,
      "loss": 0.4154,
      "step": 11480
    },
    {
      "epoch": 34.5,
      "grad_norm": 8.60441780090332,
      "learning_rate": 8.84984984984985e-06,
      "loss": 0.3467,
      "step": 11490
    },
    {
      "epoch": 34.53,
      "grad_norm": 9.848461151123047,
      "learning_rate": 8.84884884884885e-06,
      "loss": 0.363,
      "step": 11500
    },
    {
      "epoch": 34.56,
      "grad_norm": 22.63396453857422,
      "learning_rate": 8.84784784784785e-06,
      "loss": 0.4687,
      "step": 11510
    },
    {
      "epoch": 34.59,
      "grad_norm": 17.529346466064453,
      "learning_rate": 8.846846846846848e-06,
      "loss": 0.4313,
      "step": 11520
    },
    {
      "epoch": 34.62,
      "grad_norm": 14.798726081848145,
      "learning_rate": 8.845845845845847e-06,
      "loss": 0.4441,
      "step": 11530
    },
    {
      "epoch": 34.65,
      "grad_norm": 7.219150543212891,
      "learning_rate": 8.844844844844845e-06,
      "loss": 0.3626,
      "step": 11540
    },
    {
      "epoch": 34.68,
      "grad_norm": 7.736437797546387,
      "learning_rate": 8.843843843843844e-06,
      "loss": 0.3977,
      "step": 11550
    },
    {
      "epoch": 34.71,
      "grad_norm": 7.897708892822266,
      "learning_rate": 8.842842842842844e-06,
      "loss": 0.4214,
      "step": 11560
    },
    {
      "epoch": 34.74,
      "grad_norm": 10.178305625915527,
      "learning_rate": 8.841841841841843e-06,
      "loss": 0.4245,
      "step": 11570
    },
    {
      "epoch": 34.77,
      "grad_norm": 12.364108085632324,
      "learning_rate": 8.840840840840841e-06,
      "loss": 0.4341,
      "step": 11580
    },
    {
      "epoch": 34.8,
      "grad_norm": 11.002373695373535,
      "learning_rate": 8.839839839839841e-06,
      "loss": 0.4514,
      "step": 11590
    },
    {
      "epoch": 34.83,
      "grad_norm": 9.911009788513184,
      "learning_rate": 8.838838838838838e-06,
      "loss": 0.4079,
      "step": 11600
    },
    {
      "epoch": 34.86,
      "grad_norm": 9.020291328430176,
      "learning_rate": 8.837837837837839e-06,
      "loss": 0.3755,
      "step": 11610
    },
    {
      "epoch": 34.89,
      "grad_norm": 7.697541236877441,
      "learning_rate": 8.836836836836837e-06,
      "loss": 0.38,
      "step": 11620
    },
    {
      "epoch": 34.92,
      "grad_norm": 9.926669120788574,
      "learning_rate": 8.835835835835836e-06,
      "loss": 0.4466,
      "step": 11630
    },
    {
      "epoch": 34.95,
      "grad_norm": 8.414145469665527,
      "learning_rate": 8.834834834834836e-06,
      "loss": 0.4046,
      "step": 11640
    },
    {
      "epoch": 34.98,
      "grad_norm": 12.950366020202637,
      "learning_rate": 8.833833833833835e-06,
      "loss": 0.4252,
      "step": 11650
    },
    {
      "epoch": 35.0,
      "eval_accuracy": 0.925,
      "eval_loss": 0.2602948546409607,
      "eval_runtime": 30.4318,
      "eval_samples_per_second": 328.603,
      "eval_steps_per_second": 1.314,
      "step": 11655
    },
    {
      "epoch": 35.02,
      "grad_norm": 8.139337539672852,
      "learning_rate": 8.832832832832833e-06,
      "loss": 0.3753,
      "step": 11660
    },
    {
      "epoch": 35.05,
      "grad_norm": 19.609962463378906,
      "learning_rate": 8.831831831831834e-06,
      "loss": 0.3776,
      "step": 11670
    },
    {
      "epoch": 35.08,
      "grad_norm": 13.31221866607666,
      "learning_rate": 8.83083083083083e-06,
      "loss": 0.4545,
      "step": 11680
    },
    {
      "epoch": 35.11,
      "grad_norm": 6.971845626831055,
      "learning_rate": 8.82982982982983e-06,
      "loss": 0.4126,
      "step": 11690
    },
    {
      "epoch": 35.14,
      "grad_norm": 8.801642417907715,
      "learning_rate": 8.82882882882883e-06,
      "loss": 0.4099,
      "step": 11700
    },
    {
      "epoch": 35.17,
      "grad_norm": 11.066483497619629,
      "learning_rate": 8.827827827827828e-06,
      "loss": 0.434,
      "step": 11710
    },
    {
      "epoch": 35.2,
      "grad_norm": 11.330470085144043,
      "learning_rate": 8.826826826826828e-06,
      "loss": 0.4099,
      "step": 11720
    },
    {
      "epoch": 35.23,
      "grad_norm": 15.0160493850708,
      "learning_rate": 8.825825825825827e-06,
      "loss": 0.411,
      "step": 11730
    },
    {
      "epoch": 35.26,
      "grad_norm": 8.729792594909668,
      "learning_rate": 8.824824824824825e-06,
      "loss": 0.4198,
      "step": 11740
    },
    {
      "epoch": 35.29,
      "grad_norm": 16.530698776245117,
      "learning_rate": 8.823823823823826e-06,
      "loss": 0.4032,
      "step": 11750
    },
    {
      "epoch": 35.32,
      "grad_norm": 11.7534761428833,
      "learning_rate": 8.822822822822824e-06,
      "loss": 0.4211,
      "step": 11760
    },
    {
      "epoch": 35.35,
      "grad_norm": 13.237783432006836,
      "learning_rate": 8.821821821821823e-06,
      "loss": 0.4198,
      "step": 11770
    },
    {
      "epoch": 35.38,
      "grad_norm": 15.49626636505127,
      "learning_rate": 8.820820820820822e-06,
      "loss": 0.4464,
      "step": 11780
    },
    {
      "epoch": 35.41,
      "grad_norm": 18.350927352905273,
      "learning_rate": 8.81981981981982e-06,
      "loss": 0.4756,
      "step": 11790
    },
    {
      "epoch": 35.44,
      "grad_norm": 10.492507934570312,
      "learning_rate": 8.818818818818819e-06,
      "loss": 0.427,
      "step": 11800
    },
    {
      "epoch": 35.47,
      "grad_norm": 9.446090698242188,
      "learning_rate": 8.817817817817819e-06,
      "loss": 0.4392,
      "step": 11810
    },
    {
      "epoch": 35.5,
      "grad_norm": 8.739493370056152,
      "learning_rate": 8.816816816816818e-06,
      "loss": 0.4848,
      "step": 11820
    },
    {
      "epoch": 35.53,
      "grad_norm": 9.116613388061523,
      "learning_rate": 8.815815815815816e-06,
      "loss": 0.4331,
      "step": 11830
    },
    {
      "epoch": 35.56,
      "grad_norm": 14.439440727233887,
      "learning_rate": 8.814814814814817e-06,
      "loss": 0.4283,
      "step": 11840
    },
    {
      "epoch": 35.59,
      "grad_norm": 11.005097389221191,
      "learning_rate": 8.813813813813813e-06,
      "loss": 0.4383,
      "step": 11850
    },
    {
      "epoch": 35.62,
      "grad_norm": 8.86245346069336,
      "learning_rate": 8.812812812812814e-06,
      "loss": 0.3952,
      "step": 11860
    },
    {
      "epoch": 35.65,
      "grad_norm": 10.16314697265625,
      "learning_rate": 8.811811811811812e-06,
      "loss": 0.4468,
      "step": 11870
    },
    {
      "epoch": 35.68,
      "grad_norm": 15.824507713317871,
      "learning_rate": 8.810810810810811e-06,
      "loss": 0.411,
      "step": 11880
    },
    {
      "epoch": 35.71,
      "grad_norm": 13.30593490600586,
      "learning_rate": 8.809809809809811e-06,
      "loss": 0.4557,
      "step": 11890
    },
    {
      "epoch": 35.74,
      "grad_norm": 8.765216827392578,
      "learning_rate": 8.80880880880881e-06,
      "loss": 0.4089,
      "step": 11900
    },
    {
      "epoch": 35.77,
      "grad_norm": 8.380579948425293,
      "learning_rate": 8.807807807807808e-06,
      "loss": 0.406,
      "step": 11910
    },
    {
      "epoch": 35.8,
      "grad_norm": 17.145652770996094,
      "learning_rate": 8.806806806806809e-06,
      "loss": 0.4429,
      "step": 11920
    },
    {
      "epoch": 35.83,
      "grad_norm": 9.908981323242188,
      "learning_rate": 8.805805805805806e-06,
      "loss": 0.3746,
      "step": 11930
    },
    {
      "epoch": 35.86,
      "grad_norm": 29.383146286010742,
      "learning_rate": 8.804804804804806e-06,
      "loss": 0.4198,
      "step": 11940
    },
    {
      "epoch": 35.89,
      "grad_norm": 12.33173942565918,
      "learning_rate": 8.803803803803804e-06,
      "loss": 0.4129,
      "step": 11950
    },
    {
      "epoch": 35.92,
      "grad_norm": 15.631083488464355,
      "learning_rate": 8.802802802802803e-06,
      "loss": 0.4468,
      "step": 11960
    },
    {
      "epoch": 35.95,
      "grad_norm": 9.646747589111328,
      "learning_rate": 8.801801801801803e-06,
      "loss": 0.3624,
      "step": 11970
    },
    {
      "epoch": 35.98,
      "grad_norm": 11.596949577331543,
      "learning_rate": 8.800800800800802e-06,
      "loss": 0.4393,
      "step": 11980
    },
    {
      "epoch": 36.0,
      "eval_accuracy": 0.9288,
      "eval_loss": 0.25744786858558655,
      "eval_runtime": 30.2289,
      "eval_samples_per_second": 330.809,
      "eval_steps_per_second": 1.323,
      "step": 11988
    },
    {
      "epoch": 36.01,
      "grad_norm": 15.940702438354492,
      "learning_rate": 8.7997997997998e-06,
      "loss": 0.3525,
      "step": 11990
    },
    {
      "epoch": 36.04,
      "grad_norm": 18.273988723754883,
      "learning_rate": 8.798798798798799e-06,
      "loss": 0.3523,
      "step": 12000
    },
    {
      "epoch": 36.07,
      "grad_norm": 11.517987251281738,
      "learning_rate": 8.797797797797798e-06,
      "loss": 0.4488,
      "step": 12010
    },
    {
      "epoch": 36.1,
      "grad_norm": 9.518025398254395,
      "learning_rate": 8.796796796796796e-06,
      "loss": 0.3793,
      "step": 12020
    },
    {
      "epoch": 36.13,
      "grad_norm": 13.380279541015625,
      "learning_rate": 8.795795795795797e-06,
      "loss": 0.4446,
      "step": 12030
    },
    {
      "epoch": 36.16,
      "grad_norm": 8.703398704528809,
      "learning_rate": 8.794794794794795e-06,
      "loss": 0.4429,
      "step": 12040
    },
    {
      "epoch": 36.19,
      "grad_norm": 7.529371738433838,
      "learning_rate": 8.793793793793794e-06,
      "loss": 0.3442,
      "step": 12050
    },
    {
      "epoch": 36.22,
      "grad_norm": 6.543487548828125,
      "learning_rate": 8.792792792792794e-06,
      "loss": 0.416,
      "step": 12060
    },
    {
      "epoch": 36.25,
      "grad_norm": 10.498993873596191,
      "learning_rate": 8.791791791791793e-06,
      "loss": 0.3902,
      "step": 12070
    },
    {
      "epoch": 36.28,
      "grad_norm": 9.740803718566895,
      "learning_rate": 8.790790790790791e-06,
      "loss": 0.4194,
      "step": 12080
    },
    {
      "epoch": 36.31,
      "grad_norm": 9.331558227539062,
      "learning_rate": 8.789789789789792e-06,
      "loss": 0.3863,
      "step": 12090
    },
    {
      "epoch": 36.34,
      "grad_norm": 11.235783576965332,
      "learning_rate": 8.788788788788788e-06,
      "loss": 0.3495,
      "step": 12100
    },
    {
      "epoch": 36.37,
      "grad_norm": 10.817569732666016,
      "learning_rate": 8.787787787787789e-06,
      "loss": 0.3825,
      "step": 12110
    },
    {
      "epoch": 36.4,
      "grad_norm": 6.002957820892334,
      "learning_rate": 8.786786786786787e-06,
      "loss": 0.4623,
      "step": 12120
    },
    {
      "epoch": 36.43,
      "grad_norm": 12.816144943237305,
      "learning_rate": 8.785785785785786e-06,
      "loss": 0.4443,
      "step": 12130
    },
    {
      "epoch": 36.46,
      "grad_norm": 16.187400817871094,
      "learning_rate": 8.784784784784786e-06,
      "loss": 0.3743,
      "step": 12140
    },
    {
      "epoch": 36.49,
      "grad_norm": 10.328038215637207,
      "learning_rate": 8.783783783783785e-06,
      "loss": 0.3773,
      "step": 12150
    },
    {
      "epoch": 36.52,
      "grad_norm": 9.40977954864502,
      "learning_rate": 8.782782782782783e-06,
      "loss": 0.4189,
      "step": 12160
    },
    {
      "epoch": 36.55,
      "grad_norm": 14.900984764099121,
      "learning_rate": 8.781781781781784e-06,
      "loss": 0.4194,
      "step": 12170
    },
    {
      "epoch": 36.58,
      "grad_norm": 8.397542953491211,
      "learning_rate": 8.78078078078078e-06,
      "loss": 0.3905,
      "step": 12180
    },
    {
      "epoch": 36.61,
      "grad_norm": 13.33767318725586,
      "learning_rate": 8.779779779779781e-06,
      "loss": 0.4135,
      "step": 12190
    },
    {
      "epoch": 36.64,
      "grad_norm": 20.183990478515625,
      "learning_rate": 8.77877877877878e-06,
      "loss": 0.3927,
      "step": 12200
    },
    {
      "epoch": 36.67,
      "grad_norm": 9.835906982421875,
      "learning_rate": 8.777777777777778e-06,
      "loss": 0.4258,
      "step": 12210
    },
    {
      "epoch": 36.7,
      "grad_norm": 8.524311065673828,
      "learning_rate": 8.776776776776778e-06,
      "loss": 0.3815,
      "step": 12220
    },
    {
      "epoch": 36.73,
      "grad_norm": 7.198403835296631,
      "learning_rate": 8.775775775775777e-06,
      "loss": 0.3816,
      "step": 12230
    },
    {
      "epoch": 36.76,
      "grad_norm": 13.140053749084473,
      "learning_rate": 8.774774774774776e-06,
      "loss": 0.3481,
      "step": 12240
    },
    {
      "epoch": 36.79,
      "grad_norm": 13.378637313842773,
      "learning_rate": 8.773773773773774e-06,
      "loss": 0.4309,
      "step": 12250
    },
    {
      "epoch": 36.82,
      "grad_norm": 13.015869140625,
      "learning_rate": 8.772772772772773e-06,
      "loss": 0.4183,
      "step": 12260
    },
    {
      "epoch": 36.85,
      "grad_norm": 9.255596160888672,
      "learning_rate": 8.771771771771771e-06,
      "loss": 0.384,
      "step": 12270
    },
    {
      "epoch": 36.88,
      "grad_norm": 21.15357780456543,
      "learning_rate": 8.770770770770772e-06,
      "loss": 0.4241,
      "step": 12280
    },
    {
      "epoch": 36.91,
      "grad_norm": 41.609375,
      "learning_rate": 8.76976976976977e-06,
      "loss": 0.4053,
      "step": 12290
    },
    {
      "epoch": 36.94,
      "grad_norm": 10.68895435333252,
      "learning_rate": 8.768768768768769e-06,
      "loss": 0.3459,
      "step": 12300
    },
    {
      "epoch": 36.97,
      "grad_norm": 8.08588981628418,
      "learning_rate": 8.767767767767769e-06,
      "loss": 0.4098,
      "step": 12310
    },
    {
      "epoch": 37.0,
      "grad_norm": 10.934639930725098,
      "learning_rate": 8.766766766766768e-06,
      "loss": 0.3949,
      "step": 12320
    },
    {
      "epoch": 37.0,
      "eval_accuracy": 0.9279,
      "eval_loss": 0.26377877593040466,
      "eval_runtime": 30.2658,
      "eval_samples_per_second": 330.406,
      "eval_steps_per_second": 1.322,
      "step": 12321
    },
    {
      "epoch": 37.03,
      "grad_norm": 10.90078067779541,
      "learning_rate": 8.765765765765766e-06,
      "loss": 0.3728,
      "step": 12330
    },
    {
      "epoch": 37.06,
      "grad_norm": 11.749211311340332,
      "learning_rate": 8.764764764764767e-06,
      "loss": 0.4014,
      "step": 12340
    },
    {
      "epoch": 37.09,
      "grad_norm": 23.710084915161133,
      "learning_rate": 8.763763763763763e-06,
      "loss": 0.4004,
      "step": 12350
    },
    {
      "epoch": 37.12,
      "grad_norm": 11.619260787963867,
      "learning_rate": 8.762762762762764e-06,
      "loss": 0.4439,
      "step": 12360
    },
    {
      "epoch": 37.15,
      "grad_norm": 10.997756958007812,
      "learning_rate": 8.761761761761762e-06,
      "loss": 0.3705,
      "step": 12370
    },
    {
      "epoch": 37.18,
      "grad_norm": 8.640237808227539,
      "learning_rate": 8.760760760760761e-06,
      "loss": 0.4049,
      "step": 12380
    },
    {
      "epoch": 37.21,
      "grad_norm": 13.680813789367676,
      "learning_rate": 8.759759759759761e-06,
      "loss": 0.4087,
      "step": 12390
    },
    {
      "epoch": 37.24,
      "grad_norm": 14.672085762023926,
      "learning_rate": 8.75875875875876e-06,
      "loss": 0.4006,
      "step": 12400
    },
    {
      "epoch": 37.27,
      "grad_norm": 8.288801193237305,
      "learning_rate": 8.757757757757758e-06,
      "loss": 0.4078,
      "step": 12410
    },
    {
      "epoch": 37.3,
      "grad_norm": 5.074464797973633,
      "learning_rate": 8.756756756756759e-06,
      "loss": 0.3809,
      "step": 12420
    },
    {
      "epoch": 37.33,
      "grad_norm": 15.01401138305664,
      "learning_rate": 8.755755755755756e-06,
      "loss": 0.4182,
      "step": 12430
    },
    {
      "epoch": 37.36,
      "grad_norm": 8.366719245910645,
      "learning_rate": 8.754754754754756e-06,
      "loss": 0.4188,
      "step": 12440
    },
    {
      "epoch": 37.39,
      "grad_norm": 11.154147148132324,
      "learning_rate": 8.753753753753755e-06,
      "loss": 0.3938,
      "step": 12450
    },
    {
      "epoch": 37.42,
      "grad_norm": 11.058762550354004,
      "learning_rate": 8.752752752752753e-06,
      "loss": 0.3139,
      "step": 12460
    },
    {
      "epoch": 37.45,
      "grad_norm": 15.326699256896973,
      "learning_rate": 8.751751751751752e-06,
      "loss": 0.4496,
      "step": 12470
    },
    {
      "epoch": 37.48,
      "grad_norm": 9.941515922546387,
      "learning_rate": 8.750750750750752e-06,
      "loss": 0.4487,
      "step": 12480
    },
    {
      "epoch": 37.51,
      "grad_norm": 11.2483491897583,
      "learning_rate": 8.74974974974975e-06,
      "loss": 0.3616,
      "step": 12490
    },
    {
      "epoch": 37.54,
      "grad_norm": 11.386582374572754,
      "learning_rate": 8.74874874874875e-06,
      "loss": 0.3986,
      "step": 12500
    },
    {
      "epoch": 37.57,
      "grad_norm": 7.893359184265137,
      "learning_rate": 8.747747747747748e-06,
      "loss": 0.3932,
      "step": 12510
    },
    {
      "epoch": 37.6,
      "grad_norm": 8.38857650756836,
      "learning_rate": 8.746746746746746e-06,
      "loss": 0.4209,
      "step": 12520
    },
    {
      "epoch": 37.63,
      "grad_norm": 12.934131622314453,
      "learning_rate": 8.745745745745747e-06,
      "loss": 0.4506,
      "step": 12530
    },
    {
      "epoch": 37.66,
      "grad_norm": 11.205546379089355,
      "learning_rate": 8.744744744744745e-06,
      "loss": 0.4014,
      "step": 12540
    },
    {
      "epoch": 37.69,
      "grad_norm": 51.77103042602539,
      "learning_rate": 8.743743743743744e-06,
      "loss": 0.3818,
      "step": 12550
    },
    {
      "epoch": 37.72,
      "grad_norm": 12.370538711547852,
      "learning_rate": 8.742742742742744e-06,
      "loss": 0.3402,
      "step": 12560
    },
    {
      "epoch": 37.75,
      "grad_norm": 8.91227912902832,
      "learning_rate": 8.741741741741743e-06,
      "loss": 0.3446,
      "step": 12570
    },
    {
      "epoch": 37.78,
      "grad_norm": 9.033913612365723,
      "learning_rate": 8.740740740740741e-06,
      "loss": 0.3339,
      "step": 12580
    },
    {
      "epoch": 37.81,
      "grad_norm": 15.76467514038086,
      "learning_rate": 8.739739739739742e-06,
      "loss": 0.4333,
      "step": 12590
    },
    {
      "epoch": 37.84,
      "grad_norm": 12.027488708496094,
      "learning_rate": 8.738738738738739e-06,
      "loss": 0.353,
      "step": 12600
    },
    {
      "epoch": 37.87,
      "grad_norm": 11.376276016235352,
      "learning_rate": 8.737737737737739e-06,
      "loss": 0.4121,
      "step": 12610
    },
    {
      "epoch": 37.9,
      "grad_norm": 9.924150466918945,
      "learning_rate": 8.736736736736737e-06,
      "loss": 0.3352,
      "step": 12620
    },
    {
      "epoch": 37.93,
      "grad_norm": 11.061473846435547,
      "learning_rate": 8.735735735735736e-06,
      "loss": 0.4188,
      "step": 12630
    },
    {
      "epoch": 37.96,
      "grad_norm": 9.54149341583252,
      "learning_rate": 8.734734734734736e-06,
      "loss": 0.4167,
      "step": 12640
    },
    {
      "epoch": 37.99,
      "grad_norm": 10.981230735778809,
      "learning_rate": 8.733733733733735e-06,
      "loss": 0.4458,
      "step": 12650
    },
    {
      "epoch": 38.0,
      "eval_accuracy": 0.9281,
      "eval_loss": 0.2582415044307709,
      "eval_runtime": 30.5044,
      "eval_samples_per_second": 327.822,
      "eval_steps_per_second": 1.311,
      "step": 12654
    },
    {
      "epoch": 38.02,
      "grad_norm": 21.35448455810547,
      "learning_rate": 8.732732732732733e-06,
      "loss": 0.3786,
      "step": 12660
    },
    {
      "epoch": 38.05,
      "grad_norm": 11.46558952331543,
      "learning_rate": 8.731731731731734e-06,
      "loss": 0.3863,
      "step": 12670
    },
    {
      "epoch": 38.08,
      "grad_norm": 10.586058616638184,
      "learning_rate": 8.73073073073073e-06,
      "loss": 0.3799,
      "step": 12680
    },
    {
      "epoch": 38.11,
      "grad_norm": 10.60461139678955,
      "learning_rate": 8.72972972972973e-06,
      "loss": 0.3601,
      "step": 12690
    },
    {
      "epoch": 38.14,
      "grad_norm": 9.384730339050293,
      "learning_rate": 8.72872872872873e-06,
      "loss": 0.4018,
      "step": 12700
    },
    {
      "epoch": 38.17,
      "grad_norm": 8.258257865905762,
      "learning_rate": 8.727727727727728e-06,
      "loss": 0.3671,
      "step": 12710
    },
    {
      "epoch": 38.2,
      "grad_norm": 8.778036117553711,
      "learning_rate": 8.726726726726727e-06,
      "loss": 0.4141,
      "step": 12720
    },
    {
      "epoch": 38.23,
      "grad_norm": 23.05480194091797,
      "learning_rate": 8.725725725725727e-06,
      "loss": 0.3664,
      "step": 12730
    },
    {
      "epoch": 38.26,
      "grad_norm": 14.267335891723633,
      "learning_rate": 8.724724724724726e-06,
      "loss": 0.3887,
      "step": 12740
    },
    {
      "epoch": 38.29,
      "grad_norm": 10.668434143066406,
      "learning_rate": 8.723723723723724e-06,
      "loss": 0.4095,
      "step": 12750
    },
    {
      "epoch": 38.32,
      "grad_norm": 9.362272262573242,
      "learning_rate": 8.722722722722723e-06,
      "loss": 0.4313,
      "step": 12760
    },
    {
      "epoch": 38.35,
      "grad_norm": 9.018074989318848,
      "learning_rate": 8.721721721721721e-06,
      "loss": 0.4027,
      "step": 12770
    },
    {
      "epoch": 38.38,
      "grad_norm": 6.7046051025390625,
      "learning_rate": 8.720720720720722e-06,
      "loss": 0.3636,
      "step": 12780
    },
    {
      "epoch": 38.41,
      "grad_norm": 14.318246841430664,
      "learning_rate": 8.71971971971972e-06,
      "loss": 0.3956,
      "step": 12790
    },
    {
      "epoch": 38.44,
      "grad_norm": 14.366408348083496,
      "learning_rate": 8.718718718718719e-06,
      "loss": 0.412,
      "step": 12800
    },
    {
      "epoch": 38.47,
      "grad_norm": 10.622831344604492,
      "learning_rate": 8.71771771771772e-06,
      "loss": 0.3799,
      "step": 12810
    },
    {
      "epoch": 38.5,
      "grad_norm": 7.679351806640625,
      "learning_rate": 8.716716716716718e-06,
      "loss": 0.406,
      "step": 12820
    },
    {
      "epoch": 38.53,
      "grad_norm": 10.854029655456543,
      "learning_rate": 8.715715715715716e-06,
      "loss": 0.3781,
      "step": 12830
    },
    {
      "epoch": 38.56,
      "grad_norm": 10.242359161376953,
      "learning_rate": 8.714714714714717e-06,
      "loss": 0.3905,
      "step": 12840
    },
    {
      "epoch": 38.59,
      "grad_norm": 17.268033981323242,
      "learning_rate": 8.713713713713714e-06,
      "loss": 0.3843,
      "step": 12850
    },
    {
      "epoch": 38.62,
      "grad_norm": 14.737871170043945,
      "learning_rate": 8.712712712712714e-06,
      "loss": 0.435,
      "step": 12860
    },
    {
      "epoch": 38.65,
      "grad_norm": 9.059165954589844,
      "learning_rate": 8.711711711711712e-06,
      "loss": 0.3855,
      "step": 12870
    },
    {
      "epoch": 38.68,
      "grad_norm": 10.406525611877441,
      "learning_rate": 8.710710710710711e-06,
      "loss": 0.4175,
      "step": 12880
    },
    {
      "epoch": 38.71,
      "grad_norm": 7.349973678588867,
      "learning_rate": 8.709709709709711e-06,
      "loss": 0.4328,
      "step": 12890
    },
    {
      "epoch": 38.74,
      "grad_norm": 12.82719612121582,
      "learning_rate": 8.70870870870871e-06,
      "loss": 0.4047,
      "step": 12900
    },
    {
      "epoch": 38.77,
      "grad_norm": 7.317464828491211,
      "learning_rate": 8.707707707707708e-06,
      "loss": 0.3422,
      "step": 12910
    },
    {
      "epoch": 38.8,
      "grad_norm": 8.046574592590332,
      "learning_rate": 8.706706706706707e-06,
      "loss": 0.3867,
      "step": 12920
    },
    {
      "epoch": 38.83,
      "grad_norm": 7.7277913093566895,
      "learning_rate": 8.705705705705706e-06,
      "loss": 0.378,
      "step": 12930
    },
    {
      "epoch": 38.86,
      "grad_norm": 8.408747673034668,
      "learning_rate": 8.704704704704704e-06,
      "loss": 0.3773,
      "step": 12940
    },
    {
      "epoch": 38.89,
      "grad_norm": 32.56877899169922,
      "learning_rate": 8.703703703703705e-06,
      "loss": 0.4154,
      "step": 12950
    },
    {
      "epoch": 38.92,
      "grad_norm": 6.650607585906982,
      "learning_rate": 8.702702702702703e-06,
      "loss": 0.4141,
      "step": 12960
    },
    {
      "epoch": 38.95,
      "grad_norm": 76.25609588623047,
      "learning_rate": 8.701701701701702e-06,
      "loss": 0.4162,
      "step": 12970
    },
    {
      "epoch": 38.98,
      "grad_norm": 15.734679222106934,
      "learning_rate": 8.700700700700702e-06,
      "loss": 0.3999,
      "step": 12980
    },
    {
      "epoch": 39.0,
      "eval_accuracy": 0.928,
      "eval_loss": 0.2580805718898773,
      "eval_runtime": 30.6861,
      "eval_samples_per_second": 325.88,
      "eval_steps_per_second": 1.304,
      "step": 12987
    },
    {
      "epoch": 39.01,
      "grad_norm": 13.652566909790039,
      "learning_rate": 8.6996996996997e-06,
      "loss": 0.3645,
      "step": 12990
    },
    {
      "epoch": 39.04,
      "grad_norm": 9.84151840209961,
      "learning_rate": 8.6986986986987e-06,
      "loss": 0.3847,
      "step": 13000
    },
    {
      "epoch": 39.07,
      "grad_norm": 9.637248039245605,
      "learning_rate": 8.697697697697698e-06,
      "loss": 0.3613,
      "step": 13010
    },
    {
      "epoch": 39.1,
      "grad_norm": 11.930158615112305,
      "learning_rate": 8.696696696696696e-06,
      "loss": 0.435,
      "step": 13020
    },
    {
      "epoch": 39.13,
      "grad_norm": 24.212404251098633,
      "learning_rate": 8.695695695695697e-06,
      "loss": 0.3543,
      "step": 13030
    },
    {
      "epoch": 39.16,
      "grad_norm": 8.99075984954834,
      "learning_rate": 8.694694694694695e-06,
      "loss": 0.4157,
      "step": 13040
    },
    {
      "epoch": 39.19,
      "grad_norm": 10.34455680847168,
      "learning_rate": 8.693693693693694e-06,
      "loss": 0.3976,
      "step": 13050
    },
    {
      "epoch": 39.22,
      "grad_norm": 11.143070220947266,
      "learning_rate": 8.692692692692694e-06,
      "loss": 0.3574,
      "step": 13060
    },
    {
      "epoch": 39.25,
      "grad_norm": 11.767276763916016,
      "learning_rate": 8.691691691691693e-06,
      "loss": 0.3517,
      "step": 13070
    },
    {
      "epoch": 39.28,
      "grad_norm": 6.911677837371826,
      "learning_rate": 8.690690690690691e-06,
      "loss": 0.3652,
      "step": 13080
    },
    {
      "epoch": 39.31,
      "grad_norm": 16.10271453857422,
      "learning_rate": 8.689689689689692e-06,
      "loss": 0.4658,
      "step": 13090
    },
    {
      "epoch": 39.34,
      "grad_norm": 11.662455558776855,
      "learning_rate": 8.688688688688689e-06,
      "loss": 0.4222,
      "step": 13100
    },
    {
      "epoch": 39.37,
      "grad_norm": 10.91153335571289,
      "learning_rate": 8.687687687687689e-06,
      "loss": 0.4105,
      "step": 13110
    },
    {
      "epoch": 39.4,
      "grad_norm": 23.97760772705078,
      "learning_rate": 8.686686686686687e-06,
      "loss": 0.3721,
      "step": 13120
    },
    {
      "epoch": 39.43,
      "grad_norm": 7.614724159240723,
      "learning_rate": 8.685685685685686e-06,
      "loss": 0.424,
      "step": 13130
    },
    {
      "epoch": 39.46,
      "grad_norm": 12.252328872680664,
      "learning_rate": 8.684684684684686e-06,
      "loss": 0.3959,
      "step": 13140
    },
    {
      "epoch": 39.49,
      "grad_norm": 10.605676651000977,
      "learning_rate": 8.683683683683685e-06,
      "loss": 0.3853,
      "step": 13150
    },
    {
      "epoch": 39.52,
      "grad_norm": 9.702637672424316,
      "learning_rate": 8.682682682682684e-06,
      "loss": 0.3547,
      "step": 13160
    },
    {
      "epoch": 39.55,
      "grad_norm": 8.899740219116211,
      "learning_rate": 8.681681681681682e-06,
      "loss": 0.4048,
      "step": 13170
    },
    {
      "epoch": 39.58,
      "grad_norm": 11.565757751464844,
      "learning_rate": 8.68068068068068e-06,
      "loss": 0.4059,
      "step": 13180
    },
    {
      "epoch": 39.61,
      "grad_norm": 10.45947551727295,
      "learning_rate": 8.67967967967968e-06,
      "loss": 0.388,
      "step": 13190
    },
    {
      "epoch": 39.64,
      "grad_norm": 8.317366600036621,
      "learning_rate": 8.67867867867868e-06,
      "loss": 0.4108,
      "step": 13200
    },
    {
      "epoch": 39.67,
      "grad_norm": 6.8623433113098145,
      "learning_rate": 8.677677677677678e-06,
      "loss": 0.4123,
      "step": 13210
    },
    {
      "epoch": 39.7,
      "grad_norm": 5.949448108673096,
      "learning_rate": 8.676676676676677e-06,
      "loss": 0.407,
      "step": 13220
    },
    {
      "epoch": 39.73,
      "grad_norm": 9.582085609436035,
      "learning_rate": 8.675675675675677e-06,
      "loss": 0.4188,
      "step": 13230
    },
    {
      "epoch": 39.76,
      "grad_norm": 10.572515487670898,
      "learning_rate": 8.674674674674676e-06,
      "loss": 0.4289,
      "step": 13240
    },
    {
      "epoch": 39.79,
      "grad_norm": 17.77850341796875,
      "learning_rate": 8.673673673673674e-06,
      "loss": 0.4613,
      "step": 13250
    },
    {
      "epoch": 39.82,
      "grad_norm": 17.842029571533203,
      "learning_rate": 8.672672672672673e-06,
      "loss": 0.4173,
      "step": 13260
    },
    {
      "epoch": 39.85,
      "grad_norm": 11.029363632202148,
      "learning_rate": 8.671671671671671e-06,
      "loss": 0.4433,
      "step": 13270
    },
    {
      "epoch": 39.88,
      "grad_norm": 11.918964385986328,
      "learning_rate": 8.670670670670672e-06,
      "loss": 0.3568,
      "step": 13280
    },
    {
      "epoch": 39.91,
      "grad_norm": 13.341336250305176,
      "learning_rate": 8.66966966966967e-06,
      "loss": 0.3835,
      "step": 13290
    },
    {
      "epoch": 39.94,
      "grad_norm": 6.750630855560303,
      "learning_rate": 8.668668668668669e-06,
      "loss": 0.3711,
      "step": 13300
    },
    {
      "epoch": 39.97,
      "grad_norm": 23.237773895263672,
      "learning_rate": 8.66766766766767e-06,
      "loss": 0.3866,
      "step": 13310
    },
    {
      "epoch": 40.0,
      "grad_norm": 6.4752349853515625,
      "learning_rate": 8.666666666666668e-06,
      "loss": 0.3887,
      "step": 13320
    },
    {
      "epoch": 40.0,
      "eval_accuracy": 0.9301,
      "eval_loss": 0.25753843784332275,
      "eval_runtime": 31.0782,
      "eval_samples_per_second": 321.769,
      "eval_steps_per_second": 1.287,
      "step": 13320
    },
    {
      "epoch": 40.03,
      "grad_norm": 8.84365463256836,
      "learning_rate": 8.665665665665666e-06,
      "loss": 0.4027,
      "step": 13330
    },
    {
      "epoch": 40.06,
      "grad_norm": 11.843667030334473,
      "learning_rate": 8.664664664664665e-06,
      "loss": 0.4208,
      "step": 13340
    },
    {
      "epoch": 40.09,
      "grad_norm": 13.39307689666748,
      "learning_rate": 8.663663663663664e-06,
      "loss": 0.3953,
      "step": 13350
    },
    {
      "epoch": 40.12,
      "grad_norm": 16.940399169921875,
      "learning_rate": 8.662662662662664e-06,
      "loss": 0.4185,
      "step": 13360
    },
    {
      "epoch": 40.15,
      "grad_norm": 29.849838256835938,
      "learning_rate": 8.661661661661662e-06,
      "loss": 0.4294,
      "step": 13370
    },
    {
      "epoch": 40.18,
      "grad_norm": 19.348379135131836,
      "learning_rate": 8.660660660660661e-06,
      "loss": 0.4291,
      "step": 13380
    },
    {
      "epoch": 40.21,
      "grad_norm": 11.171587944030762,
      "learning_rate": 8.65965965965966e-06,
      "loss": 0.3925,
      "step": 13390
    },
    {
      "epoch": 40.24,
      "grad_norm": 10.444419860839844,
      "learning_rate": 8.65865865865866e-06,
      "loss": 0.3902,
      "step": 13400
    },
    {
      "epoch": 40.27,
      "grad_norm": 9.350363731384277,
      "learning_rate": 8.657657657657659e-06,
      "loss": 0.3961,
      "step": 13410
    },
    {
      "epoch": 40.3,
      "grad_norm": 10.349617958068848,
      "learning_rate": 8.656656656656657e-06,
      "loss": 0.3664,
      "step": 13420
    },
    {
      "epoch": 40.33,
      "grad_norm": 13.25471019744873,
      "learning_rate": 8.655655655655656e-06,
      "loss": 0.3899,
      "step": 13430
    },
    {
      "epoch": 40.36,
      "grad_norm": 8.114799499511719,
      "learning_rate": 8.654654654654654e-06,
      "loss": 0.3527,
      "step": 13440
    },
    {
      "epoch": 40.39,
      "grad_norm": 12.502910614013672,
      "learning_rate": 8.653653653653655e-06,
      "loss": 0.4763,
      "step": 13450
    },
    {
      "epoch": 40.42,
      "grad_norm": 7.282385349273682,
      "learning_rate": 8.652652652652653e-06,
      "loss": 0.4069,
      "step": 13460
    },
    {
      "epoch": 40.45,
      "grad_norm": 13.283368110656738,
      "learning_rate": 8.651651651651652e-06,
      "loss": 0.3433,
      "step": 13470
    },
    {
      "epoch": 40.48,
      "grad_norm": 9.808856964111328,
      "learning_rate": 8.650650650650652e-06,
      "loss": 0.3796,
      "step": 13480
    },
    {
      "epoch": 40.51,
      "grad_norm": 7.393431186676025,
      "learning_rate": 8.64964964964965e-06,
      "loss": 0.3883,
      "step": 13490
    },
    {
      "epoch": 40.54,
      "grad_norm": 9.233526229858398,
      "learning_rate": 8.64864864864865e-06,
      "loss": 0.3624,
      "step": 13500
    },
    {
      "epoch": 40.57,
      "grad_norm": 6.700456142425537,
      "learning_rate": 8.647647647647648e-06,
      "loss": 0.4107,
      "step": 13510
    },
    {
      "epoch": 40.6,
      "grad_norm": 9.253952980041504,
      "learning_rate": 8.646646646646646e-06,
      "loss": 0.3815,
      "step": 13520
    },
    {
      "epoch": 40.63,
      "grad_norm": 8.06972885131836,
      "learning_rate": 8.645645645645647e-06,
      "loss": 0.3665,
      "step": 13530
    },
    {
      "epoch": 40.66,
      "grad_norm": 11.388467788696289,
      "learning_rate": 8.644644644644645e-06,
      "loss": 0.3937,
      "step": 13540
    },
    {
      "epoch": 40.69,
      "grad_norm": 10.508780479431152,
      "learning_rate": 8.643643643643644e-06,
      "loss": 0.3994,
      "step": 13550
    },
    {
      "epoch": 40.72,
      "grad_norm": 9.98808479309082,
      "learning_rate": 8.642642642642644e-06,
      "loss": 0.3899,
      "step": 13560
    },
    {
      "epoch": 40.75,
      "grad_norm": 12.760663986206055,
      "learning_rate": 8.641641641641643e-06,
      "loss": 0.3792,
      "step": 13570
    },
    {
      "epoch": 40.78,
      "grad_norm": 7.717315196990967,
      "learning_rate": 8.640640640640641e-06,
      "loss": 0.346,
      "step": 13580
    },
    {
      "epoch": 40.81,
      "grad_norm": 11.570375442504883,
      "learning_rate": 8.63963963963964e-06,
      "loss": 0.3796,
      "step": 13590
    },
    {
      "epoch": 40.84,
      "grad_norm": 32.82620620727539,
      "learning_rate": 8.638638638638639e-06,
      "loss": 0.3973,
      "step": 13600
    },
    {
      "epoch": 40.87,
      "grad_norm": 14.685288429260254,
      "learning_rate": 8.637637637637637e-06,
      "loss": 0.3947,
      "step": 13610
    },
    {
      "epoch": 40.9,
      "grad_norm": 9.27966022491455,
      "learning_rate": 8.636636636636638e-06,
      "loss": 0.3774,
      "step": 13620
    },
    {
      "epoch": 40.93,
      "grad_norm": 8.751079559326172,
      "learning_rate": 8.635635635635636e-06,
      "loss": 0.4394,
      "step": 13630
    },
    {
      "epoch": 40.96,
      "grad_norm": 7.750463485717773,
      "learning_rate": 8.634634634634635e-06,
      "loss": 0.3635,
      "step": 13640
    },
    {
      "epoch": 40.99,
      "grad_norm": 9.266974449157715,
      "learning_rate": 8.633633633633635e-06,
      "loss": 0.4101,
      "step": 13650
    },
    {
      "epoch": 41.0,
      "eval_accuracy": 0.9278,
      "eval_loss": 0.2619706690311432,
      "eval_runtime": 30.5563,
      "eval_samples_per_second": 327.265,
      "eval_steps_per_second": 1.309,
      "step": 13653
    },
    {
      "epoch": 41.02,
      "grad_norm": 6.554638385772705,
      "learning_rate": 8.632632632632634e-06,
      "loss": 0.3542,
      "step": 13660
    },
    {
      "epoch": 41.05,
      "grad_norm": 7.143919467926025,
      "learning_rate": 8.631631631631632e-06,
      "loss": 0.3327,
      "step": 13670
    },
    {
      "epoch": 41.08,
      "grad_norm": 11.283164024353027,
      "learning_rate": 8.63063063063063e-06,
      "loss": 0.3718,
      "step": 13680
    },
    {
      "epoch": 41.11,
      "grad_norm": 14.787487030029297,
      "learning_rate": 8.62962962962963e-06,
      "loss": 0.3343,
      "step": 13690
    },
    {
      "epoch": 41.14,
      "grad_norm": 21.484725952148438,
      "learning_rate": 8.62862862862863e-06,
      "loss": 0.3884,
      "step": 13700
    },
    {
      "epoch": 41.17,
      "grad_norm": 10.64600944519043,
      "learning_rate": 8.627627627627628e-06,
      "loss": 0.4114,
      "step": 13710
    },
    {
      "epoch": 41.2,
      "grad_norm": 11.60208797454834,
      "learning_rate": 8.626626626626627e-06,
      "loss": 0.3539,
      "step": 13720
    },
    {
      "epoch": 41.23,
      "grad_norm": 11.907629013061523,
      "learning_rate": 8.625625625625627e-06,
      "loss": 0.3628,
      "step": 13730
    },
    {
      "epoch": 41.26,
      "grad_norm": 13.026467323303223,
      "learning_rate": 8.624624624624626e-06,
      "loss": 0.4061,
      "step": 13740
    },
    {
      "epoch": 41.29,
      "grad_norm": 6.0617451667785645,
      "learning_rate": 8.623623623623624e-06,
      "loss": 0.3874,
      "step": 13750
    },
    {
      "epoch": 41.32,
      "grad_norm": 8.921332359313965,
      "learning_rate": 8.622622622622623e-06,
      "loss": 0.3738,
      "step": 13760
    },
    {
      "epoch": 41.35,
      "grad_norm": 10.881823539733887,
      "learning_rate": 8.621621621621622e-06,
      "loss": 0.3578,
      "step": 13770
    },
    {
      "epoch": 41.38,
      "grad_norm": 25.958070755004883,
      "learning_rate": 8.620620620620622e-06,
      "loss": 0.4732,
      "step": 13780
    },
    {
      "epoch": 41.41,
      "grad_norm": 10.085030555725098,
      "learning_rate": 8.61961961961962e-06,
      "loss": 0.3866,
      "step": 13790
    },
    {
      "epoch": 41.44,
      "grad_norm": 10.821370124816895,
      "learning_rate": 8.618618618618619e-06,
      "loss": 0.3676,
      "step": 13800
    },
    {
      "epoch": 41.47,
      "grad_norm": 13.257043838500977,
      "learning_rate": 8.61761761761762e-06,
      "loss": 0.4041,
      "step": 13810
    },
    {
      "epoch": 41.5,
      "grad_norm": 24.02643585205078,
      "learning_rate": 8.616616616616618e-06,
      "loss": 0.3675,
      "step": 13820
    },
    {
      "epoch": 41.53,
      "grad_norm": 12.339000701904297,
      "learning_rate": 8.615615615615616e-06,
      "loss": 0.4197,
      "step": 13830
    },
    {
      "epoch": 41.56,
      "grad_norm": 8.537544250488281,
      "learning_rate": 8.614614614614615e-06,
      "loss": 0.401,
      "step": 13840
    },
    {
      "epoch": 41.59,
      "grad_norm": 14.730378150939941,
      "learning_rate": 8.613613613613614e-06,
      "loss": 0.3784,
      "step": 13850
    },
    {
      "epoch": 41.62,
      "grad_norm": 6.818876266479492,
      "learning_rate": 8.612612612612612e-06,
      "loss": 0.3464,
      "step": 13860
    },
    {
      "epoch": 41.65,
      "grad_norm": 16.45130157470703,
      "learning_rate": 8.611611611611613e-06,
      "loss": 0.3819,
      "step": 13870
    },
    {
      "epoch": 41.68,
      "grad_norm": 9.518233299255371,
      "learning_rate": 8.610610610610611e-06,
      "loss": 0.3519,
      "step": 13880
    },
    {
      "epoch": 41.71,
      "grad_norm": 11.735335350036621,
      "learning_rate": 8.60960960960961e-06,
      "loss": 0.3539,
      "step": 13890
    },
    {
      "epoch": 41.74,
      "grad_norm": 9.027347564697266,
      "learning_rate": 8.60860860860861e-06,
      "loss": 0.4067,
      "step": 13900
    },
    {
      "epoch": 41.77,
      "grad_norm": 11.253780364990234,
      "learning_rate": 8.607607607607609e-06,
      "loss": 0.4049,
      "step": 13910
    },
    {
      "epoch": 41.8,
      "grad_norm": 14.579793930053711,
      "learning_rate": 8.606606606606607e-06,
      "loss": 0.4027,
      "step": 13920
    },
    {
      "epoch": 41.83,
      "grad_norm": 12.6600980758667,
      "learning_rate": 8.605605605605606e-06,
      "loss": 0.3902,
      "step": 13930
    },
    {
      "epoch": 41.86,
      "grad_norm": 10.267448425292969,
      "learning_rate": 8.604604604604604e-06,
      "loss": 0.3513,
      "step": 13940
    },
    {
      "epoch": 41.89,
      "grad_norm": 8.520174980163574,
      "learning_rate": 8.603603603603605e-06,
      "loss": 0.3551,
      "step": 13950
    },
    {
      "epoch": 41.92,
      "grad_norm": 8.369441032409668,
      "learning_rate": 8.602602602602603e-06,
      "loss": 0.352,
      "step": 13960
    },
    {
      "epoch": 41.95,
      "grad_norm": 14.12592887878418,
      "learning_rate": 8.601601601601602e-06,
      "loss": 0.3594,
      "step": 13970
    },
    {
      "epoch": 41.98,
      "grad_norm": 8.744529724121094,
      "learning_rate": 8.600600600600602e-06,
      "loss": 0.3987,
      "step": 13980
    },
    {
      "epoch": 42.0,
      "eval_accuracy": 0.928,
      "eval_loss": 0.2626796364784241,
      "eval_runtime": 30.0792,
      "eval_samples_per_second": 332.456,
      "eval_steps_per_second": 1.33,
      "step": 13986
    },
    {
      "epoch": 42.01,
      "grad_norm": 10.510129928588867,
      "learning_rate": 8.5995995995996e-06,
      "loss": 0.3248,
      "step": 13990
    },
    {
      "epoch": 42.04,
      "grad_norm": 9.703568458557129,
      "learning_rate": 8.5985985985986e-06,
      "loss": 0.378,
      "step": 14000
    },
    {
      "epoch": 42.07,
      "grad_norm": 8.442028045654297,
      "learning_rate": 8.597597597597598e-06,
      "loss": 0.3951,
      "step": 14010
    },
    {
      "epoch": 42.1,
      "grad_norm": 11.244294166564941,
      "learning_rate": 8.596596596596597e-06,
      "loss": 0.3742,
      "step": 14020
    },
    {
      "epoch": 42.13,
      "grad_norm": 13.495356559753418,
      "learning_rate": 8.595595595595597e-06,
      "loss": 0.3729,
      "step": 14030
    },
    {
      "epoch": 42.16,
      "grad_norm": 13.685050964355469,
      "learning_rate": 8.594594594594595e-06,
      "loss": 0.3698,
      "step": 14040
    },
    {
      "epoch": 42.19,
      "grad_norm": 12.698522567749023,
      "learning_rate": 8.593593593593594e-06,
      "loss": 0.4271,
      "step": 14050
    },
    {
      "epoch": 42.22,
      "grad_norm": 11.278382301330566,
      "learning_rate": 8.592592592592593e-06,
      "loss": 0.354,
      "step": 14060
    },
    {
      "epoch": 42.25,
      "grad_norm": 9.81973648071289,
      "learning_rate": 8.591591591591593e-06,
      "loss": 0.4203,
      "step": 14070
    },
    {
      "epoch": 42.28,
      "grad_norm": 14.25061321258545,
      "learning_rate": 8.590590590590592e-06,
      "loss": 0.3662,
      "step": 14080
    },
    {
      "epoch": 42.31,
      "grad_norm": 10.91623592376709,
      "learning_rate": 8.58958958958959e-06,
      "loss": 0.4054,
      "step": 14090
    },
    {
      "epoch": 42.34,
      "grad_norm": 7.691105365753174,
      "learning_rate": 8.588588588588589e-06,
      "loss": 0.4248,
      "step": 14100
    },
    {
      "epoch": 42.37,
      "grad_norm": 11.796515464782715,
      "learning_rate": 8.587587587587587e-06,
      "loss": 0.3449,
      "step": 14110
    },
    {
      "epoch": 42.4,
      "grad_norm": 10.914915084838867,
      "learning_rate": 8.586586586586588e-06,
      "loss": 0.4013,
      "step": 14120
    },
    {
      "epoch": 42.43,
      "grad_norm": 8.735807418823242,
      "learning_rate": 8.585585585585586e-06,
      "loss": 0.3167,
      "step": 14130
    },
    {
      "epoch": 42.46,
      "grad_norm": 9.286930084228516,
      "learning_rate": 8.584584584584585e-06,
      "loss": 0.3557,
      "step": 14140
    },
    {
      "epoch": 42.49,
      "grad_norm": 11.611909866333008,
      "learning_rate": 8.583583583583585e-06,
      "loss": 0.3632,
      "step": 14150
    },
    {
      "epoch": 42.52,
      "grad_norm": 8.441341400146484,
      "learning_rate": 8.582582582582584e-06,
      "loss": 0.4399,
      "step": 14160
    },
    {
      "epoch": 42.55,
      "grad_norm": 11.103727340698242,
      "learning_rate": 8.581581581581582e-06,
      "loss": 0.434,
      "step": 14170
    },
    {
      "epoch": 42.58,
      "grad_norm": 10.919816017150879,
      "learning_rate": 8.580580580580581e-06,
      "loss": 0.3968,
      "step": 14180
    },
    {
      "epoch": 42.61,
      "grad_norm": 13.054405212402344,
      "learning_rate": 8.57957957957958e-06,
      "loss": 0.3848,
      "step": 14190
    },
    {
      "epoch": 42.64,
      "grad_norm": 9.721436500549316,
      "learning_rate": 8.57857857857858e-06,
      "loss": 0.3634,
      "step": 14200
    },
    {
      "epoch": 42.67,
      "grad_norm": 13.500628471374512,
      "learning_rate": 8.577577577577578e-06,
      "loss": 0.4341,
      "step": 14210
    },
    {
      "epoch": 42.7,
      "grad_norm": 8.16736888885498,
      "learning_rate": 8.576576576576577e-06,
      "loss": 0.3648,
      "step": 14220
    },
    {
      "epoch": 42.73,
      "grad_norm": 10.515711784362793,
      "learning_rate": 8.575575575575577e-06,
      "loss": 0.3518,
      "step": 14230
    },
    {
      "epoch": 42.76,
      "grad_norm": 15.003644943237305,
      "learning_rate": 8.574574574574576e-06,
      "loss": 0.409,
      "step": 14240
    },
    {
      "epoch": 42.79,
      "grad_norm": 11.138490676879883,
      "learning_rate": 8.573573573573574e-06,
      "loss": 0.4029,
      "step": 14250
    },
    {
      "epoch": 42.82,
      "grad_norm": 8.122265815734863,
      "learning_rate": 8.572572572572573e-06,
      "loss": 0.402,
      "step": 14260
    },
    {
      "epoch": 42.85,
      "grad_norm": 8.130270004272461,
      "learning_rate": 8.571571571571572e-06,
      "loss": 0.3486,
      "step": 14270
    },
    {
      "epoch": 42.88,
      "grad_norm": 7.230751991271973,
      "learning_rate": 8.570570570570572e-06,
      "loss": 0.3372,
      "step": 14280
    },
    {
      "epoch": 42.91,
      "grad_norm": 11.794231414794922,
      "learning_rate": 8.56956956956957e-06,
      "loss": 0.4256,
      "step": 14290
    },
    {
      "epoch": 42.94,
      "grad_norm": 10.275702476501465,
      "learning_rate": 8.568568568568569e-06,
      "loss": 0.3843,
      "step": 14300
    },
    {
      "epoch": 42.97,
      "grad_norm": 10.751091957092285,
      "learning_rate": 8.567567567567568e-06,
      "loss": 0.3514,
      "step": 14310
    },
    {
      "epoch": 43.0,
      "eval_accuracy": 0.9287,
      "eval_loss": 0.265987366437912,
      "eval_runtime": 29.984,
      "eval_samples_per_second": 333.511,
      "eval_steps_per_second": 1.334,
      "step": 14319
    },
    {
      "epoch": 43.0,
      "grad_norm": 10.239645004272461,
      "learning_rate": 8.566566566566568e-06,
      "loss": 0.3879,
      "step": 14320
    },
    {
      "epoch": 43.03,
      "grad_norm": 7.008784770965576,
      "learning_rate": 8.565565565565567e-06,
      "loss": 0.3856,
      "step": 14330
    },
    {
      "epoch": 43.06,
      "grad_norm": 12.353731155395508,
      "learning_rate": 8.564564564564565e-06,
      "loss": 0.4436,
      "step": 14340
    },
    {
      "epoch": 43.09,
      "grad_norm": 11.370824813842773,
      "learning_rate": 8.563563563563564e-06,
      "loss": 0.394,
      "step": 14350
    },
    {
      "epoch": 43.12,
      "grad_norm": 8.463004112243652,
      "learning_rate": 8.562562562562562e-06,
      "loss": 0.3474,
      "step": 14360
    },
    {
      "epoch": 43.15,
      "grad_norm": 20.94791030883789,
      "learning_rate": 8.561561561561563e-06,
      "loss": 0.3789,
      "step": 14370
    },
    {
      "epoch": 43.18,
      "grad_norm": 10.64284896850586,
      "learning_rate": 8.560560560560561e-06,
      "loss": 0.3546,
      "step": 14380
    },
    {
      "epoch": 43.21,
      "grad_norm": 12.647034645080566,
      "learning_rate": 8.55955955955956e-06,
      "loss": 0.3999,
      "step": 14390
    },
    {
      "epoch": 43.24,
      "grad_norm": 18.13320541381836,
      "learning_rate": 8.55855855855856e-06,
      "loss": 0.3239,
      "step": 14400
    },
    {
      "epoch": 43.27,
      "grad_norm": 20.47588348388672,
      "learning_rate": 8.557557557557559e-06,
      "loss": 0.4029,
      "step": 14410
    },
    {
      "epoch": 43.3,
      "grad_norm": 9.143036842346191,
      "learning_rate": 8.556556556556557e-06,
      "loss": 0.3399,
      "step": 14420
    },
    {
      "epoch": 43.33,
      "grad_norm": 14.88934326171875,
      "learning_rate": 8.555555555555556e-06,
      "loss": 0.4084,
      "step": 14430
    },
    {
      "epoch": 43.36,
      "grad_norm": 8.715350151062012,
      "learning_rate": 8.554554554554554e-06,
      "loss": 0.3694,
      "step": 14440
    },
    {
      "epoch": 43.39,
      "grad_norm": 9.536194801330566,
      "learning_rate": 8.553553553553555e-06,
      "loss": 0.3866,
      "step": 14450
    },
    {
      "epoch": 43.42,
      "grad_norm": 17.111906051635742,
      "learning_rate": 8.552552552552553e-06,
      "loss": 0.3892,
      "step": 14460
    },
    {
      "epoch": 43.45,
      "grad_norm": 13.305580139160156,
      "learning_rate": 8.551551551551552e-06,
      "loss": 0.3456,
      "step": 14470
    },
    {
      "epoch": 43.48,
      "grad_norm": 6.74718713760376,
      "learning_rate": 8.550550550550552e-06,
      "loss": 0.4062,
      "step": 14480
    },
    {
      "epoch": 43.51,
      "grad_norm": 9.589884757995605,
      "learning_rate": 8.549549549549551e-06,
      "loss": 0.3,
      "step": 14490
    },
    {
      "epoch": 43.54,
      "grad_norm": 6.705572605133057,
      "learning_rate": 8.54854854854855e-06,
      "loss": 0.3563,
      "step": 14500
    },
    {
      "epoch": 43.57,
      "grad_norm": 11.034214973449707,
      "learning_rate": 8.547547547547548e-06,
      "loss": 0.405,
      "step": 14510
    },
    {
      "epoch": 43.6,
      "grad_norm": 8.227956771850586,
      "learning_rate": 8.546546546546547e-06,
      "loss": 0.33,
      "step": 14520
    },
    {
      "epoch": 43.63,
      "grad_norm": 8.188857078552246,
      "learning_rate": 8.545545545545545e-06,
      "loss": 0.3673,
      "step": 14530
    },
    {
      "epoch": 43.66,
      "grad_norm": 12.041780471801758,
      "learning_rate": 8.544544544544546e-06,
      "loss": 0.3325,
      "step": 14540
    },
    {
      "epoch": 43.69,
      "grad_norm": 12.44257640838623,
      "learning_rate": 8.543543543543544e-06,
      "loss": 0.3968,
      "step": 14550
    },
    {
      "epoch": 43.72,
      "grad_norm": 14.330779075622559,
      "learning_rate": 8.542542542542543e-06,
      "loss": 0.2969,
      "step": 14560
    },
    {
      "epoch": 43.75,
      "grad_norm": 11.589505195617676,
      "learning_rate": 8.541541541541543e-06,
      "loss": 0.4342,
      "step": 14570
    },
    {
      "epoch": 43.78,
      "grad_norm": 10.879201889038086,
      "learning_rate": 8.540540540540542e-06,
      "loss": 0.3693,
      "step": 14580
    },
    {
      "epoch": 43.81,
      "grad_norm": 9.256011009216309,
      "learning_rate": 8.53953953953954e-06,
      "loss": 0.3673,
      "step": 14590
    },
    {
      "epoch": 43.84,
      "grad_norm": 10.0084810256958,
      "learning_rate": 8.538538538538539e-06,
      "loss": 0.4052,
      "step": 14600
    },
    {
      "epoch": 43.87,
      "grad_norm": 13.138761520385742,
      "learning_rate": 8.537537537537537e-06,
      "loss": 0.3528,
      "step": 14610
    },
    {
      "epoch": 43.9,
      "grad_norm": 19.79391098022461,
      "learning_rate": 8.536536536536538e-06,
      "loss": 0.4033,
      "step": 14620
    },
    {
      "epoch": 43.93,
      "grad_norm": 8.447429656982422,
      "learning_rate": 8.535535535535536e-06,
      "loss": 0.3662,
      "step": 14630
    },
    {
      "epoch": 43.96,
      "grad_norm": 13.001323699951172,
      "learning_rate": 8.534534534534535e-06,
      "loss": 0.3755,
      "step": 14640
    },
    {
      "epoch": 43.99,
      "grad_norm": 16.58340835571289,
      "learning_rate": 8.533533533533535e-06,
      "loss": 0.3916,
      "step": 14650
    },
    {
      "epoch": 44.0,
      "eval_accuracy": 0.9287,
      "eval_loss": 0.26698553562164307,
      "eval_runtime": 30.7047,
      "eval_samples_per_second": 325.684,
      "eval_steps_per_second": 1.303,
      "step": 14652
    },
    {
      "epoch": 44.02,
      "grad_norm": 6.501995086669922,
      "learning_rate": 8.532532532532534e-06,
      "loss": 0.3338,
      "step": 14660
    },
    {
      "epoch": 44.05,
      "grad_norm": 8.728530883789062,
      "learning_rate": 8.531531531531532e-06,
      "loss": 0.3562,
      "step": 14670
    },
    {
      "epoch": 44.08,
      "grad_norm": 17.454566955566406,
      "learning_rate": 8.530530530530531e-06,
      "loss": 0.3691,
      "step": 14680
    },
    {
      "epoch": 44.11,
      "grad_norm": 5.932002544403076,
      "learning_rate": 8.52952952952953e-06,
      "loss": 0.42,
      "step": 14690
    },
    {
      "epoch": 44.14,
      "grad_norm": 14.146564483642578,
      "learning_rate": 8.52852852852853e-06,
      "loss": 0.3663,
      "step": 14700
    },
    {
      "epoch": 44.17,
      "grad_norm": 8.058493614196777,
      "learning_rate": 8.527527527527528e-06,
      "loss": 0.3943,
      "step": 14710
    },
    {
      "epoch": 44.2,
      "grad_norm": 7.26616907119751,
      "learning_rate": 8.526526526526527e-06,
      "loss": 0.3452,
      "step": 14720
    },
    {
      "epoch": 44.23,
      "grad_norm": 23.906394958496094,
      "learning_rate": 8.525525525525527e-06,
      "loss": 0.344,
      "step": 14730
    },
    {
      "epoch": 44.26,
      "grad_norm": 9.583316802978516,
      "learning_rate": 8.524524524524526e-06,
      "loss": 0.4037,
      "step": 14740
    },
    {
      "epoch": 44.29,
      "grad_norm": 7.560810565948486,
      "learning_rate": 8.523523523523524e-06,
      "loss": 0.341,
      "step": 14750
    },
    {
      "epoch": 44.32,
      "grad_norm": 13.786479949951172,
      "learning_rate": 8.522522522522523e-06,
      "loss": 0.399,
      "step": 14760
    },
    {
      "epoch": 44.35,
      "grad_norm": 12.419435501098633,
      "learning_rate": 8.521521521521522e-06,
      "loss": 0.3437,
      "step": 14770
    },
    {
      "epoch": 44.38,
      "grad_norm": 12.295540809631348,
      "learning_rate": 8.52052052052052e-06,
      "loss": 0.3329,
      "step": 14780
    },
    {
      "epoch": 44.41,
      "grad_norm": 14.559554100036621,
      "learning_rate": 8.51951951951952e-06,
      "loss": 0.3736,
      "step": 14790
    },
    {
      "epoch": 44.44,
      "grad_norm": 15.304993629455566,
      "learning_rate": 8.518518518518519e-06,
      "loss": 0.4397,
      "step": 14800
    },
    {
      "epoch": 44.47,
      "grad_norm": 13.740938186645508,
      "learning_rate": 8.517517517517518e-06,
      "loss": 0.3907,
      "step": 14810
    },
    {
      "epoch": 44.5,
      "grad_norm": 7.240555286407471,
      "learning_rate": 8.516516516516518e-06,
      "loss": 0.3606,
      "step": 14820
    },
    {
      "epoch": 44.53,
      "grad_norm": 8.767868995666504,
      "learning_rate": 8.515515515515517e-06,
      "loss": 0.3659,
      "step": 14830
    },
    {
      "epoch": 44.56,
      "grad_norm": 11.760705947875977,
      "learning_rate": 8.514514514514515e-06,
      "loss": 0.3352,
      "step": 14840
    },
    {
      "epoch": 44.59,
      "grad_norm": 12.177059173583984,
      "learning_rate": 8.513513513513514e-06,
      "loss": 0.3652,
      "step": 14850
    },
    {
      "epoch": 44.62,
      "grad_norm": 12.623312950134277,
      "learning_rate": 8.512512512512512e-06,
      "loss": 0.3389,
      "step": 14860
    },
    {
      "epoch": 44.65,
      "grad_norm": 8.402901649475098,
      "learning_rate": 8.511511511511513e-06,
      "loss": 0.3352,
      "step": 14870
    },
    {
      "epoch": 44.68,
      "grad_norm": 13.718243598937988,
      "learning_rate": 8.510510510510511e-06,
      "loss": 0.3787,
      "step": 14880
    },
    {
      "epoch": 44.71,
      "grad_norm": 7.328113079071045,
      "learning_rate": 8.50950950950951e-06,
      "loss": 0.3507,
      "step": 14890
    },
    {
      "epoch": 44.74,
      "grad_norm": 11.263354301452637,
      "learning_rate": 8.50850850850851e-06,
      "loss": 0.3918,
      "step": 14900
    },
    {
      "epoch": 44.77,
      "grad_norm": 7.800790309906006,
      "learning_rate": 8.507507507507509e-06,
      "loss": 0.3882,
      "step": 14910
    },
    {
      "epoch": 44.8,
      "grad_norm": 63.04912185668945,
      "learning_rate": 8.506506506506507e-06,
      "loss": 0.3717,
      "step": 14920
    },
    {
      "epoch": 44.83,
      "grad_norm": 10.114217758178711,
      "learning_rate": 8.505505505505506e-06,
      "loss": 0.344,
      "step": 14930
    },
    {
      "epoch": 44.86,
      "grad_norm": 9.859519004821777,
      "learning_rate": 8.504504504504505e-06,
      "loss": 0.3775,
      "step": 14940
    },
    {
      "epoch": 44.89,
      "grad_norm": 8.783178329467773,
      "learning_rate": 8.503503503503505e-06,
      "loss": 0.3368,
      "step": 14950
    },
    {
      "epoch": 44.92,
      "grad_norm": 6.376849174499512,
      "learning_rate": 8.502502502502503e-06,
      "loss": 0.3648,
      "step": 14960
    },
    {
      "epoch": 44.95,
      "grad_norm": 49.76519012451172,
      "learning_rate": 8.501501501501502e-06,
      "loss": 0.4463,
      "step": 14970
    },
    {
      "epoch": 44.98,
      "grad_norm": 10.114041328430176,
      "learning_rate": 8.5005005005005e-06,
      "loss": 0.3798,
      "step": 14980
    },
    {
      "epoch": 45.0,
      "eval_accuracy": 0.9272,
      "eval_loss": 0.2695719301700592,
      "eval_runtime": 30.6158,
      "eval_samples_per_second": 326.629,
      "eval_steps_per_second": 1.307,
      "step": 14985
    },
    {
      "epoch": 45.02,
      "grad_norm": 19.853479385375977,
      "learning_rate": 8.499499499499501e-06,
      "loss": 0.5018,
      "step": 14990
    },
    {
      "epoch": 45.05,
      "grad_norm": 6.986163139343262,
      "learning_rate": 8.4984984984985e-06,
      "loss": 0.4074,
      "step": 15000
    },
    {
      "epoch": 45.08,
      "grad_norm": 6.741523742675781,
      "learning_rate": 8.497497497497498e-06,
      "loss": 0.3343,
      "step": 15010
    },
    {
      "epoch": 45.11,
      "grad_norm": 9.414925575256348,
      "learning_rate": 8.496496496496497e-06,
      "loss": 0.3852,
      "step": 15020
    },
    {
      "epoch": 45.14,
      "grad_norm": 11.96799373626709,
      "learning_rate": 8.495495495495495e-06,
      "loss": 0.3585,
      "step": 15030
    },
    {
      "epoch": 45.17,
      "grad_norm": 12.851645469665527,
      "learning_rate": 8.494494494494496e-06,
      "loss": 0.3555,
      "step": 15040
    },
    {
      "epoch": 45.2,
      "grad_norm": 7.806941032409668,
      "learning_rate": 8.493493493493494e-06,
      "loss": 0.3825,
      "step": 15050
    },
    {
      "epoch": 45.23,
      "grad_norm": 15.623090744018555,
      "learning_rate": 8.492492492492493e-06,
      "loss": 0.3384,
      "step": 15060
    },
    {
      "epoch": 45.26,
      "grad_norm": 7.867834091186523,
      "learning_rate": 8.491491491491493e-06,
      "loss": 0.3397,
      "step": 15070
    },
    {
      "epoch": 45.29,
      "grad_norm": 8.234654426574707,
      "learning_rate": 8.490490490490492e-06,
      "loss": 0.389,
      "step": 15080
    },
    {
      "epoch": 45.32,
      "grad_norm": 7.707438945770264,
      "learning_rate": 8.48948948948949e-06,
      "loss": 0.37,
      "step": 15090
    },
    {
      "epoch": 45.35,
      "grad_norm": 10.944669723510742,
      "learning_rate": 8.488488488488489e-06,
      "loss": 0.3094,
      "step": 15100
    },
    {
      "epoch": 45.38,
      "grad_norm": 9.753617286682129,
      "learning_rate": 8.487487487487487e-06,
      "loss": 0.4437,
      "step": 15110
    },
    {
      "epoch": 45.41,
      "grad_norm": 21.45918083190918,
      "learning_rate": 8.486486486486488e-06,
      "loss": 0.3658,
      "step": 15120
    },
    {
      "epoch": 45.44,
      "grad_norm": 8.101509094238281,
      "learning_rate": 8.485485485485486e-06,
      "loss": 0.3215,
      "step": 15130
    },
    {
      "epoch": 45.47,
      "grad_norm": 9.842667579650879,
      "learning_rate": 8.484484484484485e-06,
      "loss": 0.3651,
      "step": 15140
    },
    {
      "epoch": 45.5,
      "grad_norm": 24.71480941772461,
      "learning_rate": 8.483483483483485e-06,
      "loss": 0.3862,
      "step": 15150
    },
    {
      "epoch": 45.53,
      "grad_norm": 19.40270233154297,
      "learning_rate": 8.482482482482484e-06,
      "loss": 0.3531,
      "step": 15160
    },
    {
      "epoch": 45.56,
      "grad_norm": 8.077888488769531,
      "learning_rate": 8.481481481481482e-06,
      "loss": 0.3795,
      "step": 15170
    },
    {
      "epoch": 45.59,
      "grad_norm": 10.256925582885742,
      "learning_rate": 8.480480480480481e-06,
      "loss": 0.3319,
      "step": 15180
    },
    {
      "epoch": 45.62,
      "grad_norm": 11.087464332580566,
      "learning_rate": 8.47947947947948e-06,
      "loss": 0.3998,
      "step": 15190
    },
    {
      "epoch": 45.65,
      "grad_norm": 7.9510297775268555,
      "learning_rate": 8.47847847847848e-06,
      "loss": 0.3297,
      "step": 15200
    },
    {
      "epoch": 45.68,
      "grad_norm": 11.740805625915527,
      "learning_rate": 8.477477477477478e-06,
      "loss": 0.3829,
      "step": 15210
    },
    {
      "epoch": 45.71,
      "grad_norm": 7.823885440826416,
      "learning_rate": 8.476476476476477e-06,
      "loss": 0.2877,
      "step": 15220
    },
    {
      "epoch": 45.74,
      "grad_norm": 9.014694213867188,
      "learning_rate": 8.475475475475476e-06,
      "loss": 0.3428,
      "step": 15230
    },
    {
      "epoch": 45.77,
      "grad_norm": 5.919668197631836,
      "learning_rate": 8.474474474474476e-06,
      "loss": 0.3446,
      "step": 15240
    },
    {
      "epoch": 45.8,
      "grad_norm": 12.919995307922363,
      "learning_rate": 8.473473473473475e-06,
      "loss": 0.3371,
      "step": 15250
    },
    {
      "epoch": 45.83,
      "grad_norm": 16.191570281982422,
      "learning_rate": 8.472472472472473e-06,
      "loss": 0.3459,
      "step": 15260
    },
    {
      "epoch": 45.86,
      "grad_norm": 17.091819763183594,
      "learning_rate": 8.471471471471472e-06,
      "loss": 0.3441,
      "step": 15270
    },
    {
      "epoch": 45.89,
      "grad_norm": 8.407232284545898,
      "learning_rate": 8.47047047047047e-06,
      "loss": 0.3944,
      "step": 15280
    },
    {
      "epoch": 45.92,
      "grad_norm": 9.913002967834473,
      "learning_rate": 8.46946946946947e-06,
      "loss": 0.3806,
      "step": 15290
    },
    {
      "epoch": 45.95,
      "grad_norm": 7.731622695922852,
      "learning_rate": 8.46846846846847e-06,
      "loss": 0.335,
      "step": 15300
    },
    {
      "epoch": 45.98,
      "grad_norm": 12.056951522827148,
      "learning_rate": 8.467467467467468e-06,
      "loss": 0.3352,
      "step": 15310
    },
    {
      "epoch": 46.0,
      "eval_accuracy": 0.9264,
      "eval_loss": 0.27261993288993835,
      "eval_runtime": 30.6223,
      "eval_samples_per_second": 326.56,
      "eval_steps_per_second": 1.306,
      "step": 15318
    },
    {
      "epoch": 46.01,
      "grad_norm": 9.661555290222168,
      "learning_rate": 8.466466466466468e-06,
      "loss": 0.3206,
      "step": 15320
    },
    {
      "epoch": 46.04,
      "grad_norm": 10.553648948669434,
      "learning_rate": 8.465465465465467e-06,
      "loss": 0.3464,
      "step": 15330
    },
    {
      "epoch": 46.07,
      "grad_norm": 7.8503570556640625,
      "learning_rate": 8.464464464464465e-06,
      "loss": 0.3584,
      "step": 15340
    },
    {
      "epoch": 46.1,
      "grad_norm": 9.41447639465332,
      "learning_rate": 8.463463463463464e-06,
      "loss": 0.3242,
      "step": 15350
    },
    {
      "epoch": 46.13,
      "grad_norm": 7.716496467590332,
      "learning_rate": 8.462462462462462e-06,
      "loss": 0.3594,
      "step": 15360
    },
    {
      "epoch": 46.16,
      "grad_norm": 11.896690368652344,
      "learning_rate": 8.461461461461463e-06,
      "loss": 0.3598,
      "step": 15370
    },
    {
      "epoch": 46.19,
      "grad_norm": 18.825668334960938,
      "learning_rate": 8.460460460460461e-06,
      "loss": 0.365,
      "step": 15380
    },
    {
      "epoch": 46.22,
      "grad_norm": 8.321016311645508,
      "learning_rate": 8.45945945945946e-06,
      "loss": 0.3294,
      "step": 15390
    },
    {
      "epoch": 46.25,
      "grad_norm": 12.811546325683594,
      "learning_rate": 8.45845845845846e-06,
      "loss": 0.3828,
      "step": 15400
    },
    {
      "epoch": 46.28,
      "grad_norm": 10.27764892578125,
      "learning_rate": 8.457457457457459e-06,
      "loss": 0.3902,
      "step": 15410
    },
    {
      "epoch": 46.31,
      "grad_norm": 11.06362247467041,
      "learning_rate": 8.456456456456457e-06,
      "loss": 0.371,
      "step": 15420
    },
    {
      "epoch": 46.34,
      "grad_norm": 15.446944236755371,
      "learning_rate": 8.455455455455456e-06,
      "loss": 0.3941,
      "step": 15430
    },
    {
      "epoch": 46.37,
      "grad_norm": 14.598814010620117,
      "learning_rate": 8.454454454454455e-06,
      "loss": 0.3562,
      "step": 15440
    },
    {
      "epoch": 46.4,
      "grad_norm": 10.383121490478516,
      "learning_rate": 8.453453453453453e-06,
      "loss": 0.3684,
      "step": 15450
    },
    {
      "epoch": 46.43,
      "grad_norm": 8.936776161193848,
      "learning_rate": 8.452452452452454e-06,
      "loss": 0.3569,
      "step": 15460
    },
    {
      "epoch": 46.46,
      "grad_norm": 16.903728485107422,
      "learning_rate": 8.451451451451452e-06,
      "loss": 0.37,
      "step": 15470
    },
    {
      "epoch": 46.49,
      "grad_norm": 11.004768371582031,
      "learning_rate": 8.45045045045045e-06,
      "loss": 0.3663,
      "step": 15480
    },
    {
      "epoch": 46.52,
      "grad_norm": 9.848750114440918,
      "learning_rate": 8.449449449449451e-06,
      "loss": 0.342,
      "step": 15490
    },
    {
      "epoch": 46.55,
      "grad_norm": 9.40990924835205,
      "learning_rate": 8.44844844844845e-06,
      "loss": 0.3648,
      "step": 15500
    },
    {
      "epoch": 46.58,
      "grad_norm": 7.498935222625732,
      "learning_rate": 8.447447447447448e-06,
      "loss": 0.342,
      "step": 15510
    },
    {
      "epoch": 46.61,
      "grad_norm": 12.804777145385742,
      "learning_rate": 8.446446446446447e-06,
      "loss": 0.343,
      "step": 15520
    },
    {
      "epoch": 46.64,
      "grad_norm": 11.312024116516113,
      "learning_rate": 8.445445445445445e-06,
      "loss": 0.4227,
      "step": 15530
    },
    {
      "epoch": 46.67,
      "grad_norm": 29.805234909057617,
      "learning_rate": 8.444444444444446e-06,
      "loss": 0.3775,
      "step": 15540
    },
    {
      "epoch": 46.7,
      "grad_norm": 7.685587406158447,
      "learning_rate": 8.443443443443444e-06,
      "loss": 0.3919,
      "step": 15550
    },
    {
      "epoch": 46.73,
      "grad_norm": 14.661360740661621,
      "learning_rate": 8.442442442442443e-06,
      "loss": 0.3344,
      "step": 15560
    },
    {
      "epoch": 46.76,
      "grad_norm": 13.445524215698242,
      "learning_rate": 8.441441441441443e-06,
      "loss": 0.4226,
      "step": 15570
    },
    {
      "epoch": 46.79,
      "grad_norm": 10.72363567352295,
      "learning_rate": 8.440440440440442e-06,
      "loss": 0.3481,
      "step": 15580
    },
    {
      "epoch": 46.82,
      "grad_norm": 15.855146408081055,
      "learning_rate": 8.43943943943944e-06,
      "loss": 0.3663,
      "step": 15590
    },
    {
      "epoch": 46.85,
      "grad_norm": 8.282866477966309,
      "learning_rate": 8.438438438438439e-06,
      "loss": 0.4499,
      "step": 15600
    },
    {
      "epoch": 46.88,
      "grad_norm": 14.692459106445312,
      "learning_rate": 8.437437437437438e-06,
      "loss": 0.4026,
      "step": 15610
    },
    {
      "epoch": 46.91,
      "grad_norm": 18.538326263427734,
      "learning_rate": 8.436436436436438e-06,
      "loss": 0.3366,
      "step": 15620
    },
    {
      "epoch": 46.94,
      "grad_norm": 12.595952987670898,
      "learning_rate": 8.435435435435436e-06,
      "loss": 0.3636,
      "step": 15630
    },
    {
      "epoch": 46.97,
      "grad_norm": 10.842432022094727,
      "learning_rate": 8.434434434434435e-06,
      "loss": 0.3303,
      "step": 15640
    },
    {
      "epoch": 47.0,
      "grad_norm": 11.322273254394531,
      "learning_rate": 8.433433433433435e-06,
      "loss": 0.3703,
      "step": 15650
    },
    {
      "epoch": 47.0,
      "eval_accuracy": 0.926,
      "eval_loss": 0.2780829071998596,
      "eval_runtime": 30.6056,
      "eval_samples_per_second": 326.737,
      "eval_steps_per_second": 1.307,
      "step": 15651
    },
    {
      "epoch": 47.03,
      "grad_norm": 20.886505126953125,
      "learning_rate": 8.432432432432434e-06,
      "loss": 0.4436,
      "step": 15660
    },
    {
      "epoch": 47.06,
      "grad_norm": 8.133129119873047,
      "learning_rate": 8.43143143143143e-06,
      "loss": 0.3805,
      "step": 15670
    },
    {
      "epoch": 47.09,
      "grad_norm": 8.90296459197998,
      "learning_rate": 8.430430430430431e-06,
      "loss": 0.3558,
      "step": 15680
    },
    {
      "epoch": 47.12,
      "grad_norm": 4.936026096343994,
      "learning_rate": 8.42942942942943e-06,
      "loss": 0.3681,
      "step": 15690
    },
    {
      "epoch": 47.15,
      "grad_norm": 11.208409309387207,
      "learning_rate": 8.428428428428428e-06,
      "loss": 0.4096,
      "step": 15700
    },
    {
      "epoch": 47.18,
      "grad_norm": 11.841787338256836,
      "learning_rate": 8.427427427427429e-06,
      "loss": 0.3508,
      "step": 15710
    },
    {
      "epoch": 47.21,
      "grad_norm": 22.14638328552246,
      "learning_rate": 8.426426426426427e-06,
      "loss": 0.4108,
      "step": 15720
    },
    {
      "epoch": 47.24,
      "grad_norm": 9.453351020812988,
      "learning_rate": 8.425425425425426e-06,
      "loss": 0.4079,
      "step": 15730
    },
    {
      "epoch": 47.27,
      "grad_norm": 11.26319694519043,
      "learning_rate": 8.424424424424426e-06,
      "loss": 0.3982,
      "step": 15740
    },
    {
      "epoch": 47.3,
      "grad_norm": 115.33975982666016,
      "learning_rate": 8.423423423423423e-06,
      "loss": 0.358,
      "step": 15750
    },
    {
      "epoch": 47.33,
      "grad_norm": 9.413263320922852,
      "learning_rate": 8.422422422422423e-06,
      "loss": 0.3863,
      "step": 15760
    },
    {
      "epoch": 47.36,
      "grad_norm": 12.096962928771973,
      "learning_rate": 8.421421421421422e-06,
      "loss": 0.4338,
      "step": 15770
    },
    {
      "epoch": 47.39,
      "grad_norm": 8.80749797821045,
      "learning_rate": 8.42042042042042e-06,
      "loss": 0.3751,
      "step": 15780
    },
    {
      "epoch": 47.42,
      "grad_norm": 18.937952041625977,
      "learning_rate": 8.41941941941942e-06,
      "loss": 0.387,
      "step": 15790
    },
    {
      "epoch": 47.45,
      "grad_norm": 7.813028812408447,
      "learning_rate": 8.41841841841842e-06,
      "loss": 0.3625,
      "step": 15800
    },
    {
      "epoch": 47.48,
      "grad_norm": 8.949049949645996,
      "learning_rate": 8.417417417417418e-06,
      "loss": 0.4068,
      "step": 15810
    },
    {
      "epoch": 47.51,
      "grad_norm": 51.165870666503906,
      "learning_rate": 8.416416416416418e-06,
      "loss": 0.3741,
      "step": 15820
    },
    {
      "epoch": 47.54,
      "grad_norm": 10.054884910583496,
      "learning_rate": 8.415415415415417e-06,
      "loss": 0.4133,
      "step": 15830
    },
    {
      "epoch": 47.57,
      "grad_norm": 6.706489562988281,
      "learning_rate": 8.414414414414415e-06,
      "loss": 0.3266,
      "step": 15840
    },
    {
      "epoch": 47.6,
      "grad_norm": 10.084938049316406,
      "learning_rate": 8.413413413413414e-06,
      "loss": 0.343,
      "step": 15850
    },
    {
      "epoch": 47.63,
      "grad_norm": 13.865789413452148,
      "learning_rate": 8.412412412412413e-06,
      "loss": 0.3646,
      "step": 15860
    },
    {
      "epoch": 47.66,
      "grad_norm": 6.685812473297119,
      "learning_rate": 8.411411411411413e-06,
      "loss": 0.3543,
      "step": 15870
    },
    {
      "epoch": 47.69,
      "grad_norm": 10.750744819641113,
      "learning_rate": 8.410410410410411e-06,
      "loss": 0.4058,
      "step": 15880
    },
    {
      "epoch": 47.72,
      "grad_norm": 5.851251602172852,
      "learning_rate": 8.40940940940941e-06,
      "loss": 0.3631,
      "step": 15890
    },
    {
      "epoch": 47.75,
      "grad_norm": 9.299747467041016,
      "learning_rate": 8.408408408408409e-06,
      "loss": 0.3309,
      "step": 15900
    },
    {
      "epoch": 47.78,
      "grad_norm": 8.037483215332031,
      "learning_rate": 8.407407407407409e-06,
      "loss": 0.3652,
      "step": 15910
    },
    {
      "epoch": 47.81,
      "grad_norm": 11.953837394714355,
      "learning_rate": 8.406406406406406e-06,
      "loss": 0.3769,
      "step": 15920
    },
    {
      "epoch": 47.84,
      "grad_norm": 10.373597145080566,
      "learning_rate": 8.405405405405406e-06,
      "loss": 0.3703,
      "step": 15930
    },
    {
      "epoch": 47.87,
      "grad_norm": 7.543815612792969,
      "learning_rate": 8.404404404404405e-06,
      "loss": 0.3121,
      "step": 15940
    },
    {
      "epoch": 47.9,
      "grad_norm": 13.849148750305176,
      "learning_rate": 8.403403403403403e-06,
      "loss": 0.3548,
      "step": 15950
    },
    {
      "epoch": 47.93,
      "grad_norm": 9.379244804382324,
      "learning_rate": 8.402402402402404e-06,
      "loss": 0.3856,
      "step": 15960
    },
    {
      "epoch": 47.96,
      "grad_norm": 13.082120895385742,
      "learning_rate": 8.401401401401402e-06,
      "loss": 0.3339,
      "step": 15970
    },
    {
      "epoch": 47.99,
      "grad_norm": 9.398272514343262,
      "learning_rate": 8.4004004004004e-06,
      "loss": 0.3596,
      "step": 15980
    },
    {
      "epoch": 48.0,
      "eval_accuracy": 0.9274,
      "eval_loss": 0.27473461627960205,
      "eval_runtime": 30.2838,
      "eval_samples_per_second": 330.21,
      "eval_steps_per_second": 1.321,
      "step": 15984
    },
    {
      "epoch": 48.02,
      "grad_norm": 9.764413833618164,
      "learning_rate": 8.399399399399401e-06,
      "loss": 0.2918,
      "step": 15990
    },
    {
      "epoch": 48.05,
      "grad_norm": 14.378108978271484,
      "learning_rate": 8.398398398398398e-06,
      "loss": 0.4361,
      "step": 16000
    },
    {
      "epoch": 48.08,
      "grad_norm": 10.890961647033691,
      "learning_rate": 8.397397397397398e-06,
      "loss": 0.3407,
      "step": 16010
    },
    {
      "epoch": 48.11,
      "grad_norm": 17.184772491455078,
      "learning_rate": 8.396396396396397e-06,
      "loss": 0.3466,
      "step": 16020
    },
    {
      "epoch": 48.14,
      "grad_norm": 6.977208137512207,
      "learning_rate": 8.395395395395395e-06,
      "loss": 0.3796,
      "step": 16030
    },
    {
      "epoch": 48.17,
      "grad_norm": 10.32593822479248,
      "learning_rate": 8.394394394394396e-06,
      "loss": 0.4624,
      "step": 16040
    },
    {
      "epoch": 48.2,
      "grad_norm": 33.85615539550781,
      "learning_rate": 8.393393393393394e-06,
      "loss": 0.3693,
      "step": 16050
    },
    {
      "epoch": 48.23,
      "grad_norm": 8.661319732666016,
      "learning_rate": 8.392392392392393e-06,
      "loss": 0.3674,
      "step": 16060
    },
    {
      "epoch": 48.26,
      "grad_norm": 10.651653289794922,
      "learning_rate": 8.391391391391393e-06,
      "loss": 0.362,
      "step": 16070
    },
    {
      "epoch": 48.29,
      "grad_norm": 10.288437843322754,
      "learning_rate": 8.390390390390392e-06,
      "loss": 0.3861,
      "step": 16080
    },
    {
      "epoch": 48.32,
      "grad_norm": 12.986308097839355,
      "learning_rate": 8.38938938938939e-06,
      "loss": 0.3748,
      "step": 16090
    },
    {
      "epoch": 48.35,
      "grad_norm": 15.581962585449219,
      "learning_rate": 8.388388388388389e-06,
      "loss": 0.3591,
      "step": 16100
    },
    {
      "epoch": 48.38,
      "grad_norm": 8.093202590942383,
      "learning_rate": 8.387387387387388e-06,
      "loss": 0.3256,
      "step": 16110
    },
    {
      "epoch": 48.41,
      "grad_norm": 12.434480667114258,
      "learning_rate": 8.386386386386386e-06,
      "loss": 0.3893,
      "step": 16120
    },
    {
      "epoch": 48.44,
      "grad_norm": 17.403512954711914,
      "learning_rate": 8.385385385385386e-06,
      "loss": 0.3575,
      "step": 16130
    },
    {
      "epoch": 48.47,
      "grad_norm": 9.287558555603027,
      "learning_rate": 8.384384384384385e-06,
      "loss": 0.3339,
      "step": 16140
    },
    {
      "epoch": 48.5,
      "grad_norm": 15.082598686218262,
      "learning_rate": 8.383383383383384e-06,
      "loss": 0.2694,
      "step": 16150
    },
    {
      "epoch": 48.53,
      "grad_norm": 9.478750228881836,
      "learning_rate": 8.382382382382384e-06,
      "loss": 0.3963,
      "step": 16160
    },
    {
      "epoch": 48.56,
      "grad_norm": 12.289950370788574,
      "learning_rate": 8.381381381381381e-06,
      "loss": 0.3847,
      "step": 16170
    },
    {
      "epoch": 48.59,
      "grad_norm": 7.738155841827393,
      "learning_rate": 8.380380380380381e-06,
      "loss": 0.4068,
      "step": 16180
    },
    {
      "epoch": 48.62,
      "grad_norm": 9.716984748840332,
      "learning_rate": 8.37937937937938e-06,
      "loss": 0.3333,
      "step": 16190
    },
    {
      "epoch": 48.65,
      "grad_norm": 8.956748962402344,
      "learning_rate": 8.378378378378378e-06,
      "loss": 0.3265,
      "step": 16200
    },
    {
      "epoch": 48.68,
      "grad_norm": 17.094547271728516,
      "learning_rate": 8.377377377377379e-06,
      "loss": 0.3573,
      "step": 16210
    },
    {
      "epoch": 48.71,
      "grad_norm": 9.4926118850708,
      "learning_rate": 8.376376376376377e-06,
      "loss": 0.3555,
      "step": 16220
    },
    {
      "epoch": 48.74,
      "grad_norm": 11.15902328491211,
      "learning_rate": 8.375375375375376e-06,
      "loss": 0.3573,
      "step": 16230
    },
    {
      "epoch": 48.77,
      "grad_norm": 14.139511108398438,
      "learning_rate": 8.374374374374376e-06,
      "loss": 0.3624,
      "step": 16240
    },
    {
      "epoch": 48.8,
      "grad_norm": 10.704151153564453,
      "learning_rate": 8.373373373373373e-06,
      "loss": 0.3831,
      "step": 16250
    },
    {
      "epoch": 48.83,
      "grad_norm": 16.601850509643555,
      "learning_rate": 8.372372372372373e-06,
      "loss": 0.3301,
      "step": 16260
    },
    {
      "epoch": 48.86,
      "grad_norm": 7.870049476623535,
      "learning_rate": 8.371371371371372e-06,
      "loss": 0.4061,
      "step": 16270
    },
    {
      "epoch": 48.89,
      "grad_norm": 14.30327033996582,
      "learning_rate": 8.37037037037037e-06,
      "loss": 0.3136,
      "step": 16280
    },
    {
      "epoch": 48.92,
      "grad_norm": 15.52718734741211,
      "learning_rate": 8.36936936936937e-06,
      "loss": 0.4083,
      "step": 16290
    },
    {
      "epoch": 48.95,
      "grad_norm": 14.457443237304688,
      "learning_rate": 8.36836836836837e-06,
      "loss": 0.3547,
      "step": 16300
    },
    {
      "epoch": 48.98,
      "grad_norm": 9.259882926940918,
      "learning_rate": 8.367367367367368e-06,
      "loss": 0.385,
      "step": 16310
    },
    {
      "epoch": 49.0,
      "eval_accuracy": 0.9257,
      "eval_loss": 0.2769479751586914,
      "eval_runtime": 30.4492,
      "eval_samples_per_second": 328.416,
      "eval_steps_per_second": 1.314,
      "step": 16317
    },
    {
      "epoch": 49.01,
      "grad_norm": 13.306678771972656,
      "learning_rate": 8.366366366366368e-06,
      "loss": 0.4705,
      "step": 16320
    },
    {
      "epoch": 49.04,
      "grad_norm": 11.276318550109863,
      "learning_rate": 8.365365365365367e-06,
      "loss": 0.3878,
      "step": 16330
    },
    {
      "epoch": 49.07,
      "grad_norm": 7.781841278076172,
      "learning_rate": 8.364364364364365e-06,
      "loss": 0.34,
      "step": 16340
    },
    {
      "epoch": 49.1,
      "grad_norm": 10.791153907775879,
      "learning_rate": 8.363363363363364e-06,
      "loss": 0.3522,
      "step": 16350
    },
    {
      "epoch": 49.13,
      "grad_norm": 10.386406898498535,
      "learning_rate": 8.362362362362363e-06,
      "loss": 0.3984,
      "step": 16360
    },
    {
      "epoch": 49.16,
      "grad_norm": 9.338409423828125,
      "learning_rate": 8.361361361361361e-06,
      "loss": 0.4048,
      "step": 16370
    },
    {
      "epoch": 49.19,
      "grad_norm": 13.448527336120605,
      "learning_rate": 8.360360360360362e-06,
      "loss": 0.3255,
      "step": 16380
    },
    {
      "epoch": 49.22,
      "grad_norm": 9.818233489990234,
      "learning_rate": 8.35935935935936e-06,
      "loss": 0.3709,
      "step": 16390
    },
    {
      "epoch": 49.25,
      "grad_norm": 14.008875846862793,
      "learning_rate": 8.358358358358359e-06,
      "loss": 0.353,
      "step": 16400
    },
    {
      "epoch": 49.28,
      "grad_norm": 10.472732543945312,
      "learning_rate": 8.357357357357359e-06,
      "loss": 0.3383,
      "step": 16410
    },
    {
      "epoch": 49.31,
      "grad_norm": 27.522315979003906,
      "learning_rate": 8.356356356356356e-06,
      "loss": 0.3465,
      "step": 16420
    },
    {
      "epoch": 49.34,
      "grad_norm": 8.551373481750488,
      "learning_rate": 8.355355355355356e-06,
      "loss": 0.3412,
      "step": 16430
    },
    {
      "epoch": 49.37,
      "grad_norm": 10.435650825500488,
      "learning_rate": 8.354354354354355e-06,
      "loss": 0.3753,
      "step": 16440
    },
    {
      "epoch": 49.4,
      "grad_norm": 6.001415729522705,
      "learning_rate": 8.353353353353353e-06,
      "loss": 0.382,
      "step": 16450
    },
    {
      "epoch": 49.43,
      "grad_norm": 12.369171142578125,
      "learning_rate": 8.352352352352354e-06,
      "loss": 0.3839,
      "step": 16460
    },
    {
      "epoch": 49.46,
      "grad_norm": 10.009113311767578,
      "learning_rate": 8.351351351351352e-06,
      "loss": 0.4223,
      "step": 16470
    },
    {
      "epoch": 49.49,
      "grad_norm": 8.928821563720703,
      "learning_rate": 8.35035035035035e-06,
      "loss": 0.3703,
      "step": 16480
    },
    {
      "epoch": 49.52,
      "grad_norm": 8.522218704223633,
      "learning_rate": 8.349349349349351e-06,
      "loss": 0.3766,
      "step": 16490
    },
    {
      "epoch": 49.55,
      "grad_norm": 10.840795516967773,
      "learning_rate": 8.348348348348348e-06,
      "loss": 0.3367,
      "step": 16500
    },
    {
      "epoch": 49.58,
      "grad_norm": 10.618741989135742,
      "learning_rate": 8.347347347347348e-06,
      "loss": 0.3304,
      "step": 16510
    },
    {
      "epoch": 49.61,
      "grad_norm": 12.1637601852417,
      "learning_rate": 8.346346346346347e-06,
      "loss": 0.4155,
      "step": 16520
    },
    {
      "epoch": 49.64,
      "grad_norm": 11.082706451416016,
      "learning_rate": 8.345345345345346e-06,
      "loss": 0.3395,
      "step": 16530
    },
    {
      "epoch": 49.67,
      "grad_norm": 11.907086372375488,
      "learning_rate": 8.344344344344346e-06,
      "loss": 0.3612,
      "step": 16540
    },
    {
      "epoch": 49.7,
      "grad_norm": 9.972537994384766,
      "learning_rate": 8.343343343343344e-06,
      "loss": 0.3894,
      "step": 16550
    },
    {
      "epoch": 49.73,
      "grad_norm": 8.022310256958008,
      "learning_rate": 8.342342342342343e-06,
      "loss": 0.3955,
      "step": 16560
    },
    {
      "epoch": 49.76,
      "grad_norm": 8.814326286315918,
      "learning_rate": 8.341341341341343e-06,
      "loss": 0.3759,
      "step": 16570
    },
    {
      "epoch": 49.79,
      "grad_norm": 9.324073791503906,
      "learning_rate": 8.340340340340342e-06,
      "loss": 0.3793,
      "step": 16580
    },
    {
      "epoch": 49.82,
      "grad_norm": 26.732759475708008,
      "learning_rate": 8.339339339339339e-06,
      "loss": 0.3809,
      "step": 16590
    },
    {
      "epoch": 49.85,
      "grad_norm": 11.249847412109375,
      "learning_rate": 8.338338338338339e-06,
      "loss": 0.3438,
      "step": 16600
    },
    {
      "epoch": 49.88,
      "grad_norm": 27.152463912963867,
      "learning_rate": 8.337337337337338e-06,
      "loss": 0.3384,
      "step": 16610
    },
    {
      "epoch": 49.91,
      "grad_norm": 7.959740161895752,
      "learning_rate": 8.336336336336336e-06,
      "loss": 0.3638,
      "step": 16620
    },
    {
      "epoch": 49.94,
      "grad_norm": 14.411429405212402,
      "learning_rate": 8.335335335335337e-06,
      "loss": 0.3571,
      "step": 16630
    },
    {
      "epoch": 49.97,
      "grad_norm": 12.351994514465332,
      "learning_rate": 8.334334334334335e-06,
      "loss": 0.3277,
      "step": 16640
    },
    {
      "epoch": 50.0,
      "grad_norm": 318.62347412109375,
      "learning_rate": 8.333333333333334e-06,
      "loss": 0.3716,
      "step": 16650
    },
    {
      "epoch": 50.0,
      "eval_accuracy": 0.9268,
      "eval_loss": 0.2768508195877075,
      "eval_runtime": 30.1187,
      "eval_samples_per_second": 332.02,
      "eval_steps_per_second": 1.328,
      "step": 16650
    },
    {
      "epoch": 50.03,
      "grad_norm": 26.71160888671875,
      "learning_rate": 8.332332332332334e-06,
      "loss": 0.3339,
      "step": 16660
    },
    {
      "epoch": 50.06,
      "grad_norm": 11.198226928710938,
      "learning_rate": 8.331331331331331e-06,
      "loss": 0.415,
      "step": 16670
    },
    {
      "epoch": 50.09,
      "grad_norm": 11.483070373535156,
      "learning_rate": 8.330330330330331e-06,
      "loss": 0.363,
      "step": 16680
    },
    {
      "epoch": 50.12,
      "grad_norm": 5.863163948059082,
      "learning_rate": 8.32932932932933e-06,
      "loss": 0.346,
      "step": 16690
    },
    {
      "epoch": 50.15,
      "grad_norm": 8.948847770690918,
      "learning_rate": 8.328328328328328e-06,
      "loss": 0.3911,
      "step": 16700
    },
    {
      "epoch": 50.18,
      "grad_norm": 7.947269439697266,
      "learning_rate": 8.327327327327329e-06,
      "loss": 0.3524,
      "step": 16710
    },
    {
      "epoch": 50.21,
      "grad_norm": 7.5357208251953125,
      "learning_rate": 8.326326326326327e-06,
      "loss": 0.3718,
      "step": 16720
    },
    {
      "epoch": 50.24,
      "grad_norm": 5.048258304595947,
      "learning_rate": 8.325325325325326e-06,
      "loss": 0.3684,
      "step": 16730
    },
    {
      "epoch": 50.27,
      "grad_norm": 11.179451942443848,
      "learning_rate": 8.324324324324326e-06,
      "loss": 0.3095,
      "step": 16740
    },
    {
      "epoch": 50.3,
      "grad_norm": 10.54816722869873,
      "learning_rate": 8.323323323323323e-06,
      "loss": 0.3405,
      "step": 16750
    },
    {
      "epoch": 50.33,
      "grad_norm": 11.465095520019531,
      "learning_rate": 8.322322322322323e-06,
      "loss": 0.3746,
      "step": 16760
    },
    {
      "epoch": 50.36,
      "grad_norm": 8.204338073730469,
      "learning_rate": 8.321321321321322e-06,
      "loss": 0.4069,
      "step": 16770
    },
    {
      "epoch": 50.39,
      "grad_norm": 8.166742324829102,
      "learning_rate": 8.32032032032032e-06,
      "loss": 0.3768,
      "step": 16780
    },
    {
      "epoch": 50.42,
      "grad_norm": 7.175552845001221,
      "learning_rate": 8.31931931931932e-06,
      "loss": 0.3332,
      "step": 16790
    },
    {
      "epoch": 50.45,
      "grad_norm": 11.422401428222656,
      "learning_rate": 8.31831831831832e-06,
      "loss": 0.3532,
      "step": 16800
    },
    {
      "epoch": 50.48,
      "grad_norm": 14.208497047424316,
      "learning_rate": 8.317317317317318e-06,
      "loss": 0.3685,
      "step": 16810
    },
    {
      "epoch": 50.51,
      "grad_norm": 7.130876541137695,
      "learning_rate": 8.316316316316317e-06,
      "loss": 0.3711,
      "step": 16820
    },
    {
      "epoch": 50.54,
      "grad_norm": 10.63101863861084,
      "learning_rate": 8.315315315315317e-06,
      "loss": 0.3388,
      "step": 16830
    },
    {
      "epoch": 50.57,
      "grad_norm": 6.350536823272705,
      "learning_rate": 8.314314314314314e-06,
      "loss": 0.3229,
      "step": 16840
    },
    {
      "epoch": 50.6,
      "grad_norm": 9.816048622131348,
      "learning_rate": 8.313313313313314e-06,
      "loss": 0.3462,
      "step": 16850
    },
    {
      "epoch": 50.63,
      "grad_norm": 8.566447257995605,
      "learning_rate": 8.312312312312313e-06,
      "loss": 0.3681,
      "step": 16860
    },
    {
      "epoch": 50.66,
      "grad_norm": 6.129702568054199,
      "learning_rate": 8.311311311311311e-06,
      "loss": 0.3388,
      "step": 16870
    },
    {
      "epoch": 50.69,
      "grad_norm": 14.884392738342285,
      "learning_rate": 8.310310310310312e-06,
      "loss": 0.3734,
      "step": 16880
    },
    {
      "epoch": 50.72,
      "grad_norm": 9.583518981933594,
      "learning_rate": 8.30930930930931e-06,
      "loss": 0.3548,
      "step": 16890
    },
    {
      "epoch": 50.75,
      "grad_norm": 8.364599227905273,
      "learning_rate": 8.308308308308309e-06,
      "loss": 0.3873,
      "step": 16900
    },
    {
      "epoch": 50.78,
      "grad_norm": 9.4354887008667,
      "learning_rate": 8.307307307307309e-06,
      "loss": 0.3235,
      "step": 16910
    },
    {
      "epoch": 50.81,
      "grad_norm": 17.187345504760742,
      "learning_rate": 8.306306306306306e-06,
      "loss": 0.349,
      "step": 16920
    },
    {
      "epoch": 50.84,
      "grad_norm": 31.20932960510254,
      "learning_rate": 8.305305305305306e-06,
      "loss": 0.3685,
      "step": 16930
    },
    {
      "epoch": 50.87,
      "grad_norm": 11.474005699157715,
      "learning_rate": 8.304304304304305e-06,
      "loss": 0.397,
      "step": 16940
    },
    {
      "epoch": 50.9,
      "grad_norm": 34.92700958251953,
      "learning_rate": 8.303303303303303e-06,
      "loss": 0.3351,
      "step": 16950
    },
    {
      "epoch": 50.93,
      "grad_norm": 11.952458381652832,
      "learning_rate": 8.302302302302304e-06,
      "loss": 0.3716,
      "step": 16960
    },
    {
      "epoch": 50.96,
      "grad_norm": 11.915996551513672,
      "learning_rate": 8.301301301301302e-06,
      "loss": 0.3908,
      "step": 16970
    },
    {
      "epoch": 50.99,
      "grad_norm": 16.051668167114258,
      "learning_rate": 8.300300300300301e-06,
      "loss": 0.3305,
      "step": 16980
    },
    {
      "epoch": 51.0,
      "eval_accuracy": 0.9257,
      "eval_loss": 0.2755736708641052,
      "eval_runtime": 30.2132,
      "eval_samples_per_second": 330.981,
      "eval_steps_per_second": 1.324,
      "step": 16983
    },
    {
      "epoch": 51.02,
      "grad_norm": 7.306516170501709,
      "learning_rate": 8.299299299299301e-06,
      "loss": 0.315,
      "step": 16990
    },
    {
      "epoch": 51.05,
      "grad_norm": 11.473756790161133,
      "learning_rate": 8.298298298298298e-06,
      "loss": 0.3499,
      "step": 17000
    },
    {
      "epoch": 51.08,
      "grad_norm": 8.321928024291992,
      "learning_rate": 8.297297297297298e-06,
      "loss": 0.3128,
      "step": 17010
    },
    {
      "epoch": 51.11,
      "grad_norm": 9.783587455749512,
      "learning_rate": 8.296296296296297e-06,
      "loss": 0.4045,
      "step": 17020
    },
    {
      "epoch": 51.14,
      "grad_norm": 10.593074798583984,
      "learning_rate": 8.295295295295296e-06,
      "loss": 0.3554,
      "step": 17030
    },
    {
      "epoch": 51.17,
      "grad_norm": 13.409724235534668,
      "learning_rate": 8.294294294294294e-06,
      "loss": 0.3644,
      "step": 17040
    },
    {
      "epoch": 51.2,
      "grad_norm": 12.743864059448242,
      "learning_rate": 8.293293293293294e-06,
      "loss": 0.3996,
      "step": 17050
    },
    {
      "epoch": 51.23,
      "grad_norm": 55.76548767089844,
      "learning_rate": 8.292292292292293e-06,
      "loss": 0.362,
      "step": 17060
    },
    {
      "epoch": 51.26,
      "grad_norm": 10.252687454223633,
      "learning_rate": 8.291291291291292e-06,
      "loss": 0.3657,
      "step": 17070
    },
    {
      "epoch": 51.29,
      "grad_norm": 13.473118782043457,
      "learning_rate": 8.29029029029029e-06,
      "loss": 0.3184,
      "step": 17080
    },
    {
      "epoch": 51.32,
      "grad_norm": 11.952330589294434,
      "learning_rate": 8.289289289289289e-06,
      "loss": 0.3643,
      "step": 17090
    },
    {
      "epoch": 51.35,
      "grad_norm": 10.108327865600586,
      "learning_rate": 8.288288288288289e-06,
      "loss": 0.3385,
      "step": 17100
    },
    {
      "epoch": 51.38,
      "grad_norm": 9.861898422241211,
      "learning_rate": 8.287287287287288e-06,
      "loss": 0.3124,
      "step": 17110
    },
    {
      "epoch": 51.41,
      "grad_norm": 7.704039096832275,
      "learning_rate": 8.286286286286286e-06,
      "loss": 0.3254,
      "step": 17120
    },
    {
      "epoch": 51.44,
      "grad_norm": 14.782408714294434,
      "learning_rate": 8.285285285285287e-06,
      "loss": 0.3316,
      "step": 17130
    },
    {
      "epoch": 51.47,
      "grad_norm": 8.255253791809082,
      "learning_rate": 8.284284284284285e-06,
      "loss": 0.314,
      "step": 17140
    },
    {
      "epoch": 51.5,
      "grad_norm": 11.790773391723633,
      "learning_rate": 8.283283283283284e-06,
      "loss": 0.3256,
      "step": 17150
    },
    {
      "epoch": 51.53,
      "grad_norm": 8.404666900634766,
      "learning_rate": 8.282282282282284e-06,
      "loss": 0.369,
      "step": 17160
    },
    {
      "epoch": 51.56,
      "grad_norm": 17.87883186340332,
      "learning_rate": 8.281281281281281e-06,
      "loss": 0.3397,
      "step": 17170
    },
    {
      "epoch": 51.59,
      "grad_norm": 8.228426933288574,
      "learning_rate": 8.280280280280281e-06,
      "loss": 0.2889,
      "step": 17180
    },
    {
      "epoch": 51.62,
      "grad_norm": 9.79439640045166,
      "learning_rate": 8.27927927927928e-06,
      "loss": 0.3633,
      "step": 17190
    },
    {
      "epoch": 51.65,
      "grad_norm": 10.585335731506348,
      "learning_rate": 8.278278278278278e-06,
      "loss": 0.3461,
      "step": 17200
    },
    {
      "epoch": 51.68,
      "grad_norm": 8.645030975341797,
      "learning_rate": 8.277277277277279e-06,
      "loss": 0.3809,
      "step": 17210
    },
    {
      "epoch": 51.71,
      "grad_norm": 10.037964820861816,
      "learning_rate": 8.276276276276277e-06,
      "loss": 0.3458,
      "step": 17220
    },
    {
      "epoch": 51.74,
      "grad_norm": 11.35058879852295,
      "learning_rate": 8.275275275275276e-06,
      "loss": 0.3835,
      "step": 17230
    },
    {
      "epoch": 51.77,
      "grad_norm": 9.419187545776367,
      "learning_rate": 8.274274274274276e-06,
      "loss": 0.3153,
      "step": 17240
    },
    {
      "epoch": 51.8,
      "grad_norm": 9.138304710388184,
      "learning_rate": 8.273273273273273e-06,
      "loss": 0.3794,
      "step": 17250
    },
    {
      "epoch": 51.83,
      "grad_norm": 7.800412654876709,
      "learning_rate": 8.272272272272273e-06,
      "loss": 0.3442,
      "step": 17260
    },
    {
      "epoch": 51.86,
      "grad_norm": 11.79967975616455,
      "learning_rate": 8.271271271271272e-06,
      "loss": 0.3183,
      "step": 17270
    },
    {
      "epoch": 51.89,
      "grad_norm": 9.996403694152832,
      "learning_rate": 8.27027027027027e-06,
      "loss": 0.3466,
      "step": 17280
    },
    {
      "epoch": 51.92,
      "grad_norm": 14.834601402282715,
      "learning_rate": 8.26926926926927e-06,
      "loss": 0.3641,
      "step": 17290
    },
    {
      "epoch": 51.95,
      "grad_norm": 17.422407150268555,
      "learning_rate": 8.26826826826827e-06,
      "loss": 0.3801,
      "step": 17300
    },
    {
      "epoch": 51.98,
      "grad_norm": 9.181379318237305,
      "learning_rate": 8.267267267267268e-06,
      "loss": 0.354,
      "step": 17310
    },
    {
      "epoch": 52.0,
      "eval_accuracy": 0.927,
      "eval_loss": 0.27594617009162903,
      "eval_runtime": 30.3144,
      "eval_samples_per_second": 329.876,
      "eval_steps_per_second": 1.32,
      "step": 17316
    },
    {
      "epoch": 52.01,
      "grad_norm": 8.875324249267578,
      "learning_rate": 8.266266266266267e-06,
      "loss": 0.3863,
      "step": 17320
    },
    {
      "epoch": 52.04,
      "grad_norm": 12.330975532531738,
      "learning_rate": 8.265265265265265e-06,
      "loss": 0.3551,
      "step": 17330
    },
    {
      "epoch": 52.07,
      "grad_norm": 17.596134185791016,
      "learning_rate": 8.264264264264264e-06,
      "loss": 0.3884,
      "step": 17340
    },
    {
      "epoch": 52.1,
      "grad_norm": 11.096054077148438,
      "learning_rate": 8.263263263263264e-06,
      "loss": 0.3479,
      "step": 17350
    },
    {
      "epoch": 52.13,
      "grad_norm": 10.054850578308105,
      "learning_rate": 8.262262262262263e-06,
      "loss": 0.3449,
      "step": 17360
    },
    {
      "epoch": 52.16,
      "grad_norm": 11.562724113464355,
      "learning_rate": 8.261261261261261e-06,
      "loss": 0.3324,
      "step": 17370
    },
    {
      "epoch": 52.19,
      "grad_norm": 12.397189140319824,
      "learning_rate": 8.260260260260262e-06,
      "loss": 0.3614,
      "step": 17380
    },
    {
      "epoch": 52.22,
      "grad_norm": 10.525450706481934,
      "learning_rate": 8.25925925925926e-06,
      "loss": 0.3674,
      "step": 17390
    },
    {
      "epoch": 52.25,
      "grad_norm": 6.3356499671936035,
      "learning_rate": 8.258258258258259e-06,
      "loss": 0.3331,
      "step": 17400
    },
    {
      "epoch": 52.28,
      "grad_norm": 8.932330131530762,
      "learning_rate": 8.257257257257259e-06,
      "loss": 0.3871,
      "step": 17410
    },
    {
      "epoch": 52.31,
      "grad_norm": 24.228811264038086,
      "learning_rate": 8.256256256256256e-06,
      "loss": 0.382,
      "step": 17420
    },
    {
      "epoch": 52.34,
      "grad_norm": 8.634523391723633,
      "learning_rate": 8.255255255255256e-06,
      "loss": 0.3904,
      "step": 17430
    },
    {
      "epoch": 52.37,
      "grad_norm": 5.846917152404785,
      "learning_rate": 8.254254254254255e-06,
      "loss": 0.3456,
      "step": 17440
    },
    {
      "epoch": 52.4,
      "grad_norm": 7.984278678894043,
      "learning_rate": 8.253253253253254e-06,
      "loss": 0.3595,
      "step": 17450
    },
    {
      "epoch": 52.43,
      "grad_norm": 11.339083671569824,
      "learning_rate": 8.252252252252254e-06,
      "loss": 0.3365,
      "step": 17460
    },
    {
      "epoch": 52.46,
      "grad_norm": 6.378913402557373,
      "learning_rate": 8.251251251251252e-06,
      "loss": 0.3325,
      "step": 17470
    },
    {
      "epoch": 52.49,
      "grad_norm": 11.498917579650879,
      "learning_rate": 8.250250250250251e-06,
      "loss": 0.356,
      "step": 17480
    },
    {
      "epoch": 52.52,
      "grad_norm": 9.604192733764648,
      "learning_rate": 8.24924924924925e-06,
      "loss": 0.4113,
      "step": 17490
    },
    {
      "epoch": 52.55,
      "grad_norm": 7.645437717437744,
      "learning_rate": 8.248248248248248e-06,
      "loss": 0.3425,
      "step": 17500
    },
    {
      "epoch": 52.58,
      "grad_norm": 23.647212982177734,
      "learning_rate": 8.247247247247247e-06,
      "loss": 0.3512,
      "step": 17510
    },
    {
      "epoch": 52.61,
      "grad_norm": 7.20477819442749,
      "learning_rate": 8.246246246246247e-06,
      "loss": 0.3195,
      "step": 17520
    },
    {
      "epoch": 52.64,
      "grad_norm": 13.690117835998535,
      "learning_rate": 8.245245245245246e-06,
      "loss": 0.3772,
      "step": 17530
    },
    {
      "epoch": 52.67,
      "grad_norm": 9.354637145996094,
      "learning_rate": 8.244244244244244e-06,
      "loss": 0.3354,
      "step": 17540
    },
    {
      "epoch": 52.7,
      "grad_norm": 22.30280113220215,
      "learning_rate": 8.243243243243245e-06,
      "loss": 0.3709,
      "step": 17550
    },
    {
      "epoch": 52.73,
      "grad_norm": 10.671855926513672,
      "learning_rate": 8.242242242242243e-06,
      "loss": 0.3412,
      "step": 17560
    },
    {
      "epoch": 52.76,
      "grad_norm": 9.75735092163086,
      "learning_rate": 8.241241241241242e-06,
      "loss": 0.3419,
      "step": 17570
    },
    {
      "epoch": 52.79,
      "grad_norm": 10.60097599029541,
      "learning_rate": 8.24024024024024e-06,
      "loss": 0.3542,
      "step": 17580
    },
    {
      "epoch": 52.82,
      "grad_norm": 14.480786323547363,
      "learning_rate": 8.239239239239239e-06,
      "loss": 0.3598,
      "step": 17590
    },
    {
      "epoch": 52.85,
      "grad_norm": 7.0180206298828125,
      "learning_rate": 8.23823823823824e-06,
      "loss": 0.3961,
      "step": 17600
    },
    {
      "epoch": 52.88,
      "grad_norm": 9.885027885437012,
      "learning_rate": 8.237237237237238e-06,
      "loss": 0.3389,
      "step": 17610
    },
    {
      "epoch": 52.91,
      "grad_norm": 11.352765083312988,
      "learning_rate": 8.236236236236236e-06,
      "loss": 0.3867,
      "step": 17620
    },
    {
      "epoch": 52.94,
      "grad_norm": 32.880615234375,
      "learning_rate": 8.235235235235237e-06,
      "loss": 0.3488,
      "step": 17630
    },
    {
      "epoch": 52.97,
      "grad_norm": 8.713508605957031,
      "learning_rate": 8.234234234234235e-06,
      "loss": 0.3543,
      "step": 17640
    },
    {
      "epoch": 53.0,
      "eval_accuracy": 0.927,
      "eval_loss": 0.2825031876564026,
      "eval_runtime": 30.2771,
      "eval_samples_per_second": 330.283,
      "eval_steps_per_second": 1.321,
      "step": 17649
    },
    {
      "epoch": 53.0,
      "grad_norm": 9.282232284545898,
      "learning_rate": 8.233233233233234e-06,
      "loss": 0.3817,
      "step": 17650
    },
    {
      "epoch": 53.03,
      "grad_norm": 7.064899444580078,
      "learning_rate": 8.232232232232234e-06,
      "loss": 0.368,
      "step": 17660
    },
    {
      "epoch": 53.06,
      "grad_norm": 12.691940307617188,
      "learning_rate": 8.231231231231231e-06,
      "loss": 0.3762,
      "step": 17670
    },
    {
      "epoch": 53.09,
      "grad_norm": 14.303481101989746,
      "learning_rate": 8.230230230230231e-06,
      "loss": 0.3514,
      "step": 17680
    },
    {
      "epoch": 53.12,
      "grad_norm": 5.7945756912231445,
      "learning_rate": 8.22922922922923e-06,
      "loss": 0.2733,
      "step": 17690
    },
    {
      "epoch": 53.15,
      "grad_norm": 7.921624660491943,
      "learning_rate": 8.228228228228229e-06,
      "loss": 0.2949,
      "step": 17700
    },
    {
      "epoch": 53.18,
      "grad_norm": 9.656038284301758,
      "learning_rate": 8.227227227227229e-06,
      "loss": 0.3684,
      "step": 17710
    },
    {
      "epoch": 53.21,
      "grad_norm": 11.320347785949707,
      "learning_rate": 8.226226226226227e-06,
      "loss": 0.3203,
      "step": 17720
    },
    {
      "epoch": 53.24,
      "grad_norm": 9.5058012008667,
      "learning_rate": 8.225225225225226e-06,
      "loss": 0.352,
      "step": 17730
    },
    {
      "epoch": 53.27,
      "grad_norm": 10.746427536010742,
      "learning_rate": 8.224224224224225e-06,
      "loss": 0.3475,
      "step": 17740
    },
    {
      "epoch": 53.3,
      "grad_norm": 7.338408946990967,
      "learning_rate": 8.223223223223223e-06,
      "loss": 0.3676,
      "step": 17750
    },
    {
      "epoch": 53.33,
      "grad_norm": 9.343243598937988,
      "learning_rate": 8.222222222222222e-06,
      "loss": 0.3513,
      "step": 17760
    },
    {
      "epoch": 53.36,
      "grad_norm": 8.686745643615723,
      "learning_rate": 8.221221221221222e-06,
      "loss": 0.3689,
      "step": 17770
    },
    {
      "epoch": 53.39,
      "grad_norm": 11.284011840820312,
      "learning_rate": 8.22022022022022e-06,
      "loss": 0.3592,
      "step": 17780
    },
    {
      "epoch": 53.42,
      "grad_norm": 10.280372619628906,
      "learning_rate": 8.21921921921922e-06,
      "loss": 0.3695,
      "step": 17790
    },
    {
      "epoch": 53.45,
      "grad_norm": 9.912415504455566,
      "learning_rate": 8.21821821821822e-06,
      "loss": 0.3682,
      "step": 17800
    },
    {
      "epoch": 53.48,
      "grad_norm": 16.13150405883789,
      "learning_rate": 8.217217217217218e-06,
      "loss": 0.3602,
      "step": 17810
    },
    {
      "epoch": 53.51,
      "grad_norm": 12.205968856811523,
      "learning_rate": 8.216216216216217e-06,
      "loss": 0.3711,
      "step": 17820
    },
    {
      "epoch": 53.54,
      "grad_norm": 12.626721382141113,
      "learning_rate": 8.215215215215215e-06,
      "loss": 0.376,
      "step": 17830
    },
    {
      "epoch": 53.57,
      "grad_norm": 11.266919136047363,
      "learning_rate": 8.214214214214214e-06,
      "loss": 0.3565,
      "step": 17840
    },
    {
      "epoch": 53.6,
      "grad_norm": 42.46957015991211,
      "learning_rate": 8.213213213213214e-06,
      "loss": 0.3361,
      "step": 17850
    },
    {
      "epoch": 53.63,
      "grad_norm": 9.07781982421875,
      "learning_rate": 8.212212212212213e-06,
      "loss": 0.324,
      "step": 17860
    },
    {
      "epoch": 53.66,
      "grad_norm": 6.815561294555664,
      "learning_rate": 8.211211211211211e-06,
      "loss": 0.3641,
      "step": 17870
    },
    {
      "epoch": 53.69,
      "grad_norm": 12.077738761901855,
      "learning_rate": 8.210210210210212e-06,
      "loss": 0.4144,
      "step": 17880
    },
    {
      "epoch": 53.72,
      "grad_norm": 10.354879379272461,
      "learning_rate": 8.20920920920921e-06,
      "loss": 0.3218,
      "step": 17890
    },
    {
      "epoch": 53.75,
      "grad_norm": 5.950595855712891,
      "learning_rate": 8.208208208208209e-06,
      "loss": 0.3395,
      "step": 17900
    },
    {
      "epoch": 53.78,
      "grad_norm": 13.885716438293457,
      "learning_rate": 8.20720720720721e-06,
      "loss": 0.3419,
      "step": 17910
    },
    {
      "epoch": 53.81,
      "grad_norm": 12.477173805236816,
      "learning_rate": 8.206206206206206e-06,
      "loss": 0.3426,
      "step": 17920
    },
    {
      "epoch": 53.84,
      "grad_norm": 6.916115760803223,
      "learning_rate": 8.205205205205206e-06,
      "loss": 0.3384,
      "step": 17930
    },
    {
      "epoch": 53.87,
      "grad_norm": 10.736311912536621,
      "learning_rate": 8.204204204204205e-06,
      "loss": 0.3172,
      "step": 17940
    },
    {
      "epoch": 53.9,
      "grad_norm": 7.361443519592285,
      "learning_rate": 8.203203203203204e-06,
      "loss": 0.3585,
      "step": 17950
    },
    {
      "epoch": 53.93,
      "grad_norm": 8.18025016784668,
      "learning_rate": 8.202202202202202e-06,
      "loss": 0.3722,
      "step": 17960
    },
    {
      "epoch": 53.96,
      "grad_norm": 6.788581848144531,
      "learning_rate": 8.201201201201202e-06,
      "loss": 0.3605,
      "step": 17970
    },
    {
      "epoch": 53.99,
      "grad_norm": 6.558614253997803,
      "learning_rate": 8.200200200200201e-06,
      "loss": 0.3333,
      "step": 17980
    },
    {
      "epoch": 54.0,
      "eval_accuracy": 0.9277,
      "eval_loss": 0.2766025960445404,
      "eval_runtime": 30.3079,
      "eval_samples_per_second": 329.947,
      "eval_steps_per_second": 1.32,
      "step": 17982
    },
    {
      "epoch": 54.02,
      "grad_norm": 7.770737171173096,
      "learning_rate": 8.1991991991992e-06,
      "loss": 0.2893,
      "step": 17990
    },
    {
      "epoch": 54.05,
      "grad_norm": 10.562653541564941,
      "learning_rate": 8.198198198198198e-06,
      "loss": 0.3836,
      "step": 18000
    },
    {
      "epoch": 54.08,
      "grad_norm": 11.877084732055664,
      "learning_rate": 8.197197197197197e-06,
      "loss": 0.404,
      "step": 18010
    },
    {
      "epoch": 54.11,
      "grad_norm": 10.642131805419922,
      "learning_rate": 8.196196196196197e-06,
      "loss": 0.3348,
      "step": 18020
    },
    {
      "epoch": 54.14,
      "grad_norm": 9.956779479980469,
      "learning_rate": 8.195195195195196e-06,
      "loss": 0.3166,
      "step": 18030
    },
    {
      "epoch": 54.17,
      "grad_norm": 15.740788459777832,
      "learning_rate": 8.194194194194194e-06,
      "loss": 0.3421,
      "step": 18040
    },
    {
      "epoch": 54.2,
      "grad_norm": 11.470704078674316,
      "learning_rate": 8.193193193193195e-06,
      "loss": 0.4111,
      "step": 18050
    },
    {
      "epoch": 54.23,
      "grad_norm": 5.902108192443848,
      "learning_rate": 8.192192192192193e-06,
      "loss": 0.3599,
      "step": 18060
    },
    {
      "epoch": 54.26,
      "grad_norm": 9.207958221435547,
      "learning_rate": 8.191191191191192e-06,
      "loss": 0.3166,
      "step": 18070
    },
    {
      "epoch": 54.29,
      "grad_norm": 10.96752643585205,
      "learning_rate": 8.19019019019019e-06,
      "loss": 0.3613,
      "step": 18080
    },
    {
      "epoch": 54.32,
      "grad_norm": 9.506885528564453,
      "learning_rate": 8.189189189189189e-06,
      "loss": 0.3468,
      "step": 18090
    },
    {
      "epoch": 54.35,
      "grad_norm": 11.816739082336426,
      "learning_rate": 8.18818818818819e-06,
      "loss": 0.4038,
      "step": 18100
    },
    {
      "epoch": 54.38,
      "grad_norm": 6.664322853088379,
      "learning_rate": 8.187187187187188e-06,
      "loss": 0.3421,
      "step": 18110
    },
    {
      "epoch": 54.41,
      "grad_norm": 9.225794792175293,
      "learning_rate": 8.186186186186186e-06,
      "loss": 0.2957,
      "step": 18120
    },
    {
      "epoch": 54.44,
      "grad_norm": 10.506004333496094,
      "learning_rate": 8.185185185185187e-06,
      "loss": 0.3338,
      "step": 18130
    },
    {
      "epoch": 54.47,
      "grad_norm": 8.317553520202637,
      "learning_rate": 8.184184184184185e-06,
      "loss": 0.3878,
      "step": 18140
    },
    {
      "epoch": 54.5,
      "grad_norm": 62.524681091308594,
      "learning_rate": 8.183183183183184e-06,
      "loss": 0.3523,
      "step": 18150
    },
    {
      "epoch": 54.53,
      "grad_norm": 11.02165412902832,
      "learning_rate": 8.182182182182183e-06,
      "loss": 0.323,
      "step": 18160
    },
    {
      "epoch": 54.56,
      "grad_norm": 8.811226844787598,
      "learning_rate": 8.181181181181181e-06,
      "loss": 0.3462,
      "step": 18170
    },
    {
      "epoch": 54.59,
      "grad_norm": 20.77621078491211,
      "learning_rate": 8.18018018018018e-06,
      "loss": 0.317,
      "step": 18180
    },
    {
      "epoch": 54.62,
      "grad_norm": 9.3118257522583,
      "learning_rate": 8.17917917917918e-06,
      "loss": 0.2841,
      "step": 18190
    },
    {
      "epoch": 54.65,
      "grad_norm": 12.934521675109863,
      "learning_rate": 8.178178178178179e-06,
      "loss": 0.3342,
      "step": 18200
    },
    {
      "epoch": 54.68,
      "grad_norm": 9.960111618041992,
      "learning_rate": 8.177177177177177e-06,
      "loss": 0.4035,
      "step": 18210
    },
    {
      "epoch": 54.71,
      "grad_norm": 11.745287895202637,
      "learning_rate": 8.176176176176177e-06,
      "loss": 0.3682,
      "step": 18220
    },
    {
      "epoch": 54.74,
      "grad_norm": 11.538148880004883,
      "learning_rate": 8.175175175175176e-06,
      "loss": 0.401,
      "step": 18230
    },
    {
      "epoch": 54.77,
      "grad_norm": 9.593602180480957,
      "learning_rate": 8.174174174174175e-06,
      "loss": 0.3303,
      "step": 18240
    },
    {
      "epoch": 54.8,
      "grad_norm": 8.58217716217041,
      "learning_rate": 8.173173173173173e-06,
      "loss": 0.3147,
      "step": 18250
    },
    {
      "epoch": 54.83,
      "grad_norm": 11.687836647033691,
      "learning_rate": 8.172172172172172e-06,
      "loss": 0.3289,
      "step": 18260
    },
    {
      "epoch": 54.86,
      "grad_norm": 13.715713500976562,
      "learning_rate": 8.171171171171172e-06,
      "loss": 0.3227,
      "step": 18270
    },
    {
      "epoch": 54.89,
      "grad_norm": 11.921757698059082,
      "learning_rate": 8.17017017017017e-06,
      "loss": 0.368,
      "step": 18280
    },
    {
      "epoch": 54.92,
      "grad_norm": 10.40887451171875,
      "learning_rate": 8.16916916916917e-06,
      "loss": 0.3661,
      "step": 18290
    },
    {
      "epoch": 54.95,
      "grad_norm": 6.441638946533203,
      "learning_rate": 8.16816816816817e-06,
      "loss": 0.3795,
      "step": 18300
    },
    {
      "epoch": 54.98,
      "grad_norm": 11.503289222717285,
      "learning_rate": 8.167167167167168e-06,
      "loss": 0.3695,
      "step": 18310
    },
    {
      "epoch": 55.0,
      "eval_accuracy": 0.9293,
      "eval_loss": 0.27367380261421204,
      "eval_runtime": 30.402,
      "eval_samples_per_second": 328.926,
      "eval_steps_per_second": 1.316,
      "step": 18315
    },
    {
      "epoch": 55.02,
      "grad_norm": 7.902072429656982,
      "learning_rate": 8.166166166166167e-06,
      "loss": 0.3149,
      "step": 18320
    },
    {
      "epoch": 55.05,
      "grad_norm": 9.284063339233398,
      "learning_rate": 8.165165165165165e-06,
      "loss": 0.369,
      "step": 18330
    },
    {
      "epoch": 55.08,
      "grad_norm": 11.693553924560547,
      "learning_rate": 8.164164164164164e-06,
      "loss": 0.3414,
      "step": 18340
    },
    {
      "epoch": 55.11,
      "grad_norm": 17.607770919799805,
      "learning_rate": 8.163163163163164e-06,
      "loss": 0.3965,
      "step": 18350
    },
    {
      "epoch": 55.14,
      "grad_norm": 9.896686553955078,
      "learning_rate": 8.162162162162163e-06,
      "loss": 0.3541,
      "step": 18360
    },
    {
      "epoch": 55.17,
      "grad_norm": 6.082874298095703,
      "learning_rate": 8.161161161161161e-06,
      "loss": 0.3435,
      "step": 18370
    },
    {
      "epoch": 55.2,
      "grad_norm": 9.673128128051758,
      "learning_rate": 8.160160160160162e-06,
      "loss": 0.3182,
      "step": 18380
    },
    {
      "epoch": 55.23,
      "grad_norm": 11.71057415008545,
      "learning_rate": 8.15915915915916e-06,
      "loss": 0.3432,
      "step": 18390
    },
    {
      "epoch": 55.26,
      "grad_norm": 11.963470458984375,
      "learning_rate": 8.158158158158159e-06,
      "loss": 0.3272,
      "step": 18400
    },
    {
      "epoch": 55.29,
      "grad_norm": 14.812848091125488,
      "learning_rate": 8.157157157157158e-06,
      "loss": 0.3266,
      "step": 18410
    },
    {
      "epoch": 55.32,
      "grad_norm": 13.797843933105469,
      "learning_rate": 8.156156156156156e-06,
      "loss": 0.2972,
      "step": 18420
    },
    {
      "epoch": 55.35,
      "grad_norm": 8.497953414916992,
      "learning_rate": 8.155155155155155e-06,
      "loss": 0.3286,
      "step": 18430
    },
    {
      "epoch": 55.38,
      "grad_norm": 10.021702766418457,
      "learning_rate": 8.154154154154155e-06,
      "loss": 0.3378,
      "step": 18440
    },
    {
      "epoch": 55.41,
      "grad_norm": 16.62643051147461,
      "learning_rate": 8.153153153153154e-06,
      "loss": 0.3621,
      "step": 18450
    },
    {
      "epoch": 55.44,
      "grad_norm": 11.550724029541016,
      "learning_rate": 8.152152152152152e-06,
      "loss": 0.3122,
      "step": 18460
    },
    {
      "epoch": 55.47,
      "grad_norm": 12.565238952636719,
      "learning_rate": 8.151151151151153e-06,
      "loss": 0.2569,
      "step": 18470
    },
    {
      "epoch": 55.5,
      "grad_norm": 17.987682342529297,
      "learning_rate": 8.150150150150151e-06,
      "loss": 0.3078,
      "step": 18480
    },
    {
      "epoch": 55.53,
      "grad_norm": 13.228401184082031,
      "learning_rate": 8.14914914914915e-06,
      "loss": 0.39,
      "step": 18490
    },
    {
      "epoch": 55.56,
      "grad_norm": 8.513776779174805,
      "learning_rate": 8.148148148148148e-06,
      "loss": 0.3273,
      "step": 18500
    },
    {
      "epoch": 55.59,
      "grad_norm": 6.093255043029785,
      "learning_rate": 8.147147147147147e-06,
      "loss": 0.3185,
      "step": 18510
    },
    {
      "epoch": 55.62,
      "grad_norm": 11.292076110839844,
      "learning_rate": 8.146146146146147e-06,
      "loss": 0.3298,
      "step": 18520
    },
    {
      "epoch": 55.65,
      "grad_norm": 7.587357521057129,
      "learning_rate": 8.145145145145146e-06,
      "loss": 0.3817,
      "step": 18530
    },
    {
      "epoch": 55.68,
      "grad_norm": 15.010319709777832,
      "learning_rate": 8.144144144144144e-06,
      "loss": 0.3414,
      "step": 18540
    },
    {
      "epoch": 55.71,
      "grad_norm": 17.194854736328125,
      "learning_rate": 8.143143143143145e-06,
      "loss": 0.309,
      "step": 18550
    },
    {
      "epoch": 55.74,
      "grad_norm": 14.053430557250977,
      "learning_rate": 8.142142142142143e-06,
      "loss": 0.392,
      "step": 18560
    },
    {
      "epoch": 55.77,
      "grad_norm": 10.444727897644043,
      "learning_rate": 8.141141141141142e-06,
      "loss": 0.3595,
      "step": 18570
    },
    {
      "epoch": 55.8,
      "grad_norm": 9.290037155151367,
      "learning_rate": 8.14014014014014e-06,
      "loss": 0.333,
      "step": 18580
    },
    {
      "epoch": 55.83,
      "grad_norm": 8.587113380432129,
      "learning_rate": 8.139139139139139e-06,
      "loss": 0.3355,
      "step": 18590
    },
    {
      "epoch": 55.86,
      "grad_norm": 8.94831371307373,
      "learning_rate": 8.13813813813814e-06,
      "loss": 0.3066,
      "step": 18600
    },
    {
      "epoch": 55.89,
      "grad_norm": 9.935319900512695,
      "learning_rate": 8.137137137137138e-06,
      "loss": 0.3454,
      "step": 18610
    },
    {
      "epoch": 55.92,
      "grad_norm": 13.576254844665527,
      "learning_rate": 8.136136136136137e-06,
      "loss": 0.34,
      "step": 18620
    },
    {
      "epoch": 55.95,
      "grad_norm": 13.756028175354004,
      "learning_rate": 8.135135135135137e-06,
      "loss": 0.3466,
      "step": 18630
    },
    {
      "epoch": 55.98,
      "grad_norm": 5.098536968231201,
      "learning_rate": 8.134134134134135e-06,
      "loss": 0.3945,
      "step": 18640
    },
    {
      "epoch": 56.0,
      "eval_accuracy": 0.9282,
      "eval_loss": 0.27599674463272095,
      "eval_runtime": 30.2925,
      "eval_samples_per_second": 330.115,
      "eval_steps_per_second": 1.32,
      "step": 18648
    },
    {
      "epoch": 56.01,
      "grad_norm": 11.784525871276855,
      "learning_rate": 8.133133133133134e-06,
      "loss": 0.4047,
      "step": 18650
    },
    {
      "epoch": 56.04,
      "grad_norm": 14.310515403747559,
      "learning_rate": 8.132132132132133e-06,
      "loss": 0.3768,
      "step": 18660
    },
    {
      "epoch": 56.07,
      "grad_norm": 8.270451545715332,
      "learning_rate": 8.131131131131131e-06,
      "loss": 0.3131,
      "step": 18670
    },
    {
      "epoch": 56.1,
      "grad_norm": 8.968624114990234,
      "learning_rate": 8.13013013013013e-06,
      "loss": 0.3339,
      "step": 18680
    },
    {
      "epoch": 56.13,
      "grad_norm": 7.150795936584473,
      "learning_rate": 8.12912912912913e-06,
      "loss": 0.3095,
      "step": 18690
    },
    {
      "epoch": 56.16,
      "grad_norm": 10.851078987121582,
      "learning_rate": 8.128128128128129e-06,
      "loss": 0.3407,
      "step": 18700
    },
    {
      "epoch": 56.19,
      "grad_norm": 13.158836364746094,
      "learning_rate": 8.127127127127127e-06,
      "loss": 0.3466,
      "step": 18710
    },
    {
      "epoch": 56.22,
      "grad_norm": 10.966217994689941,
      "learning_rate": 8.126126126126128e-06,
      "loss": 0.3819,
      "step": 18720
    },
    {
      "epoch": 56.25,
      "grad_norm": 8.122036933898926,
      "learning_rate": 8.125125125125126e-06,
      "loss": 0.3462,
      "step": 18730
    },
    {
      "epoch": 56.28,
      "grad_norm": 10.184418678283691,
      "learning_rate": 8.124124124124125e-06,
      "loss": 0.3332,
      "step": 18740
    },
    {
      "epoch": 56.31,
      "grad_norm": 8.96741008758545,
      "learning_rate": 8.123123123123123e-06,
      "loss": 0.3241,
      "step": 18750
    },
    {
      "epoch": 56.34,
      "grad_norm": 12.725020408630371,
      "learning_rate": 8.122122122122122e-06,
      "loss": 0.3325,
      "step": 18760
    },
    {
      "epoch": 56.37,
      "grad_norm": 11.463525772094727,
      "learning_rate": 8.121121121121122e-06,
      "loss": 0.3146,
      "step": 18770
    },
    {
      "epoch": 56.4,
      "grad_norm": 123.87519836425781,
      "learning_rate": 8.12012012012012e-06,
      "loss": 0.3352,
      "step": 18780
    },
    {
      "epoch": 56.43,
      "grad_norm": 12.013495445251465,
      "learning_rate": 8.11911911911912e-06,
      "loss": 0.364,
      "step": 18790
    },
    {
      "epoch": 56.46,
      "grad_norm": 16.150331497192383,
      "learning_rate": 8.11811811811812e-06,
      "loss": 0.3325,
      "step": 18800
    },
    {
      "epoch": 56.49,
      "grad_norm": 9.468424797058105,
      "learning_rate": 8.117117117117118e-06,
      "loss": 0.3155,
      "step": 18810
    },
    {
      "epoch": 56.52,
      "grad_norm": 11.889431953430176,
      "learning_rate": 8.116116116116117e-06,
      "loss": 0.3669,
      "step": 18820
    },
    {
      "epoch": 56.55,
      "grad_norm": 7.2081756591796875,
      "learning_rate": 8.115115115115115e-06,
      "loss": 0.3202,
      "step": 18830
    },
    {
      "epoch": 56.58,
      "grad_norm": 7.230907917022705,
      "learning_rate": 8.114114114114114e-06,
      "loss": 0.3221,
      "step": 18840
    },
    {
      "epoch": 56.61,
      "grad_norm": 10.749356269836426,
      "learning_rate": 8.113113113113114e-06,
      "loss": 0.3688,
      "step": 18850
    },
    {
      "epoch": 56.64,
      "grad_norm": 8.51580810546875,
      "learning_rate": 8.112112112112113e-06,
      "loss": 0.3246,
      "step": 18860
    },
    {
      "epoch": 56.67,
      "grad_norm": 13.439396858215332,
      "learning_rate": 8.111111111111112e-06,
      "loss": 0.3556,
      "step": 18870
    },
    {
      "epoch": 56.7,
      "grad_norm": 11.260010719299316,
      "learning_rate": 8.11011011011011e-06,
      "loss": 0.3071,
      "step": 18880
    },
    {
      "epoch": 56.73,
      "grad_norm": 23.281349182128906,
      "learning_rate": 8.10910910910911e-06,
      "loss": 0.3538,
      "step": 18890
    },
    {
      "epoch": 56.76,
      "grad_norm": 8.378347396850586,
      "learning_rate": 8.108108108108109e-06,
      "loss": 0.3076,
      "step": 18900
    },
    {
      "epoch": 56.79,
      "grad_norm": 9.881134033203125,
      "learning_rate": 8.107107107107108e-06,
      "loss": 0.3357,
      "step": 18910
    },
    {
      "epoch": 56.82,
      "grad_norm": 11.504642486572266,
      "learning_rate": 8.106106106106106e-06,
      "loss": 0.33,
      "step": 18920
    },
    {
      "epoch": 56.85,
      "grad_norm": 6.207367420196533,
      "learning_rate": 8.105105105105105e-06,
      "loss": 0.3227,
      "step": 18930
    },
    {
      "epoch": 56.88,
      "grad_norm": 12.955135345458984,
      "learning_rate": 8.104104104104105e-06,
      "loss": 0.3534,
      "step": 18940
    },
    {
      "epoch": 56.91,
      "grad_norm": 13.467114448547363,
      "learning_rate": 8.103103103103104e-06,
      "loss": 0.3482,
      "step": 18950
    },
    {
      "epoch": 56.94,
      "grad_norm": 7.802535533905029,
      "learning_rate": 8.102102102102102e-06,
      "loss": 0.4006,
      "step": 18960
    },
    {
      "epoch": 56.97,
      "grad_norm": 9.149276733398438,
      "learning_rate": 8.101101101101103e-06,
      "loss": 0.3601,
      "step": 18970
    },
    {
      "epoch": 57.0,
      "grad_norm": 9.608295440673828,
      "learning_rate": 8.100100100100101e-06,
      "loss": 0.3315,
      "step": 18980
    },
    {
      "epoch": 57.0,
      "eval_accuracy": 0.9288,
      "eval_loss": 0.27489736676216125,
      "eval_runtime": 30.3067,
      "eval_samples_per_second": 329.96,
      "eval_steps_per_second": 1.32,
      "step": 18981
    },
    {
      "epoch": 57.03,
      "grad_norm": 9.489301681518555,
      "learning_rate": 8.0990990990991e-06,
      "loss": 0.3403,
      "step": 18990
    },
    {
      "epoch": 57.06,
      "grad_norm": 15.572275161743164,
      "learning_rate": 8.098098098098098e-06,
      "loss": 0.3677,
      "step": 19000
    },
    {
      "epoch": 57.09,
      "grad_norm": 7.044128894805908,
      "learning_rate": 8.097097097097097e-06,
      "loss": 0.3149,
      "step": 19010
    },
    {
      "epoch": 57.12,
      "grad_norm": 7.8285322189331055,
      "learning_rate": 8.096096096096097e-06,
      "loss": 0.3573,
      "step": 19020
    },
    {
      "epoch": 57.15,
      "grad_norm": 6.558124542236328,
      "learning_rate": 8.095095095095096e-06,
      "loss": 0.3102,
      "step": 19030
    },
    {
      "epoch": 57.18,
      "grad_norm": 8.140524864196777,
      "learning_rate": 8.094094094094094e-06,
      "loss": 0.3611,
      "step": 19040
    },
    {
      "epoch": 57.21,
      "grad_norm": 7.13889217376709,
      "learning_rate": 8.093093093093095e-06,
      "loss": 0.3133,
      "step": 19050
    },
    {
      "epoch": 57.24,
      "grad_norm": 12.635396003723145,
      "learning_rate": 8.092092092092093e-06,
      "loss": 0.2942,
      "step": 19060
    },
    {
      "epoch": 57.27,
      "grad_norm": 12.764729499816895,
      "learning_rate": 8.091091091091092e-06,
      "loss": 0.3351,
      "step": 19070
    },
    {
      "epoch": 57.3,
      "grad_norm": 8.1305570602417,
      "learning_rate": 8.09009009009009e-06,
      "loss": 0.3205,
      "step": 19080
    },
    {
      "epoch": 57.33,
      "grad_norm": 10.310760498046875,
      "learning_rate": 8.089089089089089e-06,
      "loss": 0.352,
      "step": 19090
    },
    {
      "epoch": 57.36,
      "grad_norm": 11.773191452026367,
      "learning_rate": 8.088088088088088e-06,
      "loss": 0.3081,
      "step": 19100
    },
    {
      "epoch": 57.39,
      "grad_norm": 7.764888286590576,
      "learning_rate": 8.087087087087088e-06,
      "loss": 0.3155,
      "step": 19110
    },
    {
      "epoch": 57.42,
      "grad_norm": 15.437882423400879,
      "learning_rate": 8.086086086086087e-06,
      "loss": 0.3283,
      "step": 19120
    },
    {
      "epoch": 57.45,
      "grad_norm": 7.997897624969482,
      "learning_rate": 8.085085085085085e-06,
      "loss": 0.3458,
      "step": 19130
    },
    {
      "epoch": 57.48,
      "grad_norm": 8.848065376281738,
      "learning_rate": 8.084084084084085e-06,
      "loss": 0.3667,
      "step": 19140
    },
    {
      "epoch": 57.51,
      "grad_norm": 22.186351776123047,
      "learning_rate": 8.083083083083084e-06,
      "loss": 0.32,
      "step": 19150
    },
    {
      "epoch": 57.54,
      "grad_norm": 15.074905395507812,
      "learning_rate": 8.082082082082083e-06,
      "loss": 0.393,
      "step": 19160
    },
    {
      "epoch": 57.57,
      "grad_norm": 14.982492446899414,
      "learning_rate": 8.081081081081081e-06,
      "loss": 0.3682,
      "step": 19170
    },
    {
      "epoch": 57.6,
      "grad_norm": 21.741924285888672,
      "learning_rate": 8.08008008008008e-06,
      "loss": 0.321,
      "step": 19180
    },
    {
      "epoch": 57.63,
      "grad_norm": 8.799239158630371,
      "learning_rate": 8.07907907907908e-06,
      "loss": 0.3473,
      "step": 19190
    },
    {
      "epoch": 57.66,
      "grad_norm": 12.218758583068848,
      "learning_rate": 8.078078078078079e-06,
      "loss": 0.3326,
      "step": 19200
    },
    {
      "epoch": 57.69,
      "grad_norm": 7.644417762756348,
      "learning_rate": 8.077077077077077e-06,
      "loss": 0.3663,
      "step": 19210
    },
    {
      "epoch": 57.72,
      "grad_norm": 13.230323791503906,
      "learning_rate": 8.076076076076078e-06,
      "loss": 0.3503,
      "step": 19220
    },
    {
      "epoch": 57.75,
      "grad_norm": 9.138818740844727,
      "learning_rate": 8.075075075075076e-06,
      "loss": 0.3462,
      "step": 19230
    },
    {
      "epoch": 57.78,
      "grad_norm": 10.958566665649414,
      "learning_rate": 8.074074074074075e-06,
      "loss": 0.2626,
      "step": 19240
    },
    {
      "epoch": 57.81,
      "grad_norm": 11.654053688049316,
      "learning_rate": 8.073073073073073e-06,
      "loss": 0.3962,
      "step": 19250
    },
    {
      "epoch": 57.84,
      "grad_norm": 10.765387535095215,
      "learning_rate": 8.072072072072072e-06,
      "loss": 0.356,
      "step": 19260
    },
    {
      "epoch": 57.87,
      "grad_norm": 13.09720230102539,
      "learning_rate": 8.071071071071072e-06,
      "loss": 0.3597,
      "step": 19270
    },
    {
      "epoch": 57.9,
      "grad_norm": 10.797560691833496,
      "learning_rate": 8.070070070070071e-06,
      "loss": 0.3954,
      "step": 19280
    },
    {
      "epoch": 57.93,
      "grad_norm": 8.639421463012695,
      "learning_rate": 8.06906906906907e-06,
      "loss": 0.3687,
      "step": 19290
    },
    {
      "epoch": 57.96,
      "grad_norm": 13.017178535461426,
      "learning_rate": 8.06806806806807e-06,
      "loss": 0.3234,
      "step": 19300
    },
    {
      "epoch": 57.99,
      "grad_norm": 10.36191463470459,
      "learning_rate": 8.067067067067068e-06,
      "loss": 0.3012,
      "step": 19310
    },
    {
      "epoch": 58.0,
      "eval_accuracy": 0.9283,
      "eval_loss": 0.274671733379364,
      "eval_runtime": 30.1191,
      "eval_samples_per_second": 332.016,
      "eval_steps_per_second": 1.328,
      "step": 19314
    },
    {
      "epoch": 58.02,
      "grad_norm": 12.083019256591797,
      "learning_rate": 8.066066066066067e-06,
      "loss": 0.3454,
      "step": 19320
    },
    {
      "epoch": 58.05,
      "grad_norm": 15.797082901000977,
      "learning_rate": 8.065065065065066e-06,
      "loss": 0.3463,
      "step": 19330
    },
    {
      "epoch": 58.08,
      "grad_norm": 9.531441688537598,
      "learning_rate": 8.064064064064064e-06,
      "loss": 0.3585,
      "step": 19340
    },
    {
      "epoch": 58.11,
      "grad_norm": 9.490439414978027,
      "learning_rate": 8.063063063063063e-06,
      "loss": 0.3776,
      "step": 19350
    },
    {
      "epoch": 58.14,
      "grad_norm": 8.471771240234375,
      "learning_rate": 8.062062062062063e-06,
      "loss": 0.3162,
      "step": 19360
    },
    {
      "epoch": 58.17,
      "grad_norm": 16.912620544433594,
      "learning_rate": 8.061061061061062e-06,
      "loss": 0.3282,
      "step": 19370
    },
    {
      "epoch": 58.2,
      "grad_norm": 10.107708930969238,
      "learning_rate": 8.06006006006006e-06,
      "loss": 0.3623,
      "step": 19380
    },
    {
      "epoch": 58.23,
      "grad_norm": 14.52938175201416,
      "learning_rate": 8.05905905905906e-06,
      "loss": 0.3968,
      "step": 19390
    },
    {
      "epoch": 58.26,
      "grad_norm": 11.0515775680542,
      "learning_rate": 8.058058058058059e-06,
      "loss": 0.3512,
      "step": 19400
    },
    {
      "epoch": 58.29,
      "grad_norm": 6.04974365234375,
      "learning_rate": 8.057057057057058e-06,
      "loss": 0.3463,
      "step": 19410
    },
    {
      "epoch": 58.32,
      "grad_norm": 9.309150695800781,
      "learning_rate": 8.056056056056056e-06,
      "loss": 0.3024,
      "step": 19420
    },
    {
      "epoch": 58.35,
      "grad_norm": 12.057458877563477,
      "learning_rate": 8.055055055055055e-06,
      "loss": 0.3671,
      "step": 19430
    },
    {
      "epoch": 58.38,
      "grad_norm": 13.13753604888916,
      "learning_rate": 8.054054054054055e-06,
      "loss": 0.3363,
      "step": 19440
    },
    {
      "epoch": 58.41,
      "grad_norm": 25.988725662231445,
      "learning_rate": 8.053053053053054e-06,
      "loss": 0.327,
      "step": 19450
    },
    {
      "epoch": 58.44,
      "grad_norm": 10.23130989074707,
      "learning_rate": 8.052052052052052e-06,
      "loss": 0.3338,
      "step": 19460
    },
    {
      "epoch": 58.47,
      "grad_norm": 12.397786140441895,
      "learning_rate": 8.051051051051053e-06,
      "loss": 0.353,
      "step": 19470
    },
    {
      "epoch": 58.5,
      "grad_norm": 12.78853988647461,
      "learning_rate": 8.050050050050051e-06,
      "loss": 0.3485,
      "step": 19480
    },
    {
      "epoch": 58.53,
      "grad_norm": 7.363010406494141,
      "learning_rate": 8.04904904904905e-06,
      "loss": 0.3757,
      "step": 19490
    },
    {
      "epoch": 58.56,
      "grad_norm": 7.872521877288818,
      "learning_rate": 8.048048048048048e-06,
      "loss": 0.3481,
      "step": 19500
    },
    {
      "epoch": 58.59,
      "grad_norm": 8.132575988769531,
      "learning_rate": 8.047047047047047e-06,
      "loss": 0.3248,
      "step": 19510
    },
    {
      "epoch": 58.62,
      "grad_norm": 7.111752986907959,
      "learning_rate": 8.046046046046047e-06,
      "loss": 0.3235,
      "step": 19520
    },
    {
      "epoch": 58.65,
      "grad_norm": 10.55127239227295,
      "learning_rate": 8.045045045045046e-06,
      "loss": 0.3376,
      "step": 19530
    },
    {
      "epoch": 58.68,
      "grad_norm": 5.368039131164551,
      "learning_rate": 8.044044044044045e-06,
      "loss": 0.3506,
      "step": 19540
    },
    {
      "epoch": 58.71,
      "grad_norm": 8.934823036193848,
      "learning_rate": 8.043043043043043e-06,
      "loss": 0.3551,
      "step": 19550
    },
    {
      "epoch": 58.74,
      "grad_norm": 10.673761367797852,
      "learning_rate": 8.042042042042043e-06,
      "loss": 0.3242,
      "step": 19560
    },
    {
      "epoch": 58.77,
      "grad_norm": 8.829063415527344,
      "learning_rate": 8.041041041041042e-06,
      "loss": 0.3532,
      "step": 19570
    },
    {
      "epoch": 58.8,
      "grad_norm": 5.948292255401611,
      "learning_rate": 8.04004004004004e-06,
      "loss": 0.3689,
      "step": 19580
    },
    {
      "epoch": 58.83,
      "grad_norm": 6.607304573059082,
      "learning_rate": 8.03903903903904e-06,
      "loss": 0.3544,
      "step": 19590
    },
    {
      "epoch": 58.86,
      "grad_norm": 24.354997634887695,
      "learning_rate": 8.038038038038038e-06,
      "loss": 0.3708,
      "step": 19600
    },
    {
      "epoch": 58.89,
      "grad_norm": 7.090005397796631,
      "learning_rate": 8.037037037037038e-06,
      "loss": 0.3213,
      "step": 19610
    },
    {
      "epoch": 58.92,
      "grad_norm": 8.344409942626953,
      "learning_rate": 8.036036036036037e-06,
      "loss": 0.3277,
      "step": 19620
    },
    {
      "epoch": 58.95,
      "grad_norm": 6.210591793060303,
      "learning_rate": 8.035035035035035e-06,
      "loss": 0.3041,
      "step": 19630
    },
    {
      "epoch": 58.98,
      "grad_norm": 10.609624862670898,
      "learning_rate": 8.034034034034036e-06,
      "loss": 0.3261,
      "step": 19640
    },
    {
      "epoch": 59.0,
      "eval_accuracy": 0.9283,
      "eval_loss": 0.28119683265686035,
      "eval_runtime": 30.1725,
      "eval_samples_per_second": 331.427,
      "eval_steps_per_second": 1.326,
      "step": 19647
    },
    {
      "epoch": 59.01,
      "grad_norm": 9.268049240112305,
      "learning_rate": 8.033033033033034e-06,
      "loss": 0.3591,
      "step": 19650
    },
    {
      "epoch": 59.04,
      "grad_norm": 9.366358757019043,
      "learning_rate": 8.032032032032033e-06,
      "loss": 0.3491,
      "step": 19660
    },
    {
      "epoch": 59.07,
      "grad_norm": 9.560079574584961,
      "learning_rate": 8.031031031031031e-06,
      "loss": 0.3866,
      "step": 19670
    },
    {
      "epoch": 59.1,
      "grad_norm": 8.784268379211426,
      "learning_rate": 8.03003003003003e-06,
      "loss": 0.3211,
      "step": 19680
    },
    {
      "epoch": 59.13,
      "grad_norm": 14.688582420349121,
      "learning_rate": 8.02902902902903e-06,
      "loss": 0.3707,
      "step": 19690
    },
    {
      "epoch": 59.16,
      "grad_norm": 7.3063645362854,
      "learning_rate": 8.028028028028029e-06,
      "loss": 0.2788,
      "step": 19700
    },
    {
      "epoch": 59.19,
      "grad_norm": 12.764232635498047,
      "learning_rate": 8.027027027027027e-06,
      "loss": 0.3207,
      "step": 19710
    },
    {
      "epoch": 59.22,
      "grad_norm": 8.961464881896973,
      "learning_rate": 8.026026026026028e-06,
      "loss": 0.3495,
      "step": 19720
    },
    {
      "epoch": 59.25,
      "grad_norm": 17.970264434814453,
      "learning_rate": 8.025025025025026e-06,
      "loss": 0.3344,
      "step": 19730
    },
    {
      "epoch": 59.28,
      "grad_norm": 10.456023216247559,
      "learning_rate": 8.024024024024025e-06,
      "loss": 0.2733,
      "step": 19740
    },
    {
      "epoch": 59.31,
      "grad_norm": 13.454469680786133,
      "learning_rate": 8.023023023023023e-06,
      "loss": 0.3451,
      "step": 19750
    },
    {
      "epoch": 59.34,
      "grad_norm": 9.928228378295898,
      "learning_rate": 8.022022022022022e-06,
      "loss": 0.3103,
      "step": 19760
    },
    {
      "epoch": 59.37,
      "grad_norm": 8.149476051330566,
      "learning_rate": 8.021021021021022e-06,
      "loss": 0.35,
      "step": 19770
    },
    {
      "epoch": 59.4,
      "grad_norm": 8.944438934326172,
      "learning_rate": 8.020020020020021e-06,
      "loss": 0.364,
      "step": 19780
    },
    {
      "epoch": 59.43,
      "grad_norm": 13.062980651855469,
      "learning_rate": 8.01901901901902e-06,
      "loss": 0.3402,
      "step": 19790
    },
    {
      "epoch": 59.46,
      "grad_norm": 11.60055923461914,
      "learning_rate": 8.018018018018018e-06,
      "loss": 0.3895,
      "step": 19800
    },
    {
      "epoch": 59.49,
      "grad_norm": 21.184160232543945,
      "learning_rate": 8.017017017017018e-06,
      "loss": 0.3454,
      "step": 19810
    },
    {
      "epoch": 59.52,
      "grad_norm": 5.904134273529053,
      "learning_rate": 8.016016016016017e-06,
      "loss": 0.3279,
      "step": 19820
    },
    {
      "epoch": 59.55,
      "grad_norm": 12.97242546081543,
      "learning_rate": 8.015015015015016e-06,
      "loss": 0.2958,
      "step": 19830
    },
    {
      "epoch": 59.58,
      "grad_norm": 9.463737487792969,
      "learning_rate": 8.014014014014014e-06,
      "loss": 0.2754,
      "step": 19840
    },
    {
      "epoch": 59.61,
      "grad_norm": 11.5346040725708,
      "learning_rate": 8.013013013013013e-06,
      "loss": 0.259,
      "step": 19850
    },
    {
      "epoch": 59.64,
      "grad_norm": 9.183198928833008,
      "learning_rate": 8.012012012012013e-06,
      "loss": 0.3202,
      "step": 19860
    },
    {
      "epoch": 59.67,
      "grad_norm": 10.773748397827148,
      "learning_rate": 8.011011011011012e-06,
      "loss": 0.3302,
      "step": 19870
    },
    {
      "epoch": 59.7,
      "grad_norm": 11.06137752532959,
      "learning_rate": 8.01001001001001e-06,
      "loss": 0.3712,
      "step": 19880
    },
    {
      "epoch": 59.73,
      "grad_norm": 12.271261215209961,
      "learning_rate": 8.00900900900901e-06,
      "loss": 0.3548,
      "step": 19890
    },
    {
      "epoch": 59.76,
      "grad_norm": 11.94296932220459,
      "learning_rate": 8.00800800800801e-06,
      "loss": 0.3162,
      "step": 19900
    },
    {
      "epoch": 59.79,
      "grad_norm": 9.39575481414795,
      "learning_rate": 8.007007007007008e-06,
      "loss": 0.3625,
      "step": 19910
    },
    {
      "epoch": 59.82,
      "grad_norm": 9.609649658203125,
      "learning_rate": 8.006006006006006e-06,
      "loss": 0.3334,
      "step": 19920
    },
    {
      "epoch": 59.85,
      "grad_norm": 10.835598945617676,
      "learning_rate": 8.005005005005005e-06,
      "loss": 0.3506,
      "step": 19930
    },
    {
      "epoch": 59.88,
      "grad_norm": 11.665934562683105,
      "learning_rate": 8.004004004004005e-06,
      "loss": 0.2988,
      "step": 19940
    },
    {
      "epoch": 59.91,
      "grad_norm": 6.854680061340332,
      "learning_rate": 8.003003003003004e-06,
      "loss": 0.3361,
      "step": 19950
    },
    {
      "epoch": 59.94,
      "grad_norm": 10.184540748596191,
      "learning_rate": 8.002002002002002e-06,
      "loss": 0.2905,
      "step": 19960
    },
    {
      "epoch": 59.97,
      "grad_norm": 12.883761405944824,
      "learning_rate": 8.001001001001003e-06,
      "loss": 0.3746,
      "step": 19970
    },
    {
      "epoch": 60.0,
      "grad_norm": 2.8957912921905518,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.2817,
      "step": 19980
    },
    {
      "epoch": 60.0,
      "eval_accuracy": 0.9285,
      "eval_loss": 0.27537769079208374,
      "eval_runtime": 30.3436,
      "eval_samples_per_second": 329.559,
      "eval_steps_per_second": 1.318,
      "step": 19980
    },
    {
      "epoch": 60.03,
      "grad_norm": 20.665491104125977,
      "learning_rate": 7.998998998999e-06,
      "loss": 0.3215,
      "step": 19990
    },
    {
      "epoch": 60.06,
      "grad_norm": 6.750944137573242,
      "learning_rate": 7.997997997997999e-06,
      "loss": 0.3367,
      "step": 20000
    },
    {
      "epoch": 60.09,
      "grad_norm": 15.198680877685547,
      "learning_rate": 7.996996996996997e-06,
      "loss": 0.3207,
      "step": 20010
    },
    {
      "epoch": 60.12,
      "grad_norm": 10.891490936279297,
      "learning_rate": 7.995995995995996e-06,
      "loss": 0.4046,
      "step": 20020
    },
    {
      "epoch": 60.15,
      "grad_norm": 8.051135063171387,
      "learning_rate": 7.994994994994996e-06,
      "loss": 0.3493,
      "step": 20030
    },
    {
      "epoch": 60.18,
      "grad_norm": 9.352787017822266,
      "learning_rate": 7.993993993993995e-06,
      "loss": 0.3035,
      "step": 20040
    },
    {
      "epoch": 60.21,
      "grad_norm": 10.874890327453613,
      "learning_rate": 7.992992992992993e-06,
      "loss": 0.3549,
      "step": 20050
    },
    {
      "epoch": 60.24,
      "grad_norm": 13.040722846984863,
      "learning_rate": 7.991991991991993e-06,
      "loss": 0.3379,
      "step": 20060
    },
    {
      "epoch": 60.27,
      "grad_norm": 6.828490734100342,
      "learning_rate": 7.990990990990992e-06,
      "loss": 0.375,
      "step": 20070
    },
    {
      "epoch": 60.3,
      "grad_norm": 8.598703384399414,
      "learning_rate": 7.98998998998999e-06,
      "loss": 0.3453,
      "step": 20080
    },
    {
      "epoch": 60.33,
      "grad_norm": 10.424263000488281,
      "learning_rate": 7.98898898898899e-06,
      "loss": 0.3273,
      "step": 20090
    },
    {
      "epoch": 60.36,
      "grad_norm": 8.203944206237793,
      "learning_rate": 7.987987987987988e-06,
      "loss": 0.3261,
      "step": 20100
    },
    {
      "epoch": 60.39,
      "grad_norm": 10.265007019042969,
      "learning_rate": 7.986986986986988e-06,
      "loss": 0.3408,
      "step": 20110
    },
    {
      "epoch": 60.42,
      "grad_norm": 19.575834274291992,
      "learning_rate": 7.985985985985987e-06,
      "loss": 0.3343,
      "step": 20120
    },
    {
      "epoch": 60.45,
      "grad_norm": 10.629146575927734,
      "learning_rate": 7.984984984984985e-06,
      "loss": 0.3365,
      "step": 20130
    },
    {
      "epoch": 60.48,
      "grad_norm": 17.273698806762695,
      "learning_rate": 7.983983983983986e-06,
      "loss": 0.3241,
      "step": 20140
    },
    {
      "epoch": 60.51,
      "grad_norm": 6.781803607940674,
      "learning_rate": 7.982982982982984e-06,
      "loss": 0.3046,
      "step": 20150
    },
    {
      "epoch": 60.54,
      "grad_norm": 6.762634754180908,
      "learning_rate": 7.981981981981983e-06,
      "loss": 0.2996,
      "step": 20160
    },
    {
      "epoch": 60.57,
      "grad_norm": 15.066496849060059,
      "learning_rate": 7.980980980980981e-06,
      "loss": 0.307,
      "step": 20170
    },
    {
      "epoch": 60.6,
      "grad_norm": 13.316816329956055,
      "learning_rate": 7.97997997997998e-06,
      "loss": 0.3009,
      "step": 20180
    },
    {
      "epoch": 60.63,
      "grad_norm": 9.425992965698242,
      "learning_rate": 7.97897897897898e-06,
      "loss": 0.3465,
      "step": 20190
    },
    {
      "epoch": 60.66,
      "grad_norm": 11.520425796508789,
      "learning_rate": 7.977977977977979e-06,
      "loss": 0.3164,
      "step": 20200
    },
    {
      "epoch": 60.69,
      "grad_norm": 8.337925910949707,
      "learning_rate": 7.976976976976977e-06,
      "loss": 0.3179,
      "step": 20210
    },
    {
      "epoch": 60.72,
      "grad_norm": 7.222460746765137,
      "learning_rate": 7.975975975975978e-06,
      "loss": 0.3384,
      "step": 20220
    },
    {
      "epoch": 60.75,
      "grad_norm": 28.688295364379883,
      "learning_rate": 7.974974974974976e-06,
      "loss": 0.3388,
      "step": 20230
    },
    {
      "epoch": 60.78,
      "grad_norm": 21.542186737060547,
      "learning_rate": 7.973973973973973e-06,
      "loss": 0.3309,
      "step": 20240
    },
    {
      "epoch": 60.81,
      "grad_norm": 25.37692642211914,
      "learning_rate": 7.972972972972974e-06,
      "loss": 0.3361,
      "step": 20250
    },
    {
      "epoch": 60.84,
      "grad_norm": 9.750048637390137,
      "learning_rate": 7.971971971971972e-06,
      "loss": 0.3572,
      "step": 20260
    },
    {
      "epoch": 60.87,
      "grad_norm": 6.711831569671631,
      "learning_rate": 7.97097097097097e-06,
      "loss": 0.2961,
      "step": 20270
    },
    {
      "epoch": 60.9,
      "grad_norm": 6.276210308074951,
      "learning_rate": 7.969969969969971e-06,
      "loss": 0.352,
      "step": 20280
    },
    {
      "epoch": 60.93,
      "grad_norm": 15.82321548461914,
      "learning_rate": 7.96896896896897e-06,
      "loss": 0.3216,
      "step": 20290
    },
    {
      "epoch": 60.96,
      "grad_norm": 7.726736545562744,
      "learning_rate": 7.967967967967968e-06,
      "loss": 0.3175,
      "step": 20300
    },
    {
      "epoch": 60.99,
      "grad_norm": 9.728242874145508,
      "learning_rate": 7.966966966966969e-06,
      "loss": 0.3039,
      "step": 20310
    },
    {
      "epoch": 61.0,
      "eval_accuracy": 0.929,
      "eval_loss": 0.28187862038612366,
      "eval_runtime": 30.4712,
      "eval_samples_per_second": 328.179,
      "eval_steps_per_second": 1.313,
      "step": 20313
    },
    {
      "epoch": 61.02,
      "grad_norm": 6.557823657989502,
      "learning_rate": 7.965965965965967e-06,
      "loss": 0.3423,
      "step": 20320
    },
    {
      "epoch": 61.05,
      "grad_norm": 11.972136497497559,
      "learning_rate": 7.964964964964966e-06,
      "loss": 0.3372,
      "step": 20330
    },
    {
      "epoch": 61.08,
      "grad_norm": 8.896626472473145,
      "learning_rate": 7.963963963963964e-06,
      "loss": 0.3347,
      "step": 20340
    },
    {
      "epoch": 61.11,
      "grad_norm": 8.908828735351562,
      "learning_rate": 7.962962962962963e-06,
      "loss": 0.3366,
      "step": 20350
    },
    {
      "epoch": 61.14,
      "grad_norm": 12.13779067993164,
      "learning_rate": 7.961961961961963e-06,
      "loss": 0.2991,
      "step": 20360
    },
    {
      "epoch": 61.17,
      "grad_norm": 11.553640365600586,
      "learning_rate": 7.960960960960962e-06,
      "loss": 0.3402,
      "step": 20370
    },
    {
      "epoch": 61.2,
      "grad_norm": 14.877877235412598,
      "learning_rate": 7.95995995995996e-06,
      "loss": 0.32,
      "step": 20380
    },
    {
      "epoch": 61.23,
      "grad_norm": 8.307705879211426,
      "learning_rate": 7.95895895895896e-06,
      "loss": 0.3485,
      "step": 20390
    },
    {
      "epoch": 61.26,
      "grad_norm": 9.19522476196289,
      "learning_rate": 7.95795795795796e-06,
      "loss": 0.3019,
      "step": 20400
    },
    {
      "epoch": 61.29,
      "grad_norm": 11.361839294433594,
      "learning_rate": 7.956956956956958e-06,
      "loss": 0.3625,
      "step": 20410
    },
    {
      "epoch": 61.32,
      "grad_norm": 7.35371732711792,
      "learning_rate": 7.955955955955956e-06,
      "loss": 0.3109,
      "step": 20420
    },
    {
      "epoch": 61.35,
      "grad_norm": 10.084836959838867,
      "learning_rate": 7.954954954954955e-06,
      "loss": 0.3937,
      "step": 20430
    },
    {
      "epoch": 61.38,
      "grad_norm": 10.313196182250977,
      "learning_rate": 7.953953953953955e-06,
      "loss": 0.3199,
      "step": 20440
    },
    {
      "epoch": 61.41,
      "grad_norm": 11.044137954711914,
      "learning_rate": 7.952952952952954e-06,
      "loss": 0.3741,
      "step": 20450
    },
    {
      "epoch": 61.44,
      "grad_norm": 7.936108112335205,
      "learning_rate": 7.951951951951953e-06,
      "loss": 0.3466,
      "step": 20460
    },
    {
      "epoch": 61.47,
      "grad_norm": 7.13407039642334,
      "learning_rate": 7.950950950950951e-06,
      "loss": 0.3065,
      "step": 20470
    },
    {
      "epoch": 61.5,
      "grad_norm": 8.788211822509766,
      "learning_rate": 7.949949949949951e-06,
      "loss": 0.3505,
      "step": 20480
    },
    {
      "epoch": 61.53,
      "grad_norm": 8.706562995910645,
      "learning_rate": 7.948948948948948e-06,
      "loss": 0.361,
      "step": 20490
    },
    {
      "epoch": 61.56,
      "grad_norm": 8.077728271484375,
      "learning_rate": 7.947947947947949e-06,
      "loss": 0.3122,
      "step": 20500
    },
    {
      "epoch": 61.59,
      "grad_norm": 8.773408889770508,
      "learning_rate": 7.946946946946947e-06,
      "loss": 0.3299,
      "step": 20510
    },
    {
      "epoch": 61.62,
      "grad_norm": 11.315860748291016,
      "learning_rate": 7.945945945945946e-06,
      "loss": 0.3479,
      "step": 20520
    },
    {
      "epoch": 61.65,
      "grad_norm": 7.7536749839782715,
      "learning_rate": 7.944944944944946e-06,
      "loss": 0.3361,
      "step": 20530
    },
    {
      "epoch": 61.68,
      "grad_norm": 9.027618408203125,
      "learning_rate": 7.943943943943945e-06,
      "loss": 0.3089,
      "step": 20540
    },
    {
      "epoch": 61.71,
      "grad_norm": 8.191462516784668,
      "learning_rate": 7.942942942942943e-06,
      "loss": 0.3456,
      "step": 20550
    },
    {
      "epoch": 61.74,
      "grad_norm": 5.5945281982421875,
      "learning_rate": 7.941941941941944e-06,
      "loss": 0.3104,
      "step": 20560
    },
    {
      "epoch": 61.77,
      "grad_norm": 8.97971248626709,
      "learning_rate": 7.940940940940942e-06,
      "loss": 0.3408,
      "step": 20570
    },
    {
      "epoch": 61.8,
      "grad_norm": 7.762706756591797,
      "learning_rate": 7.93993993993994e-06,
      "loss": 0.3397,
      "step": 20580
    },
    {
      "epoch": 61.83,
      "grad_norm": 10.23642635345459,
      "learning_rate": 7.93893893893894e-06,
      "loss": 0.3621,
      "step": 20590
    },
    {
      "epoch": 61.86,
      "grad_norm": 12.05219841003418,
      "learning_rate": 7.937937937937938e-06,
      "loss": 0.3423,
      "step": 20600
    },
    {
      "epoch": 61.89,
      "grad_norm": 11.520780563354492,
      "learning_rate": 7.936936936936938e-06,
      "loss": 0.349,
      "step": 20610
    },
    {
      "epoch": 61.92,
      "grad_norm": 11.037263870239258,
      "learning_rate": 7.935935935935937e-06,
      "loss": 0.2548,
      "step": 20620
    },
    {
      "epoch": 61.95,
      "grad_norm": 10.236679077148438,
      "learning_rate": 7.934934934934935e-06,
      "loss": 0.3059,
      "step": 20630
    },
    {
      "epoch": 61.98,
      "grad_norm": 7.142411708831787,
      "learning_rate": 7.933933933933936e-06,
      "loss": 0.306,
      "step": 20640
    },
    {
      "epoch": 62.0,
      "eval_accuracy": 0.9293,
      "eval_loss": 0.28156551718711853,
      "eval_runtime": 30.3478,
      "eval_samples_per_second": 329.513,
      "eval_steps_per_second": 1.318,
      "step": 20646
    },
    {
      "epoch": 62.01,
      "grad_norm": 11.385009765625,
      "learning_rate": 7.932932932932934e-06,
      "loss": 0.2795,
      "step": 20650
    },
    {
      "epoch": 62.04,
      "grad_norm": 9.382291793823242,
      "learning_rate": 7.931931931931933e-06,
      "loss": 0.3312,
      "step": 20660
    },
    {
      "epoch": 62.07,
      "grad_norm": 10.223499298095703,
      "learning_rate": 7.930930930930931e-06,
      "loss": 0.3431,
      "step": 20670
    },
    {
      "epoch": 62.1,
      "grad_norm": 15.718355178833008,
      "learning_rate": 7.92992992992993e-06,
      "loss": 0.365,
      "step": 20680
    },
    {
      "epoch": 62.13,
      "grad_norm": 7.110666275024414,
      "learning_rate": 7.928928928928929e-06,
      "loss": 0.3755,
      "step": 20690
    },
    {
      "epoch": 62.16,
      "grad_norm": 9.10853385925293,
      "learning_rate": 7.927927927927929e-06,
      "loss": 0.3626,
      "step": 20700
    },
    {
      "epoch": 62.19,
      "grad_norm": 14.082881927490234,
      "learning_rate": 7.926926926926928e-06,
      "loss": 0.3223,
      "step": 20710
    },
    {
      "epoch": 62.22,
      "grad_norm": 11.472249031066895,
      "learning_rate": 7.925925925925926e-06,
      "loss": 0.3142,
      "step": 20720
    },
    {
      "epoch": 62.25,
      "grad_norm": 12.035321235656738,
      "learning_rate": 7.924924924924926e-06,
      "loss": 0.3527,
      "step": 20730
    },
    {
      "epoch": 62.28,
      "grad_norm": 12.315011978149414,
      "learning_rate": 7.923923923923923e-06,
      "loss": 0.3673,
      "step": 20740
    },
    {
      "epoch": 62.31,
      "grad_norm": 10.736784934997559,
      "learning_rate": 7.922922922922924e-06,
      "loss": 0.3099,
      "step": 20750
    },
    {
      "epoch": 62.34,
      "grad_norm": 20.91286277770996,
      "learning_rate": 7.921921921921922e-06,
      "loss": 0.3215,
      "step": 20760
    },
    {
      "epoch": 62.37,
      "grad_norm": 9.482779502868652,
      "learning_rate": 7.92092092092092e-06,
      "loss": 0.2838,
      "step": 20770
    },
    {
      "epoch": 62.4,
      "grad_norm": 8.627375602722168,
      "learning_rate": 7.919919919919921e-06,
      "loss": 0.3091,
      "step": 20780
    },
    {
      "epoch": 62.43,
      "grad_norm": 11.477374076843262,
      "learning_rate": 7.91891891891892e-06,
      "loss": 0.3144,
      "step": 20790
    },
    {
      "epoch": 62.46,
      "grad_norm": 7.87100076675415,
      "learning_rate": 7.917917917917918e-06,
      "loss": 0.3548,
      "step": 20800
    },
    {
      "epoch": 62.49,
      "grad_norm": 6.039185523986816,
      "learning_rate": 7.916916916916919e-06,
      "loss": 0.307,
      "step": 20810
    },
    {
      "epoch": 62.52,
      "grad_norm": 17.07259750366211,
      "learning_rate": 7.915915915915915e-06,
      "loss": 0.3,
      "step": 20820
    },
    {
      "epoch": 62.55,
      "grad_norm": 11.636271476745605,
      "learning_rate": 7.914914914914916e-06,
      "loss": 0.3683,
      "step": 20830
    },
    {
      "epoch": 62.58,
      "grad_norm": 20.401649475097656,
      "learning_rate": 7.913913913913914e-06,
      "loss": 0.3629,
      "step": 20840
    },
    {
      "epoch": 62.61,
      "grad_norm": 12.574553489685059,
      "learning_rate": 7.912912912912913e-06,
      "loss": 0.3491,
      "step": 20850
    },
    {
      "epoch": 62.64,
      "grad_norm": 7.269466400146484,
      "learning_rate": 7.911911911911913e-06,
      "loss": 0.3222,
      "step": 20860
    },
    {
      "epoch": 62.67,
      "grad_norm": 8.476310729980469,
      "learning_rate": 7.910910910910912e-06,
      "loss": 0.3426,
      "step": 20870
    },
    {
      "epoch": 62.7,
      "grad_norm": 7.763072967529297,
      "learning_rate": 7.90990990990991e-06,
      "loss": 0.3565,
      "step": 20880
    },
    {
      "epoch": 62.73,
      "grad_norm": 6.947924613952637,
      "learning_rate": 7.90890890890891e-06,
      "loss": 0.299,
      "step": 20890
    },
    {
      "epoch": 62.76,
      "grad_norm": 5.7818779945373535,
      "learning_rate": 7.90790790790791e-06,
      "loss": 0.3011,
      "step": 20900
    },
    {
      "epoch": 62.79,
      "grad_norm": 8.12251091003418,
      "learning_rate": 7.906906906906908e-06,
      "loss": 0.4142,
      "step": 20910
    },
    {
      "epoch": 62.82,
      "grad_norm": 10.4780855178833,
      "learning_rate": 7.905905905905907e-06,
      "loss": 0.3374,
      "step": 20920
    },
    {
      "epoch": 62.85,
      "grad_norm": 13.668453216552734,
      "learning_rate": 7.904904904904905e-06,
      "loss": 0.3484,
      "step": 20930
    },
    {
      "epoch": 62.88,
      "grad_norm": 7.8147172927856445,
      "learning_rate": 7.903903903903904e-06,
      "loss": 0.3226,
      "step": 20940
    },
    {
      "epoch": 62.91,
      "grad_norm": 7.263507843017578,
      "learning_rate": 7.902902902902904e-06,
      "loss": 0.3286,
      "step": 20950
    },
    {
      "epoch": 62.94,
      "grad_norm": 7.895671367645264,
      "learning_rate": 7.901901901901903e-06,
      "loss": 0.3337,
      "step": 20960
    },
    {
      "epoch": 62.97,
      "grad_norm": 9.506521224975586,
      "learning_rate": 7.900900900900901e-06,
      "loss": 0.3166,
      "step": 20970
    },
    {
      "epoch": 63.0,
      "eval_accuracy": 0.9265,
      "eval_loss": 0.2839010953903198,
      "eval_runtime": 30.1719,
      "eval_samples_per_second": 331.434,
      "eval_steps_per_second": 1.326,
      "step": 20979
    },
    {
      "epoch": 63.0,
      "grad_norm": 8.6882963180542,
      "learning_rate": 7.899899899899901e-06,
      "loss": 0.2617,
      "step": 20980
    },
    {
      "epoch": 63.03,
      "grad_norm": 48.44770050048828,
      "learning_rate": 7.898898898898898e-06,
      "loss": 0.3401,
      "step": 20990
    },
    {
      "epoch": 63.06,
      "grad_norm": 12.187158584594727,
      "learning_rate": 7.897897897897899e-06,
      "loss": 0.3386,
      "step": 21000
    },
    {
      "epoch": 63.09,
      "grad_norm": 8.34494686126709,
      "learning_rate": 7.896896896896897e-06,
      "loss": 0.3315,
      "step": 21010
    },
    {
      "epoch": 63.12,
      "grad_norm": 10.482794761657715,
      "learning_rate": 7.895895895895896e-06,
      "loss": 0.2906,
      "step": 21020
    },
    {
      "epoch": 63.15,
      "grad_norm": 8.335676193237305,
      "learning_rate": 7.894894894894896e-06,
      "loss": 0.3315,
      "step": 21030
    },
    {
      "epoch": 63.18,
      "grad_norm": 9.097391128540039,
      "learning_rate": 7.893893893893895e-06,
      "loss": 0.3475,
      "step": 21040
    },
    {
      "epoch": 63.21,
      "grad_norm": 10.687468528747559,
      "learning_rate": 7.892892892892893e-06,
      "loss": 0.3,
      "step": 21050
    },
    {
      "epoch": 63.24,
      "grad_norm": 15.741429328918457,
      "learning_rate": 7.891891891891894e-06,
      "loss": 0.3016,
      "step": 21060
    },
    {
      "epoch": 63.27,
      "grad_norm": 9.980709075927734,
      "learning_rate": 7.89089089089089e-06,
      "loss": 0.2775,
      "step": 21070
    },
    {
      "epoch": 63.3,
      "grad_norm": 9.483716011047363,
      "learning_rate": 7.88988988988989e-06,
      "loss": 0.2942,
      "step": 21080
    },
    {
      "epoch": 63.33,
      "grad_norm": 11.17950439453125,
      "learning_rate": 7.88888888888889e-06,
      "loss": 0.3159,
      "step": 21090
    },
    {
      "epoch": 63.36,
      "grad_norm": 8.637929916381836,
      "learning_rate": 7.887887887887888e-06,
      "loss": 0.3131,
      "step": 21100
    },
    {
      "epoch": 63.39,
      "grad_norm": 13.111700057983398,
      "learning_rate": 7.886886886886888e-06,
      "loss": 0.3285,
      "step": 21110
    },
    {
      "epoch": 63.42,
      "grad_norm": 18.82198715209961,
      "learning_rate": 7.885885885885887e-06,
      "loss": 0.3571,
      "step": 21120
    },
    {
      "epoch": 63.45,
      "grad_norm": 11.237526893615723,
      "learning_rate": 7.884884884884885e-06,
      "loss": 0.3453,
      "step": 21130
    },
    {
      "epoch": 63.48,
      "grad_norm": 14.27531623840332,
      "learning_rate": 7.883883883883886e-06,
      "loss": 0.3856,
      "step": 21140
    },
    {
      "epoch": 63.51,
      "grad_norm": 9.593804359436035,
      "learning_rate": 7.882882882882884e-06,
      "loss": 0.32,
      "step": 21150
    },
    {
      "epoch": 63.54,
      "grad_norm": 9.467520713806152,
      "learning_rate": 7.881881881881881e-06,
      "loss": 0.3034,
      "step": 21160
    },
    {
      "epoch": 63.57,
      "grad_norm": 14.424857139587402,
      "learning_rate": 7.880880880880882e-06,
      "loss": 0.3217,
      "step": 21170
    },
    {
      "epoch": 63.6,
      "grad_norm": 9.851648330688477,
      "learning_rate": 7.87987987987988e-06,
      "loss": 0.3242,
      "step": 21180
    },
    {
      "epoch": 63.63,
      "grad_norm": 9.80981731414795,
      "learning_rate": 7.878878878878879e-06,
      "loss": 0.334,
      "step": 21190
    },
    {
      "epoch": 63.66,
      "grad_norm": 9.978001594543457,
      "learning_rate": 7.877877877877879e-06,
      "loss": 0.3306,
      "step": 21200
    },
    {
      "epoch": 63.69,
      "grad_norm": 12.903973579406738,
      "learning_rate": 7.876876876876878e-06,
      "loss": 0.3602,
      "step": 21210
    },
    {
      "epoch": 63.72,
      "grad_norm": 20.115964889526367,
      "learning_rate": 7.875875875875876e-06,
      "loss": 0.3137,
      "step": 21220
    },
    {
      "epoch": 63.75,
      "grad_norm": 11.106460571289062,
      "learning_rate": 7.874874874874877e-06,
      "loss": 0.2826,
      "step": 21230
    },
    {
      "epoch": 63.78,
      "grad_norm": 29.17913246154785,
      "learning_rate": 7.873873873873873e-06,
      "loss": 0.3261,
      "step": 21240
    },
    {
      "epoch": 63.81,
      "grad_norm": 11.208224296569824,
      "learning_rate": 7.872872872872874e-06,
      "loss": 0.3092,
      "step": 21250
    },
    {
      "epoch": 63.84,
      "grad_norm": 13.784420013427734,
      "learning_rate": 7.871871871871872e-06,
      "loss": 0.3025,
      "step": 21260
    },
    {
      "epoch": 63.87,
      "grad_norm": 9.042902946472168,
      "learning_rate": 7.870870870870871e-06,
      "loss": 0.3304,
      "step": 21270
    },
    {
      "epoch": 63.9,
      "grad_norm": 10.465621948242188,
      "learning_rate": 7.869869869869871e-06,
      "loss": 0.2962,
      "step": 21280
    },
    {
      "epoch": 63.93,
      "grad_norm": 14.038422584533691,
      "learning_rate": 7.86886886886887e-06,
      "loss": 0.3416,
      "step": 21290
    },
    {
      "epoch": 63.96,
      "grad_norm": 7.32805871963501,
      "learning_rate": 7.867867867867868e-06,
      "loss": 0.3419,
      "step": 21300
    },
    {
      "epoch": 63.99,
      "grad_norm": 8.779979705810547,
      "learning_rate": 7.866866866866869e-06,
      "loss": 0.3471,
      "step": 21310
    },
    {
      "epoch": 64.0,
      "eval_accuracy": 0.9289,
      "eval_loss": 0.27988043427467346,
      "eval_runtime": 29.7905,
      "eval_samples_per_second": 335.678,
      "eval_steps_per_second": 1.343,
      "step": 21312
    },
    {
      "epoch": 64.02,
      "grad_norm": 12.8053560256958,
      "learning_rate": 7.865865865865866e-06,
      "loss": 0.3297,
      "step": 21320
    },
    {
      "epoch": 64.05,
      "grad_norm": 156.92623901367188,
      "learning_rate": 7.864864864864866e-06,
      "loss": 0.3127,
      "step": 21330
    },
    {
      "epoch": 64.08,
      "grad_norm": 14.451507568359375,
      "learning_rate": 7.863863863863864e-06,
      "loss": 0.3685,
      "step": 21340
    },
    {
      "epoch": 64.11,
      "grad_norm": 5.955498218536377,
      "learning_rate": 7.862862862862863e-06,
      "loss": 0.2973,
      "step": 21350
    },
    {
      "epoch": 64.14,
      "grad_norm": 9.780427932739258,
      "learning_rate": 7.861861861861863e-06,
      "loss": 0.3495,
      "step": 21360
    },
    {
      "epoch": 64.17,
      "grad_norm": 9.540242195129395,
      "learning_rate": 7.860860860860862e-06,
      "loss": 0.3328,
      "step": 21370
    },
    {
      "epoch": 64.2,
      "grad_norm": 7.687532901763916,
      "learning_rate": 7.85985985985986e-06,
      "loss": 0.3378,
      "step": 21380
    },
    {
      "epoch": 64.23,
      "grad_norm": 9.998066902160645,
      "learning_rate": 7.858858858858859e-06,
      "loss": 0.3347,
      "step": 21390
    },
    {
      "epoch": 64.26,
      "grad_norm": 12.642586708068848,
      "learning_rate": 7.85785785785786e-06,
      "loss": 0.3299,
      "step": 21400
    },
    {
      "epoch": 64.29,
      "grad_norm": 14.062581062316895,
      "learning_rate": 7.856856856856856e-06,
      "loss": 0.2705,
      "step": 21410
    },
    {
      "epoch": 64.32,
      "grad_norm": 8.216482162475586,
      "learning_rate": 7.855855855855857e-06,
      "loss": 0.297,
      "step": 21420
    },
    {
      "epoch": 64.35,
      "grad_norm": 11.485599517822266,
      "learning_rate": 7.854854854854855e-06,
      "loss": 0.2803,
      "step": 21430
    },
    {
      "epoch": 64.38,
      "grad_norm": 14.045943260192871,
      "learning_rate": 7.853853853853854e-06,
      "loss": 0.3241,
      "step": 21440
    },
    {
      "epoch": 64.41,
      "grad_norm": 10.054695129394531,
      "learning_rate": 7.852852852852854e-06,
      "loss": 0.3495,
      "step": 21450
    },
    {
      "epoch": 64.44,
      "grad_norm": 9.065461158752441,
      "learning_rate": 7.851851851851853e-06,
      "loss": 0.3306,
      "step": 21460
    },
    {
      "epoch": 64.47,
      "grad_norm": 6.7041401863098145,
      "learning_rate": 7.850850850850851e-06,
      "loss": 0.3082,
      "step": 21470
    },
    {
      "epoch": 64.5,
      "grad_norm": 9.173362731933594,
      "learning_rate": 7.849849849849852e-06,
      "loss": 0.3013,
      "step": 21480
    },
    {
      "epoch": 64.53,
      "grad_norm": 6.9233808517456055,
      "learning_rate": 7.848848848848848e-06,
      "loss": 0.3073,
      "step": 21490
    },
    {
      "epoch": 64.56,
      "grad_norm": 7.436712741851807,
      "learning_rate": 7.847847847847849e-06,
      "loss": 0.3757,
      "step": 21500
    },
    {
      "epoch": 64.59,
      "grad_norm": 8.92714786529541,
      "learning_rate": 7.846846846846847e-06,
      "loss": 0.3401,
      "step": 21510
    },
    {
      "epoch": 64.62,
      "grad_norm": 11.673639297485352,
      "learning_rate": 7.845845845845846e-06,
      "loss": 0.3022,
      "step": 21520
    },
    {
      "epoch": 64.65,
      "grad_norm": 12.044401168823242,
      "learning_rate": 7.844844844844846e-06,
      "loss": 0.3602,
      "step": 21530
    },
    {
      "epoch": 64.68,
      "grad_norm": 7.265422821044922,
      "learning_rate": 7.843843843843845e-06,
      "loss": 0.3104,
      "step": 21540
    },
    {
      "epoch": 64.71,
      "grad_norm": 10.175910949707031,
      "learning_rate": 7.842842842842843e-06,
      "loss": 0.3418,
      "step": 21550
    },
    {
      "epoch": 64.74,
      "grad_norm": 14.005911827087402,
      "learning_rate": 7.841841841841844e-06,
      "loss": 0.2867,
      "step": 21560
    },
    {
      "epoch": 64.77,
      "grad_norm": 7.230674743652344,
      "learning_rate": 7.84084084084084e-06,
      "loss": 0.2927,
      "step": 21570
    },
    {
      "epoch": 64.8,
      "grad_norm": 13.946086883544922,
      "learning_rate": 7.839839839839841e-06,
      "loss": 0.3362,
      "step": 21580
    },
    {
      "epoch": 64.83,
      "grad_norm": 10.583385467529297,
      "learning_rate": 7.83883883883884e-06,
      "loss": 0.3275,
      "step": 21590
    },
    {
      "epoch": 64.86,
      "grad_norm": 11.054278373718262,
      "learning_rate": 7.837837837837838e-06,
      "loss": 0.3684,
      "step": 21600
    },
    {
      "epoch": 64.89,
      "grad_norm": 27.29981803894043,
      "learning_rate": 7.836836836836837e-06,
      "loss": 0.3579,
      "step": 21610
    },
    {
      "epoch": 64.92,
      "grad_norm": 9.092497825622559,
      "learning_rate": 7.835835835835837e-06,
      "loss": 0.3591,
      "step": 21620
    },
    {
      "epoch": 64.95,
      "grad_norm": 5.665831565856934,
      "learning_rate": 7.834834834834836e-06,
      "loss": 0.3234,
      "step": 21630
    },
    {
      "epoch": 64.98,
      "grad_norm": 39.293724060058594,
      "learning_rate": 7.833833833833834e-06,
      "loss": 0.2996,
      "step": 21640
    },
    {
      "epoch": 65.0,
      "eval_accuracy": 0.9259,
      "eval_loss": 0.28480955958366394,
      "eval_runtime": 30.2862,
      "eval_samples_per_second": 330.184,
      "eval_steps_per_second": 1.321,
      "step": 21645
    },
    {
      "epoch": 65.02,
      "grad_norm": 15.635560035705566,
      "learning_rate": 7.832832832832834e-06,
      "loss": 0.3412,
      "step": 21650
    },
    {
      "epoch": 65.05,
      "grad_norm": 7.564380645751953,
      "learning_rate": 7.831831831831831e-06,
      "loss": 0.2862,
      "step": 21660
    },
    {
      "epoch": 65.08,
      "grad_norm": 17.617725372314453,
      "learning_rate": 7.830830830830832e-06,
      "loss": 0.3009,
      "step": 21670
    },
    {
      "epoch": 65.11,
      "grad_norm": 5.260299205780029,
      "learning_rate": 7.82982982982983e-06,
      "loss": 0.3063,
      "step": 21680
    },
    {
      "epoch": 65.14,
      "grad_norm": 10.332259178161621,
      "learning_rate": 7.828828828828829e-06,
      "loss": 0.3213,
      "step": 21690
    },
    {
      "epoch": 65.17,
      "grad_norm": 19.753875732421875,
      "learning_rate": 7.827827827827829e-06,
      "loss": 0.3487,
      "step": 21700
    },
    {
      "epoch": 65.2,
      "grad_norm": 16.12872314453125,
      "learning_rate": 7.826826826826828e-06,
      "loss": 0.3321,
      "step": 21710
    },
    {
      "epoch": 65.23,
      "grad_norm": 14.514760971069336,
      "learning_rate": 7.825825825825826e-06,
      "loss": 0.3197,
      "step": 21720
    },
    {
      "epoch": 65.26,
      "grad_norm": 7.914402961730957,
      "learning_rate": 7.824824824824827e-06,
      "loss": 0.328,
      "step": 21730
    },
    {
      "epoch": 65.29,
      "grad_norm": 9.121522903442383,
      "learning_rate": 7.823823823823823e-06,
      "loss": 0.3203,
      "step": 21740
    },
    {
      "epoch": 65.32,
      "grad_norm": 12.540960311889648,
      "learning_rate": 7.822822822822824e-06,
      "loss": 0.3011,
      "step": 21750
    },
    {
      "epoch": 65.35,
      "grad_norm": 11.059178352355957,
      "learning_rate": 7.821821821821822e-06,
      "loss": 0.3773,
      "step": 21760
    },
    {
      "epoch": 65.38,
      "grad_norm": 7.422905445098877,
      "learning_rate": 7.820820820820821e-06,
      "loss": 0.2896,
      "step": 21770
    },
    {
      "epoch": 65.41,
      "grad_norm": 8.292637825012207,
      "learning_rate": 7.819819819819821e-06,
      "loss": 0.3112,
      "step": 21780
    },
    {
      "epoch": 65.44,
      "grad_norm": 6.934098720550537,
      "learning_rate": 7.81881881881882e-06,
      "loss": 0.2691,
      "step": 21790
    },
    {
      "epoch": 65.47,
      "grad_norm": 11.148361206054688,
      "learning_rate": 7.817817817817818e-06,
      "loss": 0.3007,
      "step": 21800
    },
    {
      "epoch": 65.5,
      "grad_norm": 9.407459259033203,
      "learning_rate": 7.816816816816819e-06,
      "loss": 0.316,
      "step": 21810
    },
    {
      "epoch": 65.53,
      "grad_norm": 8.743988990783691,
      "learning_rate": 7.815815815815816e-06,
      "loss": 0.3053,
      "step": 21820
    },
    {
      "epoch": 65.56,
      "grad_norm": 10.809152603149414,
      "learning_rate": 7.814814814814816e-06,
      "loss": 0.3172,
      "step": 21830
    },
    {
      "epoch": 65.59,
      "grad_norm": 5.7762908935546875,
      "learning_rate": 7.813813813813815e-06,
      "loss": 0.3437,
      "step": 21840
    },
    {
      "epoch": 65.62,
      "grad_norm": 7.504060745239258,
      "learning_rate": 7.812812812812813e-06,
      "loss": 0.3031,
      "step": 21850
    },
    {
      "epoch": 65.65,
      "grad_norm": 10.572630882263184,
      "learning_rate": 7.811811811811812e-06,
      "loss": 0.358,
      "step": 21860
    },
    {
      "epoch": 65.68,
      "grad_norm": 8.651265144348145,
      "learning_rate": 7.810810810810812e-06,
      "loss": 0.3191,
      "step": 21870
    },
    {
      "epoch": 65.71,
      "grad_norm": 7.8468217849731445,
      "learning_rate": 7.80980980980981e-06,
      "loss": 0.2866,
      "step": 21880
    },
    {
      "epoch": 65.74,
      "grad_norm": 6.5179033279418945,
      "learning_rate": 7.80880880880881e-06,
      "loss": 0.25,
      "step": 21890
    },
    {
      "epoch": 65.77,
      "grad_norm": 9.003945350646973,
      "learning_rate": 7.807807807807808e-06,
      "loss": 0.3377,
      "step": 21900
    },
    {
      "epoch": 65.8,
      "grad_norm": 10.532355308532715,
      "learning_rate": 7.806806806806806e-06,
      "loss": 0.3523,
      "step": 21910
    },
    {
      "epoch": 65.83,
      "grad_norm": 11.053046226501465,
      "learning_rate": 7.805805805805807e-06,
      "loss": 0.3022,
      "step": 21920
    },
    {
      "epoch": 65.86,
      "grad_norm": 14.968323707580566,
      "learning_rate": 7.804804804804805e-06,
      "loss": 0.3783,
      "step": 21930
    },
    {
      "epoch": 65.89,
      "grad_norm": 10.6726713180542,
      "learning_rate": 7.803803803803804e-06,
      "loss": 0.3424,
      "step": 21940
    },
    {
      "epoch": 65.92,
      "grad_norm": 7.953042030334473,
      "learning_rate": 7.802802802802804e-06,
      "loss": 0.3419,
      "step": 21950
    },
    {
      "epoch": 65.95,
      "grad_norm": 12.139482498168945,
      "learning_rate": 7.801801801801803e-06,
      "loss": 0.4105,
      "step": 21960
    },
    {
      "epoch": 65.98,
      "grad_norm": 8.202539443969727,
      "learning_rate": 7.800800800800801e-06,
      "loss": 0.3252,
      "step": 21970
    },
    {
      "epoch": 66.0,
      "eval_accuracy": 0.9278,
      "eval_loss": 0.2840452492237091,
      "eval_runtime": 30.1322,
      "eval_samples_per_second": 331.871,
      "eval_steps_per_second": 1.327,
      "step": 21978
    },
    {
      "epoch": 66.01,
      "grad_norm": 7.949310302734375,
      "learning_rate": 7.799799799799802e-06,
      "loss": 0.4774,
      "step": 21980
    },
    {
      "epoch": 66.04,
      "grad_norm": 18.162630081176758,
      "learning_rate": 7.798798798798799e-06,
      "loss": 0.2908,
      "step": 21990
    },
    {
      "epoch": 66.07,
      "grad_norm": 7.7490339279174805,
      "learning_rate": 7.797797797797799e-06,
      "loss": 0.3226,
      "step": 22000
    },
    {
      "epoch": 66.1,
      "grad_norm": 8.122990608215332,
      "learning_rate": 7.796796796796797e-06,
      "loss": 0.3387,
      "step": 22010
    },
    {
      "epoch": 66.13,
      "grad_norm": 11.342418670654297,
      "learning_rate": 7.795795795795796e-06,
      "loss": 0.3018,
      "step": 22020
    },
    {
      "epoch": 66.16,
      "grad_norm": 14.430131912231445,
      "learning_rate": 7.794794794794796e-06,
      "loss": 0.3003,
      "step": 22030
    },
    {
      "epoch": 66.19,
      "grad_norm": 11.513663291931152,
      "learning_rate": 7.793793793793795e-06,
      "loss": 0.3134,
      "step": 22040
    },
    {
      "epoch": 66.22,
      "grad_norm": 11.189915657043457,
      "learning_rate": 7.792792792792793e-06,
      "loss": 0.2775,
      "step": 22050
    },
    {
      "epoch": 66.25,
      "grad_norm": 11.753035545349121,
      "learning_rate": 7.791791791791792e-06,
      "loss": 0.3399,
      "step": 22060
    },
    {
      "epoch": 66.28,
      "grad_norm": 10.307710647583008,
      "learning_rate": 7.79079079079079e-06,
      "loss": 0.3507,
      "step": 22070
    },
    {
      "epoch": 66.31,
      "grad_norm": 7.679218769073486,
      "learning_rate": 7.78978978978979e-06,
      "loss": 0.3308,
      "step": 22080
    },
    {
      "epoch": 66.34,
      "grad_norm": 8.841164588928223,
      "learning_rate": 7.78878878878879e-06,
      "loss": 0.3097,
      "step": 22090
    },
    {
      "epoch": 66.37,
      "grad_norm": 14.498682022094727,
      "learning_rate": 7.787787787787788e-06,
      "loss": 0.353,
      "step": 22100
    },
    {
      "epoch": 66.4,
      "grad_norm": 6.857099533081055,
      "learning_rate": 7.786786786786787e-06,
      "loss": 0.3469,
      "step": 22110
    },
    {
      "epoch": 66.43,
      "grad_norm": 13.920244216918945,
      "learning_rate": 7.785785785785787e-06,
      "loss": 0.3353,
      "step": 22120
    },
    {
      "epoch": 66.46,
      "grad_norm": 11.716530799865723,
      "learning_rate": 7.784784784784786e-06,
      "loss": 0.3221,
      "step": 22130
    },
    {
      "epoch": 66.49,
      "grad_norm": 13.277667045593262,
      "learning_rate": 7.783783783783784e-06,
      "loss": 0.3235,
      "step": 22140
    },
    {
      "epoch": 66.52,
      "grad_norm": 8.703441619873047,
      "learning_rate": 7.782782782782783e-06,
      "loss": 0.3316,
      "step": 22150
    },
    {
      "epoch": 66.55,
      "grad_norm": 8.532262802124023,
      "learning_rate": 7.781781781781781e-06,
      "loss": 0.3228,
      "step": 22160
    },
    {
      "epoch": 66.58,
      "grad_norm": 13.708900451660156,
      "learning_rate": 7.780780780780782e-06,
      "loss": 0.3436,
      "step": 22170
    },
    {
      "epoch": 66.61,
      "grad_norm": 8.877799987792969,
      "learning_rate": 7.77977977977978e-06,
      "loss": 0.2836,
      "step": 22180
    },
    {
      "epoch": 66.64,
      "grad_norm": 11.738471031188965,
      "learning_rate": 7.778778778778779e-06,
      "loss": 0.344,
      "step": 22190
    },
    {
      "epoch": 66.67,
      "grad_norm": 11.911633491516113,
      "learning_rate": 7.77777777777778e-06,
      "loss": 0.3381,
      "step": 22200
    },
    {
      "epoch": 66.7,
      "grad_norm": 13.926767349243164,
      "learning_rate": 7.776776776776778e-06,
      "loss": 0.3401,
      "step": 22210
    },
    {
      "epoch": 66.73,
      "grad_norm": 14.48157787322998,
      "learning_rate": 7.775775775775776e-06,
      "loss": 0.3057,
      "step": 22220
    },
    {
      "epoch": 66.76,
      "grad_norm": 11.972227096557617,
      "learning_rate": 7.774774774774777e-06,
      "loss": 0.3271,
      "step": 22230
    },
    {
      "epoch": 66.79,
      "grad_norm": 8.929190635681152,
      "learning_rate": 7.773773773773774e-06,
      "loss": 0.3286,
      "step": 22240
    },
    {
      "epoch": 66.82,
      "grad_norm": 54.92183303833008,
      "learning_rate": 7.772772772772774e-06,
      "loss": 0.2744,
      "step": 22250
    },
    {
      "epoch": 66.85,
      "grad_norm": 12.29891586303711,
      "learning_rate": 7.771771771771772e-06,
      "loss": 0.3465,
      "step": 22260
    },
    {
      "epoch": 66.88,
      "grad_norm": 12.343019485473633,
      "learning_rate": 7.770770770770771e-06,
      "loss": 0.3362,
      "step": 22270
    },
    {
      "epoch": 66.91,
      "grad_norm": 12.407134056091309,
      "learning_rate": 7.769769769769771e-06,
      "loss": 0.2855,
      "step": 22280
    },
    {
      "epoch": 66.94,
      "grad_norm": 7.734322547912598,
      "learning_rate": 7.76876876876877e-06,
      "loss": 0.34,
      "step": 22290
    },
    {
      "epoch": 66.97,
      "grad_norm": 12.012907981872559,
      "learning_rate": 7.767767767767769e-06,
      "loss": 0.3306,
      "step": 22300
    },
    {
      "epoch": 67.0,
      "grad_norm": 9.052247047424316,
      "learning_rate": 7.766766766766767e-06,
      "loss": 0.324,
      "step": 22310
    },
    {
      "epoch": 67.0,
      "eval_accuracy": 0.9273,
      "eval_loss": 0.28527113795280457,
      "eval_runtime": 30.4273,
      "eval_samples_per_second": 328.653,
      "eval_steps_per_second": 1.315,
      "step": 22311
    },
    {
      "epoch": 67.03,
      "grad_norm": 16.09638786315918,
      "learning_rate": 7.765765765765766e-06,
      "loss": 0.3449,
      "step": 22320
    },
    {
      "epoch": 67.06,
      "grad_norm": 7.259049892425537,
      "learning_rate": 7.764764764764764e-06,
      "loss": 0.3113,
      "step": 22330
    },
    {
      "epoch": 67.09,
      "grad_norm": 9.85209846496582,
      "learning_rate": 7.763763763763765e-06,
      "loss": 0.3012,
      "step": 22340
    },
    {
      "epoch": 67.12,
      "grad_norm": 9.127687454223633,
      "learning_rate": 7.762762762762763e-06,
      "loss": 0.3184,
      "step": 22350
    },
    {
      "epoch": 67.15,
      "grad_norm": 8.984769821166992,
      "learning_rate": 7.761761761761762e-06,
      "loss": 0.3241,
      "step": 22360
    },
    {
      "epoch": 67.18,
      "grad_norm": 7.669935703277588,
      "learning_rate": 7.760760760760762e-06,
      "loss": 0.2745,
      "step": 22370
    },
    {
      "epoch": 67.21,
      "grad_norm": 9.03403091430664,
      "learning_rate": 7.75975975975976e-06,
      "loss": 0.3386,
      "step": 22380
    },
    {
      "epoch": 67.24,
      "grad_norm": 8.973670959472656,
      "learning_rate": 7.75875875875876e-06,
      "loss": 0.2947,
      "step": 22390
    },
    {
      "epoch": 67.27,
      "grad_norm": 10.151505470275879,
      "learning_rate": 7.757757757757758e-06,
      "loss": 0.3756,
      "step": 22400
    },
    {
      "epoch": 67.3,
      "grad_norm": 7.9546217918396,
      "learning_rate": 7.756756756756756e-06,
      "loss": 0.366,
      "step": 22410
    },
    {
      "epoch": 67.33,
      "grad_norm": 7.906595230102539,
      "learning_rate": 7.755755755755757e-06,
      "loss": 0.3087,
      "step": 22420
    },
    {
      "epoch": 67.36,
      "grad_norm": 7.04659366607666,
      "learning_rate": 7.754754754754755e-06,
      "loss": 0.2872,
      "step": 22430
    },
    {
      "epoch": 67.39,
      "grad_norm": 7.261047840118408,
      "learning_rate": 7.753753753753754e-06,
      "loss": 0.351,
      "step": 22440
    },
    {
      "epoch": 67.42,
      "grad_norm": 8.258650779724121,
      "learning_rate": 7.752752752752754e-06,
      "loss": 0.3098,
      "step": 22450
    },
    {
      "epoch": 67.45,
      "grad_norm": 9.970067977905273,
      "learning_rate": 7.751751751751753e-06,
      "loss": 0.3158,
      "step": 22460
    },
    {
      "epoch": 67.48,
      "grad_norm": 11.386726379394531,
      "learning_rate": 7.750750750750751e-06,
      "loss": 0.2727,
      "step": 22470
    },
    {
      "epoch": 67.51,
      "grad_norm": 9.31470012664795,
      "learning_rate": 7.749749749749752e-06,
      "loss": 0.3222,
      "step": 22480
    },
    {
      "epoch": 67.54,
      "grad_norm": 13.707615852355957,
      "learning_rate": 7.748748748748749e-06,
      "loss": 0.307,
      "step": 22490
    },
    {
      "epoch": 67.57,
      "grad_norm": 11.800713539123535,
      "learning_rate": 7.747747747747749e-06,
      "loss": 0.3363,
      "step": 22500
    },
    {
      "epoch": 67.6,
      "grad_norm": 8.65191650390625,
      "learning_rate": 7.746746746746747e-06,
      "loss": 0.3171,
      "step": 22510
    },
    {
      "epoch": 67.63,
      "grad_norm": 18.524646759033203,
      "learning_rate": 7.745745745745746e-06,
      "loss": 0.3513,
      "step": 22520
    },
    {
      "epoch": 67.66,
      "grad_norm": 8.623818397521973,
      "learning_rate": 7.744744744744745e-06,
      "loss": 0.3107,
      "step": 22530
    },
    {
      "epoch": 67.69,
      "grad_norm": 6.415104389190674,
      "learning_rate": 7.743743743743745e-06,
      "loss": 0.2806,
      "step": 22540
    },
    {
      "epoch": 67.72,
      "grad_norm": 9.560946464538574,
      "learning_rate": 7.742742742742744e-06,
      "loss": 0.294,
      "step": 22550
    },
    {
      "epoch": 67.75,
      "grad_norm": 6.53039026260376,
      "learning_rate": 7.741741741741742e-06,
      "loss": 0.2818,
      "step": 22560
    },
    {
      "epoch": 67.78,
      "grad_norm": 8.36352252960205,
      "learning_rate": 7.74074074074074e-06,
      "loss": 0.3035,
      "step": 22570
    },
    {
      "epoch": 67.81,
      "grad_norm": 158.1710968017578,
      "learning_rate": 7.73973973973974e-06,
      "loss": 0.2538,
      "step": 22580
    },
    {
      "epoch": 67.84,
      "grad_norm": 8.739215850830078,
      "learning_rate": 7.73873873873874e-06,
      "loss": 0.3494,
      "step": 22590
    },
    {
      "epoch": 67.87,
      "grad_norm": 21.067893981933594,
      "learning_rate": 7.737737737737738e-06,
      "loss": 0.3028,
      "step": 22600
    },
    {
      "epoch": 67.9,
      "grad_norm": 7.712640285491943,
      "learning_rate": 7.736736736736737e-06,
      "loss": 0.3188,
      "step": 22610
    },
    {
      "epoch": 67.93,
      "grad_norm": 12.430156707763672,
      "learning_rate": 7.735735735735737e-06,
      "loss": 0.3374,
      "step": 22620
    },
    {
      "epoch": 67.96,
      "grad_norm": 9.382542610168457,
      "learning_rate": 7.734734734734736e-06,
      "loss": 0.3372,
      "step": 22630
    },
    {
      "epoch": 67.99,
      "grad_norm": 5.900764465332031,
      "learning_rate": 7.733733733733734e-06,
      "loss": 0.2911,
      "step": 22640
    },
    {
      "epoch": 68.0,
      "eval_accuracy": 0.9282,
      "eval_loss": 0.283931702375412,
      "eval_runtime": 30.3349,
      "eval_samples_per_second": 329.653,
      "eval_steps_per_second": 1.319,
      "step": 22644
    },
    {
      "epoch": 68.02,
      "grad_norm": 11.490135192871094,
      "learning_rate": 7.732732732732733e-06,
      "loss": 0.3605,
      "step": 22650
    },
    {
      "epoch": 68.05,
      "grad_norm": 9.129350662231445,
      "learning_rate": 7.731731731731731e-06,
      "loss": 0.3383,
      "step": 22660
    },
    {
      "epoch": 68.08,
      "grad_norm": 9.362987518310547,
      "learning_rate": 7.730730730730732e-06,
      "loss": 0.3322,
      "step": 22670
    },
    {
      "epoch": 68.11,
      "grad_norm": 10.415196418762207,
      "learning_rate": 7.72972972972973e-06,
      "loss": 0.2981,
      "step": 22680
    },
    {
      "epoch": 68.14,
      "grad_norm": 13.375238418579102,
      "learning_rate": 7.728728728728729e-06,
      "loss": 0.331,
      "step": 22690
    },
    {
      "epoch": 68.17,
      "grad_norm": 10.638209342956543,
      "learning_rate": 7.72772772772773e-06,
      "loss": 0.322,
      "step": 22700
    },
    {
      "epoch": 68.2,
      "grad_norm": 9.923174858093262,
      "learning_rate": 7.726726726726728e-06,
      "loss": 0.3621,
      "step": 22710
    },
    {
      "epoch": 68.23,
      "grad_norm": 10.636186599731445,
      "learning_rate": 7.725725725725726e-06,
      "loss": 0.3567,
      "step": 22720
    },
    {
      "epoch": 68.26,
      "grad_norm": 10.163002014160156,
      "learning_rate": 7.724724724724727e-06,
      "loss": 0.3075,
      "step": 22730
    },
    {
      "epoch": 68.29,
      "grad_norm": 6.810419082641602,
      "learning_rate": 7.723723723723724e-06,
      "loss": 0.2865,
      "step": 22740
    },
    {
      "epoch": 68.32,
      "grad_norm": 8.398218154907227,
      "learning_rate": 7.722722722722722e-06,
      "loss": 0.3395,
      "step": 22750
    },
    {
      "epoch": 68.35,
      "grad_norm": 12.866475105285645,
      "learning_rate": 7.721721721721722e-06,
      "loss": 0.3238,
      "step": 22760
    },
    {
      "epoch": 68.38,
      "grad_norm": 7.318457126617432,
      "learning_rate": 7.720720720720721e-06,
      "loss": 0.3122,
      "step": 22770
    },
    {
      "epoch": 68.41,
      "grad_norm": 9.224184036254883,
      "learning_rate": 7.71971971971972e-06,
      "loss": 0.3078,
      "step": 22780
    },
    {
      "epoch": 68.44,
      "grad_norm": 14.799285888671875,
      "learning_rate": 7.71871871871872e-06,
      "loss": 0.3177,
      "step": 22790
    },
    {
      "epoch": 68.47,
      "grad_norm": 7.923614025115967,
      "learning_rate": 7.717717717717719e-06,
      "loss": 0.2565,
      "step": 22800
    },
    {
      "epoch": 68.5,
      "grad_norm": 9.54554557800293,
      "learning_rate": 7.716716716716717e-06,
      "loss": 0.3438,
      "step": 22810
    },
    {
      "epoch": 68.53,
      "grad_norm": 7.196744918823242,
      "learning_rate": 7.715715715715716e-06,
      "loss": 0.3334,
      "step": 22820
    },
    {
      "epoch": 68.56,
      "grad_norm": 11.80366325378418,
      "learning_rate": 7.714714714714714e-06,
      "loss": 0.3024,
      "step": 22830
    },
    {
      "epoch": 68.59,
      "grad_norm": 10.038961410522461,
      "learning_rate": 7.713713713713715e-06,
      "loss": 0.3322,
      "step": 22840
    },
    {
      "epoch": 68.62,
      "grad_norm": 8.976503372192383,
      "learning_rate": 7.712712712712713e-06,
      "loss": 0.3503,
      "step": 22850
    },
    {
      "epoch": 68.65,
      "grad_norm": 7.989677906036377,
      "learning_rate": 7.711711711711712e-06,
      "loss": 0.2937,
      "step": 22860
    },
    {
      "epoch": 68.68,
      "grad_norm": 7.2822418212890625,
      "learning_rate": 7.710710710710712e-06,
      "loss": 0.3161,
      "step": 22870
    },
    {
      "epoch": 68.71,
      "grad_norm": 8.396312713623047,
      "learning_rate": 7.70970970970971e-06,
      "loss": 0.3625,
      "step": 22880
    },
    {
      "epoch": 68.74,
      "grad_norm": 7.456044673919678,
      "learning_rate": 7.70870870870871e-06,
      "loss": 0.2842,
      "step": 22890
    },
    {
      "epoch": 68.77,
      "grad_norm": 25.769044876098633,
      "learning_rate": 7.707707707707708e-06,
      "loss": 0.2877,
      "step": 22900
    },
    {
      "epoch": 68.8,
      "grad_norm": 11.653054237365723,
      "learning_rate": 7.706706706706707e-06,
      "loss": 0.297,
      "step": 22910
    },
    {
      "epoch": 68.83,
      "grad_norm": 10.513693809509277,
      "learning_rate": 7.705705705705707e-06,
      "loss": 0.3004,
      "step": 22920
    },
    {
      "epoch": 68.86,
      "grad_norm": 13.775023460388184,
      "learning_rate": 7.704704704704705e-06,
      "loss": 0.2976,
      "step": 22930
    },
    {
      "epoch": 68.89,
      "grad_norm": 31.457298278808594,
      "learning_rate": 7.703703703703704e-06,
      "loss": 0.2878,
      "step": 22940
    },
    {
      "epoch": 68.92,
      "grad_norm": 16.45201301574707,
      "learning_rate": 7.702702702702704e-06,
      "loss": 0.3202,
      "step": 22950
    },
    {
      "epoch": 68.95,
      "grad_norm": 7.051560878753662,
      "learning_rate": 7.701701701701703e-06,
      "loss": 0.3054,
      "step": 22960
    },
    {
      "epoch": 68.98,
      "grad_norm": 96.16947937011719,
      "learning_rate": 7.700700700700701e-06,
      "loss": 0.2852,
      "step": 22970
    },
    {
      "epoch": 69.0,
      "eval_accuracy": 0.927,
      "eval_loss": 0.2891680598258972,
      "eval_runtime": 30.0731,
      "eval_samples_per_second": 332.523,
      "eval_steps_per_second": 1.33,
      "step": 22977
    },
    {
      "epoch": 69.01,
      "grad_norm": 12.446845054626465,
      "learning_rate": 7.6996996996997e-06,
      "loss": 0.333,
      "step": 22980
    },
    {
      "epoch": 69.04,
      "grad_norm": 9.202837944030762,
      "learning_rate": 7.698698698698699e-06,
      "loss": 0.2908,
      "step": 22990
    },
    {
      "epoch": 69.07,
      "grad_norm": 10.546953201293945,
      "learning_rate": 7.697697697697697e-06,
      "loss": 0.2948,
      "step": 23000
    },
    {
      "epoch": 69.1,
      "grad_norm": 8.322066307067871,
      "learning_rate": 7.696696696696698e-06,
      "loss": 0.358,
      "step": 23010
    },
    {
      "epoch": 69.13,
      "grad_norm": 15.954294204711914,
      "learning_rate": 7.695695695695696e-06,
      "loss": 0.2541,
      "step": 23020
    },
    {
      "epoch": 69.16,
      "grad_norm": 6.374436378479004,
      "learning_rate": 7.694694694694695e-06,
      "loss": 0.3657,
      "step": 23030
    },
    {
      "epoch": 69.19,
      "grad_norm": 9.781026840209961,
      "learning_rate": 7.693693693693695e-06,
      "loss": 0.294,
      "step": 23040
    },
    {
      "epoch": 69.22,
      "grad_norm": 11.463104248046875,
      "learning_rate": 7.692692692692694e-06,
      "loss": 0.3549,
      "step": 23050
    },
    {
      "epoch": 69.25,
      "grad_norm": 9.437586784362793,
      "learning_rate": 7.691691691691692e-06,
      "loss": 0.2775,
      "step": 23060
    },
    {
      "epoch": 69.28,
      "grad_norm": 22.58432960510254,
      "learning_rate": 7.69069069069069e-06,
      "loss": 0.3247,
      "step": 23070
    },
    {
      "epoch": 69.31,
      "grad_norm": 8.51115894317627,
      "learning_rate": 7.68968968968969e-06,
      "loss": 0.2741,
      "step": 23080
    },
    {
      "epoch": 69.34,
      "grad_norm": 9.827244758605957,
      "learning_rate": 7.68868868868869e-06,
      "loss": 0.3296,
      "step": 23090
    },
    {
      "epoch": 69.37,
      "grad_norm": 7.950046062469482,
      "learning_rate": 7.687687687687688e-06,
      "loss": 0.3142,
      "step": 23100
    },
    {
      "epoch": 69.4,
      "grad_norm": 14.976530075073242,
      "learning_rate": 7.686686686686687e-06,
      "loss": 0.3299,
      "step": 23110
    },
    {
      "epoch": 69.43,
      "grad_norm": 10.04582405090332,
      "learning_rate": 7.685685685685687e-06,
      "loss": 0.2868,
      "step": 23120
    },
    {
      "epoch": 69.46,
      "grad_norm": 9.546643257141113,
      "learning_rate": 7.684684684684686e-06,
      "loss": 0.2959,
      "step": 23130
    },
    {
      "epoch": 69.49,
      "grad_norm": 9.432767868041992,
      "learning_rate": 7.683683683683684e-06,
      "loss": 0.3142,
      "step": 23140
    },
    {
      "epoch": 69.52,
      "grad_norm": 6.99530553817749,
      "learning_rate": 7.682682682682683e-06,
      "loss": 0.291,
      "step": 23150
    },
    {
      "epoch": 69.55,
      "grad_norm": 18.65958595275879,
      "learning_rate": 7.681681681681682e-06,
      "loss": 0.2421,
      "step": 23160
    },
    {
      "epoch": 69.58,
      "grad_norm": 9.78608226776123,
      "learning_rate": 7.680680680680682e-06,
      "loss": 0.2915,
      "step": 23170
    },
    {
      "epoch": 69.61,
      "grad_norm": 10.383846282958984,
      "learning_rate": 7.67967967967968e-06,
      "loss": 0.3034,
      "step": 23180
    },
    {
      "epoch": 69.64,
      "grad_norm": 6.185490131378174,
      "learning_rate": 7.678678678678679e-06,
      "loss": 0.334,
      "step": 23190
    },
    {
      "epoch": 69.67,
      "grad_norm": 15.207490921020508,
      "learning_rate": 7.67767767767768e-06,
      "loss": 0.3055,
      "step": 23200
    },
    {
      "epoch": 69.7,
      "grad_norm": 14.461374282836914,
      "learning_rate": 7.676676676676678e-06,
      "loss": 0.3108,
      "step": 23210
    },
    {
      "epoch": 69.73,
      "grad_norm": 10.45607852935791,
      "learning_rate": 7.675675675675676e-06,
      "loss": 0.283,
      "step": 23220
    },
    {
      "epoch": 69.76,
      "grad_norm": 10.475542068481445,
      "learning_rate": 7.674674674674675e-06,
      "loss": 0.3077,
      "step": 23230
    },
    {
      "epoch": 69.79,
      "grad_norm": 6.531900405883789,
      "learning_rate": 7.673673673673674e-06,
      "loss": 0.3335,
      "step": 23240
    },
    {
      "epoch": 69.82,
      "grad_norm": 10.646883964538574,
      "learning_rate": 7.672672672672672e-06,
      "loss": 0.2849,
      "step": 23250
    },
    {
      "epoch": 69.85,
      "grad_norm": 11.44185733795166,
      "learning_rate": 7.671671671671673e-06,
      "loss": 0.3277,
      "step": 23260
    },
    {
      "epoch": 69.88,
      "grad_norm": 9.464559555053711,
      "learning_rate": 7.670670670670671e-06,
      "loss": 0.3074,
      "step": 23270
    },
    {
      "epoch": 69.91,
      "grad_norm": 6.203151226043701,
      "learning_rate": 7.66966966966967e-06,
      "loss": 0.3545,
      "step": 23280
    },
    {
      "epoch": 69.94,
      "grad_norm": 9.596717834472656,
      "learning_rate": 7.66866866866867e-06,
      "loss": 0.3227,
      "step": 23290
    },
    {
      "epoch": 69.97,
      "grad_norm": 9.337689399719238,
      "learning_rate": 7.667667667667669e-06,
      "loss": 0.332,
      "step": 23300
    },
    {
      "epoch": 70.0,
      "grad_norm": 59.963871002197266,
      "learning_rate": 7.666666666666667e-06,
      "loss": 0.3154,
      "step": 23310
    },
    {
      "epoch": 70.0,
      "eval_accuracy": 0.928,
      "eval_loss": 0.29016348719596863,
      "eval_runtime": 29.8981,
      "eval_samples_per_second": 334.469,
      "eval_steps_per_second": 1.338,
      "step": 23310
    },
    {
      "epoch": 70.03,
      "grad_norm": 16.945310592651367,
      "learning_rate": 7.665665665665666e-06,
      "loss": 0.3113,
      "step": 23320
    },
    {
      "epoch": 70.06,
      "grad_norm": 14.200393676757812,
      "learning_rate": 7.664664664664664e-06,
      "loss": 0.3013,
      "step": 23330
    },
    {
      "epoch": 70.09,
      "grad_norm": 6.8329668045043945,
      "learning_rate": 7.663663663663665e-06,
      "loss": 0.3112,
      "step": 23340
    },
    {
      "epoch": 70.12,
      "grad_norm": 12.809391975402832,
      "learning_rate": 7.662662662662663e-06,
      "loss": 0.2929,
      "step": 23350
    },
    {
      "epoch": 70.15,
      "grad_norm": 6.787331581115723,
      "learning_rate": 7.661661661661662e-06,
      "loss": 0.317,
      "step": 23360
    },
    {
      "epoch": 70.18,
      "grad_norm": 10.48849105834961,
      "learning_rate": 7.660660660660662e-06,
      "loss": 0.3283,
      "step": 23370
    },
    {
      "epoch": 70.21,
      "grad_norm": 8.80192756652832,
      "learning_rate": 7.65965965965966e-06,
      "loss": 0.2867,
      "step": 23380
    },
    {
      "epoch": 70.24,
      "grad_norm": 7.3339524269104,
      "learning_rate": 7.65865865865866e-06,
      "loss": 0.342,
      "step": 23390
    },
    {
      "epoch": 70.27,
      "grad_norm": 9.82497501373291,
      "learning_rate": 7.657657657657658e-06,
      "loss": 0.2985,
      "step": 23400
    },
    {
      "epoch": 70.3,
      "grad_norm": 8.856953620910645,
      "learning_rate": 7.656656656656657e-06,
      "loss": 0.3527,
      "step": 23410
    },
    {
      "epoch": 70.33,
      "grad_norm": 12.057849884033203,
      "learning_rate": 7.655655655655657e-06,
      "loss": 0.2752,
      "step": 23420
    },
    {
      "epoch": 70.36,
      "grad_norm": 7.011959552764893,
      "learning_rate": 7.654654654654655e-06,
      "loss": 0.2626,
      "step": 23430
    },
    {
      "epoch": 70.39,
      "grad_norm": 13.77779769897461,
      "learning_rate": 7.653653653653654e-06,
      "loss": 0.3095,
      "step": 23440
    },
    {
      "epoch": 70.42,
      "grad_norm": 9.560693740844727,
      "learning_rate": 7.652652652652653e-06,
      "loss": 0.3195,
      "step": 23450
    },
    {
      "epoch": 70.45,
      "grad_norm": 8.36783504486084,
      "learning_rate": 7.651651651651653e-06,
      "loss": 0.2851,
      "step": 23460
    },
    {
      "epoch": 70.48,
      "grad_norm": 9.114294052124023,
      "learning_rate": 7.650650650650652e-06,
      "loss": 0.3305,
      "step": 23470
    },
    {
      "epoch": 70.51,
      "grad_norm": 7.3281097412109375,
      "learning_rate": 7.64964964964965e-06,
      "loss": 0.3021,
      "step": 23480
    },
    {
      "epoch": 70.54,
      "grad_norm": 13.46298599243164,
      "learning_rate": 7.648648648648649e-06,
      "loss": 0.337,
      "step": 23490
    },
    {
      "epoch": 70.57,
      "grad_norm": 11.58102035522461,
      "learning_rate": 7.647647647647647e-06,
      "loss": 0.2894,
      "step": 23500
    },
    {
      "epoch": 70.6,
      "grad_norm": 6.916345119476318,
      "learning_rate": 7.646646646646648e-06,
      "loss": 0.3357,
      "step": 23510
    },
    {
      "epoch": 70.63,
      "grad_norm": 11.817749977111816,
      "learning_rate": 7.645645645645646e-06,
      "loss": 0.277,
      "step": 23520
    },
    {
      "epoch": 70.66,
      "grad_norm": 10.731739044189453,
      "learning_rate": 7.644644644644645e-06,
      "loss": 0.289,
      "step": 23530
    },
    {
      "epoch": 70.69,
      "grad_norm": 6.284928321838379,
      "learning_rate": 7.643643643643645e-06,
      "loss": 0.2959,
      "step": 23540
    },
    {
      "epoch": 70.72,
      "grad_norm": 9.111356735229492,
      "learning_rate": 7.642642642642644e-06,
      "loss": 0.329,
      "step": 23550
    },
    {
      "epoch": 70.75,
      "grad_norm": 7.945197582244873,
      "learning_rate": 7.641641641641642e-06,
      "loss": 0.2164,
      "step": 23560
    },
    {
      "epoch": 70.78,
      "grad_norm": 8.268195152282715,
      "learning_rate": 7.640640640640641e-06,
      "loss": 0.2847,
      "step": 23570
    },
    {
      "epoch": 70.81,
      "grad_norm": 7.129288196563721,
      "learning_rate": 7.63963963963964e-06,
      "loss": 0.3348,
      "step": 23580
    },
    {
      "epoch": 70.84,
      "grad_norm": 9.710803985595703,
      "learning_rate": 7.63863863863864e-06,
      "loss": 0.3604,
      "step": 23590
    },
    {
      "epoch": 70.87,
      "grad_norm": 10.092866897583008,
      "learning_rate": 7.637637637637638e-06,
      "loss": 0.2845,
      "step": 23600
    },
    {
      "epoch": 70.9,
      "grad_norm": 7.188078880310059,
      "learning_rate": 7.636636636636637e-06,
      "loss": 0.3056,
      "step": 23610
    },
    {
      "epoch": 70.93,
      "grad_norm": 12.947321891784668,
      "learning_rate": 7.635635635635637e-06,
      "loss": 0.3422,
      "step": 23620
    },
    {
      "epoch": 70.96,
      "grad_norm": 8.099897384643555,
      "learning_rate": 7.634634634634636e-06,
      "loss": 0.2784,
      "step": 23630
    },
    {
      "epoch": 70.99,
      "grad_norm": 6.496668815612793,
      "learning_rate": 7.633633633633634e-06,
      "loss": 0.3402,
      "step": 23640
    },
    {
      "epoch": 71.0,
      "eval_accuracy": 0.9294,
      "eval_loss": 0.2873501777648926,
      "eval_runtime": 30.6647,
      "eval_samples_per_second": 326.108,
      "eval_steps_per_second": 1.304,
      "step": 23643
    },
    {
      "epoch": 71.02,
      "grad_norm": 6.976884365081787,
      "learning_rate": 7.632632632632633e-06,
      "loss": 0.2771,
      "step": 23650
    },
    {
      "epoch": 71.05,
      "grad_norm": 10.155166625976562,
      "learning_rate": 7.631631631631632e-06,
      "loss": 0.3143,
      "step": 23660
    },
    {
      "epoch": 71.08,
      "grad_norm": 17.801342010498047,
      "learning_rate": 7.63063063063063e-06,
      "loss": 0.3116,
      "step": 23670
    },
    {
      "epoch": 71.11,
      "grad_norm": 13.330009460449219,
      "learning_rate": 7.62962962962963e-06,
      "loss": 0.3232,
      "step": 23680
    },
    {
      "epoch": 71.14,
      "grad_norm": 11.779132843017578,
      "learning_rate": 7.628628628628629e-06,
      "loss": 0.3373,
      "step": 23690
    },
    {
      "epoch": 71.17,
      "grad_norm": 8.482049942016602,
      "learning_rate": 7.6276276276276285e-06,
      "loss": 0.2931,
      "step": 23700
    },
    {
      "epoch": 71.2,
      "grad_norm": 13.195013046264648,
      "learning_rate": 7.626626626626628e-06,
      "loss": 0.2744,
      "step": 23710
    },
    {
      "epoch": 71.23,
      "grad_norm": 6.838131427764893,
      "learning_rate": 7.6256256256256266e-06,
      "loss": 0.2896,
      "step": 23720
    },
    {
      "epoch": 71.26,
      "grad_norm": 95.7518539428711,
      "learning_rate": 7.624624624624624e-06,
      "loss": 0.3062,
      "step": 23730
    },
    {
      "epoch": 71.29,
      "grad_norm": 18.472408294677734,
      "learning_rate": 7.623623623623624e-06,
      "loss": 0.3505,
      "step": 23740
    },
    {
      "epoch": 71.32,
      "grad_norm": 8.398428916931152,
      "learning_rate": 7.622622622622623e-06,
      "loss": 0.2555,
      "step": 23750
    },
    {
      "epoch": 71.35,
      "grad_norm": 8.12890625,
      "learning_rate": 7.621621621621622e-06,
      "loss": 0.29,
      "step": 23760
    },
    {
      "epoch": 71.38,
      "grad_norm": 7.042325973510742,
      "learning_rate": 7.620620620620621e-06,
      "loss": 0.3006,
      "step": 23770
    },
    {
      "epoch": 71.41,
      "grad_norm": 8.700090408325195,
      "learning_rate": 7.619619619619621e-06,
      "loss": 0.311,
      "step": 23780
    },
    {
      "epoch": 71.44,
      "grad_norm": 15.50074577331543,
      "learning_rate": 7.618618618618619e-06,
      "loss": 0.3163,
      "step": 23790
    },
    {
      "epoch": 71.47,
      "grad_norm": 66.91822814941406,
      "learning_rate": 7.617617617617619e-06,
      "loss": 0.2737,
      "step": 23800
    },
    {
      "epoch": 71.5,
      "grad_norm": 9.421163558959961,
      "learning_rate": 7.616616616616618e-06,
      "loss": 0.337,
      "step": 23810
    },
    {
      "epoch": 71.53,
      "grad_norm": 8.495651245117188,
      "learning_rate": 7.615615615615616e-06,
      "loss": 0.307,
      "step": 23820
    },
    {
      "epoch": 71.56,
      "grad_norm": 10.914982795715332,
      "learning_rate": 7.614614614614615e-06,
      "loss": 0.3491,
      "step": 23830
    },
    {
      "epoch": 71.59,
      "grad_norm": 7.34835958480835,
      "learning_rate": 7.613613613613614e-06,
      "loss": 0.3019,
      "step": 23840
    },
    {
      "epoch": 71.62,
      "grad_norm": 8.464154243469238,
      "learning_rate": 7.612612612612613e-06,
      "loss": 0.2774,
      "step": 23850
    },
    {
      "epoch": 71.65,
      "grad_norm": 6.718542098999023,
      "learning_rate": 7.611611611611612e-06,
      "loss": 0.2915,
      "step": 23860
    },
    {
      "epoch": 71.68,
      "grad_norm": 7.739274501800537,
      "learning_rate": 7.610610610610611e-06,
      "loss": 0.2952,
      "step": 23870
    },
    {
      "epoch": 71.71,
      "grad_norm": 12.87848949432373,
      "learning_rate": 7.609609609609611e-06,
      "loss": 0.3326,
      "step": 23880
    },
    {
      "epoch": 71.74,
      "grad_norm": 13.428434371948242,
      "learning_rate": 7.6086086086086095e-06,
      "loss": 0.2478,
      "step": 23890
    },
    {
      "epoch": 71.77,
      "grad_norm": 10.098626136779785,
      "learning_rate": 7.607607607607608e-06,
      "loss": 0.3365,
      "step": 23900
    },
    {
      "epoch": 71.8,
      "grad_norm": 14.810151100158691,
      "learning_rate": 7.606606606606607e-06,
      "loss": 0.3042,
      "step": 23910
    },
    {
      "epoch": 71.83,
      "grad_norm": 10.504941940307617,
      "learning_rate": 7.605605605605606e-06,
      "loss": 0.3536,
      "step": 23920
    },
    {
      "epoch": 71.86,
      "grad_norm": 10.261408805847168,
      "learning_rate": 7.6046046046046055e-06,
      "loss": 0.3242,
      "step": 23930
    },
    {
      "epoch": 71.89,
      "grad_norm": 8.839296340942383,
      "learning_rate": 7.603603603603604e-06,
      "loss": 0.258,
      "step": 23940
    },
    {
      "epoch": 71.92,
      "grad_norm": 14.264151573181152,
      "learning_rate": 7.6026026026026036e-06,
      "loss": 0.3235,
      "step": 23950
    },
    {
      "epoch": 71.95,
      "grad_norm": 64.74969482421875,
      "learning_rate": 7.601601601601602e-06,
      "loss": 0.2807,
      "step": 23960
    },
    {
      "epoch": 71.98,
      "grad_norm": 10.489749908447266,
      "learning_rate": 7.600600600600602e-06,
      "loss": 0.3122,
      "step": 23970
    },
    {
      "epoch": 72.0,
      "eval_accuracy": 0.9258,
      "eval_loss": 0.2908500134944916,
      "eval_runtime": 30.2312,
      "eval_samples_per_second": 330.784,
      "eval_steps_per_second": 1.323,
      "step": 23976
    },
    {
      "epoch": 72.01,
      "grad_norm": 9.434158325195312,
      "learning_rate": 7.599599599599599e-06,
      "loss": 0.3345,
      "step": 23980
    },
    {
      "epoch": 72.04,
      "grad_norm": 6.079000949859619,
      "learning_rate": 7.598598598598599e-06,
      "loss": 0.3184,
      "step": 23990
    },
    {
      "epoch": 72.07,
      "grad_norm": 10.942490577697754,
      "learning_rate": 7.597597597597598e-06,
      "loss": 0.2874,
      "step": 24000
    },
    {
      "epoch": 72.1,
      "grad_norm": 16.739900588989258,
      "learning_rate": 7.596596596596597e-06,
      "loss": 0.36,
      "step": 24010
    },
    {
      "epoch": 72.13,
      "grad_norm": 11.225438117980957,
      "learning_rate": 7.595595595595596e-06,
      "loss": 0.296,
      "step": 24020
    },
    {
      "epoch": 72.16,
      "grad_norm": 8.085643768310547,
      "learning_rate": 7.594594594594596e-06,
      "loss": 0.2942,
      "step": 24030
    },
    {
      "epoch": 72.19,
      "grad_norm": 7.428586959838867,
      "learning_rate": 7.593593593593594e-06,
      "loss": 0.3069,
      "step": 24040
    },
    {
      "epoch": 72.22,
      "grad_norm": 13.574721336364746,
      "learning_rate": 7.592592592592594e-06,
      "loss": 0.3155,
      "step": 24050
    },
    {
      "epoch": 72.25,
      "grad_norm": 5.690531253814697,
      "learning_rate": 7.591591591591592e-06,
      "loss": 0.314,
      "step": 24060
    },
    {
      "epoch": 72.28,
      "grad_norm": 12.788064002990723,
      "learning_rate": 7.590590590590591e-06,
      "loss": 0.2996,
      "step": 24070
    },
    {
      "epoch": 72.31,
      "grad_norm": 6.494003772735596,
      "learning_rate": 7.5895895895895895e-06,
      "loss": 0.2576,
      "step": 24080
    },
    {
      "epoch": 72.34,
      "grad_norm": 11.88971996307373,
      "learning_rate": 7.588588588588589e-06,
      "loss": 0.3183,
      "step": 24090
    },
    {
      "epoch": 72.37,
      "grad_norm": 8.038151741027832,
      "learning_rate": 7.587587587587588e-06,
      "loss": 0.2957,
      "step": 24100
    },
    {
      "epoch": 72.4,
      "grad_norm": 5.389163017272949,
      "learning_rate": 7.586586586586587e-06,
      "loss": 0.3394,
      "step": 24110
    },
    {
      "epoch": 72.43,
      "grad_norm": 12.468687057495117,
      "learning_rate": 7.5855855855855865e-06,
      "loss": 0.3588,
      "step": 24120
    },
    {
      "epoch": 72.46,
      "grad_norm": 21.71793556213379,
      "learning_rate": 7.584584584584586e-06,
      "loss": 0.3131,
      "step": 24130
    },
    {
      "epoch": 72.49,
      "grad_norm": 8.525282859802246,
      "learning_rate": 7.5835835835835845e-06,
      "loss": 0.3444,
      "step": 24140
    },
    {
      "epoch": 72.52,
      "grad_norm": 7.688137054443359,
      "learning_rate": 7.582582582582583e-06,
      "loss": 0.3375,
      "step": 24150
    },
    {
      "epoch": 72.55,
      "grad_norm": 8.914911270141602,
      "learning_rate": 7.581581581581582e-06,
      "loss": 0.3397,
      "step": 24160
    },
    {
      "epoch": 72.58,
      "grad_norm": 10.868552207946777,
      "learning_rate": 7.580580580580581e-06,
      "loss": 0.3246,
      "step": 24170
    },
    {
      "epoch": 72.61,
      "grad_norm": 5.609339237213135,
      "learning_rate": 7.57957957957958e-06,
      "loss": 0.3139,
      "step": 24180
    },
    {
      "epoch": 72.64,
      "grad_norm": 9.647193908691406,
      "learning_rate": 7.578578578578579e-06,
      "loss": 0.3397,
      "step": 24190
    },
    {
      "epoch": 72.67,
      "grad_norm": 14.739232063293457,
      "learning_rate": 7.577577577577579e-06,
      "loss": 0.3517,
      "step": 24200
    },
    {
      "epoch": 72.7,
      "grad_norm": 5.8103485107421875,
      "learning_rate": 7.576576576576577e-06,
      "loss": 0.2397,
      "step": 24210
    },
    {
      "epoch": 72.73,
      "grad_norm": 9.883508682250977,
      "learning_rate": 7.575575575575577e-06,
      "loss": 0.3596,
      "step": 24220
    },
    {
      "epoch": 72.76,
      "grad_norm": 10.005369186401367,
      "learning_rate": 7.574574574574574e-06,
      "loss": 0.3009,
      "step": 24230
    },
    {
      "epoch": 72.79,
      "grad_norm": 13.002354621887207,
      "learning_rate": 7.573573573573574e-06,
      "loss": 0.2861,
      "step": 24240
    },
    {
      "epoch": 72.82,
      "grad_norm": 6.862220764160156,
      "learning_rate": 7.572572572572573e-06,
      "loss": 0.3221,
      "step": 24250
    },
    {
      "epoch": 72.85,
      "grad_norm": 10.221587181091309,
      "learning_rate": 7.571571571571572e-06,
      "loss": 0.3136,
      "step": 24260
    },
    {
      "epoch": 72.88,
      "grad_norm": 7.82661771774292,
      "learning_rate": 7.570570570570571e-06,
      "loss": 0.3004,
      "step": 24270
    },
    {
      "epoch": 72.91,
      "grad_norm": 7.424114227294922,
      "learning_rate": 7.569569569569571e-06,
      "loss": 0.2931,
      "step": 24280
    },
    {
      "epoch": 72.94,
      "grad_norm": 8.394424438476562,
      "learning_rate": 7.568568568568569e-06,
      "loss": 0.3145,
      "step": 24290
    },
    {
      "epoch": 72.97,
      "grad_norm": 7.180873870849609,
      "learning_rate": 7.567567567567569e-06,
      "loss": 0.3276,
      "step": 24300
    },
    {
      "epoch": 73.0,
      "eval_accuracy": 0.9281,
      "eval_loss": 0.28558769822120667,
      "eval_runtime": 30.6766,
      "eval_samples_per_second": 325.981,
      "eval_steps_per_second": 1.304,
      "step": 24309
    },
    {
      "epoch": 73.0,
      "grad_norm": 7.631088733673096,
      "learning_rate": 7.566566566566567e-06,
      "loss": 0.28,
      "step": 24310
    },
    {
      "epoch": 73.03,
      "grad_norm": 6.390652179718018,
      "learning_rate": 7.565565565565566e-06,
      "loss": 0.33,
      "step": 24320
    },
    {
      "epoch": 73.06,
      "grad_norm": 35.52358627319336,
      "learning_rate": 7.5645645645645646e-06,
      "loss": 0.3676,
      "step": 24330
    },
    {
      "epoch": 73.09,
      "grad_norm": 8.733526229858398,
      "learning_rate": 7.563563563563564e-06,
      "loss": 0.3178,
      "step": 24340
    },
    {
      "epoch": 73.12,
      "grad_norm": 13.361793518066406,
      "learning_rate": 7.5625625625625634e-06,
      "loss": 0.2956,
      "step": 24350
    },
    {
      "epoch": 73.15,
      "grad_norm": 47.71128845214844,
      "learning_rate": 7.561561561561562e-06,
      "loss": 0.3461,
      "step": 24360
    },
    {
      "epoch": 73.18,
      "grad_norm": 7.483211994171143,
      "learning_rate": 7.5605605605605615e-06,
      "loss": 0.3303,
      "step": 24370
    },
    {
      "epoch": 73.21,
      "grad_norm": 7.917505741119385,
      "learning_rate": 7.559559559559561e-06,
      "loss": 0.3331,
      "step": 24380
    },
    {
      "epoch": 73.24,
      "grad_norm": 7.731630802154541,
      "learning_rate": 7.5585585585585595e-06,
      "loss": 0.304,
      "step": 24390
    },
    {
      "epoch": 73.27,
      "grad_norm": 11.890353202819824,
      "learning_rate": 7.557557557557558e-06,
      "loss": 0.3061,
      "step": 24400
    },
    {
      "epoch": 73.3,
      "grad_norm": 7.421016216278076,
      "learning_rate": 7.556556556556557e-06,
      "loss": 0.3152,
      "step": 24410
    },
    {
      "epoch": 73.33,
      "grad_norm": 17.061004638671875,
      "learning_rate": 7.555555555555556e-06,
      "loss": 0.3623,
      "step": 24420
    },
    {
      "epoch": 73.36,
      "grad_norm": 13.53083324432373,
      "learning_rate": 7.554554554554555e-06,
      "loss": 0.337,
      "step": 24430
    },
    {
      "epoch": 73.39,
      "grad_norm": 170.4421844482422,
      "learning_rate": 7.553553553553554e-06,
      "loss": 0.3224,
      "step": 24440
    },
    {
      "epoch": 73.42,
      "grad_norm": 9.768987655639648,
      "learning_rate": 7.552552552552554e-06,
      "loss": 0.3086,
      "step": 24450
    },
    {
      "epoch": 73.45,
      "grad_norm": 30.720521926879883,
      "learning_rate": 7.551551551551552e-06,
      "loss": 0.3338,
      "step": 24460
    },
    {
      "epoch": 73.48,
      "grad_norm": 10.686890602111816,
      "learning_rate": 7.550550550550552e-06,
      "loss": 0.3043,
      "step": 24470
    },
    {
      "epoch": 73.51,
      "grad_norm": 9.670889854431152,
      "learning_rate": 7.549549549549549e-06,
      "loss": 0.3512,
      "step": 24480
    },
    {
      "epoch": 73.54,
      "grad_norm": 7.99171781539917,
      "learning_rate": 7.548548548548549e-06,
      "loss": 0.2845,
      "step": 24490
    },
    {
      "epoch": 73.57,
      "grad_norm": 7.086174011230469,
      "learning_rate": 7.547547547547548e-06,
      "loss": 0.3254,
      "step": 24500
    },
    {
      "epoch": 73.6,
      "grad_norm": 8.895134925842285,
      "learning_rate": 7.546546546546547e-06,
      "loss": 0.308,
      "step": 24510
    },
    {
      "epoch": 73.63,
      "grad_norm": 11.880653381347656,
      "learning_rate": 7.545545545545546e-06,
      "loss": 0.2878,
      "step": 24520
    },
    {
      "epoch": 73.66,
      "grad_norm": 7.345247745513916,
      "learning_rate": 7.544544544544545e-06,
      "loss": 0.279,
      "step": 24530
    },
    {
      "epoch": 73.69,
      "grad_norm": 14.1832914352417,
      "learning_rate": 7.543543543543544e-06,
      "loss": 0.3103,
      "step": 24540
    },
    {
      "epoch": 73.72,
      "grad_norm": 35.291259765625,
      "learning_rate": 7.542542542542544e-06,
      "loss": 0.3339,
      "step": 24550
    },
    {
      "epoch": 73.75,
      "grad_norm": 30.576311111450195,
      "learning_rate": 7.5415415415415416e-06,
      "loss": 0.3141,
      "step": 24560
    },
    {
      "epoch": 73.78,
      "grad_norm": 6.640713214874268,
      "learning_rate": 7.540540540540541e-06,
      "loss": 0.3027,
      "step": 24570
    },
    {
      "epoch": 73.81,
      "grad_norm": 8.383827209472656,
      "learning_rate": 7.53953953953954e-06,
      "loss": 0.3029,
      "step": 24580
    },
    {
      "epoch": 73.84,
      "grad_norm": 6.757811546325684,
      "learning_rate": 7.538538538538539e-06,
      "loss": 0.3061,
      "step": 24590
    },
    {
      "epoch": 73.87,
      "grad_norm": 8.171782493591309,
      "learning_rate": 7.5375375375375385e-06,
      "loss": 0.3179,
      "step": 24600
    },
    {
      "epoch": 73.9,
      "grad_norm": 8.461021423339844,
      "learning_rate": 7.536536536536537e-06,
      "loss": 0.3193,
      "step": 24610
    },
    {
      "epoch": 73.93,
      "grad_norm": 16.660646438598633,
      "learning_rate": 7.5355355355355365e-06,
      "loss": 0.2777,
      "step": 24620
    },
    {
      "epoch": 73.96,
      "grad_norm": 6.314070224761963,
      "learning_rate": 7.534534534534535e-06,
      "loss": 0.2922,
      "step": 24630
    },
    {
      "epoch": 73.99,
      "grad_norm": 8.28592586517334,
      "learning_rate": 7.5335335335335346e-06,
      "loss": 0.291,
      "step": 24640
    },
    {
      "epoch": 74.0,
      "eval_accuracy": 0.9257,
      "eval_loss": 0.28950878977775574,
      "eval_runtime": 30.4174,
      "eval_samples_per_second": 328.76,
      "eval_steps_per_second": 1.315,
      "step": 24642
    },
    {
      "epoch": 74.02,
      "grad_norm": 8.10717487335205,
      "learning_rate": 7.532532532532532e-06,
      "loss": 0.2899,
      "step": 24650
    },
    {
      "epoch": 74.05,
      "grad_norm": 10.250568389892578,
      "learning_rate": 7.531531531531532e-06,
      "loss": 0.2958,
      "step": 24660
    },
    {
      "epoch": 74.08,
      "grad_norm": 11.427330017089844,
      "learning_rate": 7.530530530530531e-06,
      "loss": 0.3255,
      "step": 24670
    },
    {
      "epoch": 74.11,
      "grad_norm": 16.939380645751953,
      "learning_rate": 7.52952952952953e-06,
      "loss": 0.3037,
      "step": 24680
    },
    {
      "epoch": 74.14,
      "grad_norm": 11.018939018249512,
      "learning_rate": 7.528528528528529e-06,
      "loss": 0.3142,
      "step": 24690
    },
    {
      "epoch": 74.17,
      "grad_norm": 13.198030471801758,
      "learning_rate": 7.527527527527529e-06,
      "loss": 0.2886,
      "step": 24700
    },
    {
      "epoch": 74.2,
      "grad_norm": 6.478206157684326,
      "learning_rate": 7.526526526526527e-06,
      "loss": 0.3245,
      "step": 24710
    },
    {
      "epoch": 74.23,
      "grad_norm": 10.569625854492188,
      "learning_rate": 7.525525525525527e-06,
      "loss": 0.2863,
      "step": 24720
    },
    {
      "epoch": 74.26,
      "grad_norm": 21.147260665893555,
      "learning_rate": 7.5245245245245245e-06,
      "loss": 0.3214,
      "step": 24730
    },
    {
      "epoch": 74.29,
      "grad_norm": 18.914812088012695,
      "learning_rate": 7.523523523523524e-06,
      "loss": 0.3065,
      "step": 24740
    },
    {
      "epoch": 74.32,
      "grad_norm": 8.845747947692871,
      "learning_rate": 7.5225225225225225e-06,
      "loss": 0.3063,
      "step": 24750
    },
    {
      "epoch": 74.35,
      "grad_norm": 9.638672828674316,
      "learning_rate": 7.521521521521522e-06,
      "loss": 0.3009,
      "step": 24760
    },
    {
      "epoch": 74.38,
      "grad_norm": 6.345875263214111,
      "learning_rate": 7.520520520520521e-06,
      "loss": 0.3318,
      "step": 24770
    },
    {
      "epoch": 74.41,
      "grad_norm": 12.41404914855957,
      "learning_rate": 7.51951951951952e-06,
      "loss": 0.2917,
      "step": 24780
    },
    {
      "epoch": 74.44,
      "grad_norm": 11.763190269470215,
      "learning_rate": 7.518518518518519e-06,
      "loss": 0.3084,
      "step": 24790
    },
    {
      "epoch": 74.47,
      "grad_norm": 7.976477146148682,
      "learning_rate": 7.517517517517519e-06,
      "loss": 0.2857,
      "step": 24800
    },
    {
      "epoch": 74.5,
      "grad_norm": 10.18941879272461,
      "learning_rate": 7.516516516516517e-06,
      "loss": 0.3062,
      "step": 24810
    },
    {
      "epoch": 74.53,
      "grad_norm": 7.2084503173828125,
      "learning_rate": 7.515515515515516e-06,
      "loss": 0.3344,
      "step": 24820
    },
    {
      "epoch": 74.56,
      "grad_norm": 13.412360191345215,
      "learning_rate": 7.514514514514515e-06,
      "loss": 0.2855,
      "step": 24830
    },
    {
      "epoch": 74.59,
      "grad_norm": 7.036795139312744,
      "learning_rate": 7.513513513513514e-06,
      "loss": 0.3273,
      "step": 24840
    },
    {
      "epoch": 74.62,
      "grad_norm": 7.875713348388672,
      "learning_rate": 7.5125125125125135e-06,
      "loss": 0.2723,
      "step": 24850
    },
    {
      "epoch": 74.65,
      "grad_norm": 6.918676853179932,
      "learning_rate": 7.511511511511512e-06,
      "loss": 0.2745,
      "step": 24860
    },
    {
      "epoch": 74.68,
      "grad_norm": 15.264237403869629,
      "learning_rate": 7.5105105105105116e-06,
      "loss": 0.281,
      "step": 24870
    },
    {
      "epoch": 74.71,
      "grad_norm": 12.359001159667969,
      "learning_rate": 7.50950950950951e-06,
      "loss": 0.2722,
      "step": 24880
    },
    {
      "epoch": 74.74,
      "grad_norm": 9.498245239257812,
      "learning_rate": 7.50850850850851e-06,
      "loss": 0.3004,
      "step": 24890
    },
    {
      "epoch": 74.77,
      "grad_norm": 10.370760917663574,
      "learning_rate": 7.507507507507507e-06,
      "loss": 0.3207,
      "step": 24900
    },
    {
      "epoch": 74.8,
      "grad_norm": 11.406927108764648,
      "learning_rate": 7.506506506506507e-06,
      "loss": 0.3559,
      "step": 24910
    },
    {
      "epoch": 74.83,
      "grad_norm": 12.650437355041504,
      "learning_rate": 7.505505505505506e-06,
      "loss": 0.2912,
      "step": 24920
    },
    {
      "epoch": 74.86,
      "grad_norm": 62.67543029785156,
      "learning_rate": 7.504504504504505e-06,
      "loss": 0.3006,
      "step": 24930
    },
    {
      "epoch": 74.89,
      "grad_norm": 8.565555572509766,
      "learning_rate": 7.503503503503504e-06,
      "loss": 0.3263,
      "step": 24940
    },
    {
      "epoch": 74.92,
      "grad_norm": 10.021269798278809,
      "learning_rate": 7.502502502502504e-06,
      "loss": 0.2862,
      "step": 24950
    },
    {
      "epoch": 74.95,
      "grad_norm": 9.24077033996582,
      "learning_rate": 7.501501501501502e-06,
      "loss": 0.3183,
      "step": 24960
    },
    {
      "epoch": 74.98,
      "grad_norm": 11.272571563720703,
      "learning_rate": 7.500500500500502e-06,
      "loss": 0.2939,
      "step": 24970
    },
    {
      "epoch": 75.0,
      "eval_accuracy": 0.927,
      "eval_loss": 0.2922111749649048,
      "eval_runtime": 30.4587,
      "eval_samples_per_second": 328.313,
      "eval_steps_per_second": 1.313,
      "step": 24975
    },
    {
      "epoch": 75.02,
      "grad_norm": 11.191278457641602,
      "learning_rate": 7.4994994994994995e-06,
      "loss": 0.2549,
      "step": 24980
    },
    {
      "epoch": 75.05,
      "grad_norm": 8.267064094543457,
      "learning_rate": 7.498498498498499e-06,
      "loss": 0.2448,
      "step": 24990
    },
    {
      "epoch": 75.08,
      "grad_norm": 7.4436726570129395,
      "learning_rate": 7.4974974974974975e-06,
      "loss": 0.3123,
      "step": 25000
    },
    {
      "epoch": 75.11,
      "grad_norm": 21.02184295654297,
      "learning_rate": 7.496496496496497e-06,
      "loss": 0.3598,
      "step": 25010
    },
    {
      "epoch": 75.14,
      "grad_norm": 8.168778419494629,
      "learning_rate": 7.495495495495496e-06,
      "loss": 0.2811,
      "step": 25020
    },
    {
      "epoch": 75.17,
      "grad_norm": 10.992924690246582,
      "learning_rate": 7.494494494494495e-06,
      "loss": 0.3323,
      "step": 25030
    },
    {
      "epoch": 75.2,
      "grad_norm": 12.10815715789795,
      "learning_rate": 7.4934934934934944e-06,
      "loss": 0.239,
      "step": 25040
    },
    {
      "epoch": 75.23,
      "grad_norm": 7.463057041168213,
      "learning_rate": 7.492492492492494e-06,
      "loss": 0.3294,
      "step": 25050
    },
    {
      "epoch": 75.26,
      "grad_norm": 11.298584938049316,
      "learning_rate": 7.491491491491492e-06,
      "loss": 0.2763,
      "step": 25060
    },
    {
      "epoch": 75.29,
      "grad_norm": 10.89389705657959,
      "learning_rate": 7.490490490490491e-06,
      "loss": 0.3373,
      "step": 25070
    },
    {
      "epoch": 75.32,
      "grad_norm": 8.46941089630127,
      "learning_rate": 7.48948948948949e-06,
      "loss": 0.2906,
      "step": 25080
    },
    {
      "epoch": 75.35,
      "grad_norm": 9.361191749572754,
      "learning_rate": 7.488488488488489e-06,
      "loss": 0.3439,
      "step": 25090
    },
    {
      "epoch": 75.38,
      "grad_norm": 9.169535636901855,
      "learning_rate": 7.487487487487488e-06,
      "loss": 0.312,
      "step": 25100
    },
    {
      "epoch": 75.41,
      "grad_norm": 11.417469024658203,
      "learning_rate": 7.486486486486487e-06,
      "loss": 0.3318,
      "step": 25110
    },
    {
      "epoch": 75.44,
      "grad_norm": 6.8783955574035645,
      "learning_rate": 7.485485485485487e-06,
      "loss": 0.3306,
      "step": 25120
    },
    {
      "epoch": 75.47,
      "grad_norm": 10.386122703552246,
      "learning_rate": 7.484484484484485e-06,
      "loss": 0.3445,
      "step": 25130
    },
    {
      "epoch": 75.5,
      "grad_norm": 10.814770698547363,
      "learning_rate": 7.483483483483485e-06,
      "loss": 0.324,
      "step": 25140
    },
    {
      "epoch": 75.53,
      "grad_norm": 5.083248615264893,
      "learning_rate": 7.482482482482482e-06,
      "loss": 0.2973,
      "step": 25150
    },
    {
      "epoch": 75.56,
      "grad_norm": 13.891046524047852,
      "learning_rate": 7.481481481481482e-06,
      "loss": 0.2877,
      "step": 25160
    },
    {
      "epoch": 75.59,
      "grad_norm": 11.058889389038086,
      "learning_rate": 7.480480480480481e-06,
      "loss": 0.2984,
      "step": 25170
    },
    {
      "epoch": 75.62,
      "grad_norm": 12.379912376403809,
      "learning_rate": 7.47947947947948e-06,
      "loss": 0.3211,
      "step": 25180
    },
    {
      "epoch": 75.65,
      "grad_norm": 10.40090274810791,
      "learning_rate": 7.478478478478479e-06,
      "loss": 0.2864,
      "step": 25190
    },
    {
      "epoch": 75.68,
      "grad_norm": 10.47667407989502,
      "learning_rate": 7.477477477477479e-06,
      "loss": 0.3603,
      "step": 25200
    },
    {
      "epoch": 75.71,
      "grad_norm": 10.829870223999023,
      "learning_rate": 7.476476476476477e-06,
      "loss": 0.2803,
      "step": 25210
    },
    {
      "epoch": 75.74,
      "grad_norm": 9.343204498291016,
      "learning_rate": 7.475475475475477e-06,
      "loss": 0.2886,
      "step": 25220
    },
    {
      "epoch": 75.77,
      "grad_norm": 9.49301815032959,
      "learning_rate": 7.4744744744744745e-06,
      "loss": 0.3072,
      "step": 25230
    },
    {
      "epoch": 75.8,
      "grad_norm": 7.518646240234375,
      "learning_rate": 7.473473473473474e-06,
      "loss": 0.2859,
      "step": 25240
    },
    {
      "epoch": 75.83,
      "grad_norm": 11.801888465881348,
      "learning_rate": 7.4724724724724726e-06,
      "loss": 0.2976,
      "step": 25250
    },
    {
      "epoch": 75.86,
      "grad_norm": 15.316084861755371,
      "learning_rate": 7.471471471471472e-06,
      "loss": 0.3104,
      "step": 25260
    },
    {
      "epoch": 75.89,
      "grad_norm": 11.857620239257812,
      "learning_rate": 7.4704704704704714e-06,
      "loss": 0.3084,
      "step": 25270
    },
    {
      "epoch": 75.92,
      "grad_norm": 9.12366008758545,
      "learning_rate": 7.46946946946947e-06,
      "loss": 0.3606,
      "step": 25280
    },
    {
      "epoch": 75.95,
      "grad_norm": 11.10390567779541,
      "learning_rate": 7.4684684684684695e-06,
      "loss": 0.3243,
      "step": 25290
    },
    {
      "epoch": 75.98,
      "grad_norm": 7.584151744842529,
      "learning_rate": 7.467467467467469e-06,
      "loss": 0.2853,
      "step": 25300
    },
    {
      "epoch": 76.0,
      "eval_accuracy": 0.9245,
      "eval_loss": 0.295622318983078,
      "eval_runtime": 30.2852,
      "eval_samples_per_second": 330.194,
      "eval_steps_per_second": 1.321,
      "step": 25308
    },
    {
      "epoch": 76.01,
      "grad_norm": 7.311970233917236,
      "learning_rate": 7.466466466466467e-06,
      "loss": 0.2765,
      "step": 25310
    },
    {
      "epoch": 76.04,
      "grad_norm": 6.495667934417725,
      "learning_rate": 7.465465465465466e-06,
      "loss": 0.2875,
      "step": 25320
    },
    {
      "epoch": 76.07,
      "grad_norm": 12.85489273071289,
      "learning_rate": 7.464464464464465e-06,
      "loss": 0.3263,
      "step": 25330
    },
    {
      "epoch": 76.1,
      "grad_norm": 7.4302873611450195,
      "learning_rate": 7.463463463463464e-06,
      "loss": 0.3274,
      "step": 25340
    },
    {
      "epoch": 76.13,
      "grad_norm": 5.441634654998779,
      "learning_rate": 7.462462462462463e-06,
      "loss": 0.301,
      "step": 25350
    },
    {
      "epoch": 76.16,
      "grad_norm": 6.4101386070251465,
      "learning_rate": 7.461461461461462e-06,
      "loss": 0.2728,
      "step": 25360
    },
    {
      "epoch": 76.19,
      "grad_norm": 7.09028959274292,
      "learning_rate": 7.460460460460462e-06,
      "loss": 0.2886,
      "step": 25370
    },
    {
      "epoch": 76.22,
      "grad_norm": 6.305820465087891,
      "learning_rate": 7.45945945945946e-06,
      "loss": 0.3056,
      "step": 25380
    },
    {
      "epoch": 76.25,
      "grad_norm": 11.63720989227295,
      "learning_rate": 7.45845845845846e-06,
      "loss": 0.3382,
      "step": 25390
    },
    {
      "epoch": 76.28,
      "grad_norm": 10.77578353881836,
      "learning_rate": 7.457457457457457e-06,
      "loss": 0.3243,
      "step": 25400
    },
    {
      "epoch": 76.31,
      "grad_norm": 9.650285720825195,
      "learning_rate": 7.456456456456457e-06,
      "loss": 0.303,
      "step": 25410
    },
    {
      "epoch": 76.34,
      "grad_norm": 8.544416427612305,
      "learning_rate": 7.455455455455456e-06,
      "loss": 0.2858,
      "step": 25420
    },
    {
      "epoch": 76.37,
      "grad_norm": 14.733138084411621,
      "learning_rate": 7.454454454454455e-06,
      "loss": 0.3237,
      "step": 25430
    },
    {
      "epoch": 76.4,
      "grad_norm": 8.932827949523926,
      "learning_rate": 7.453453453453454e-06,
      "loss": 0.3088,
      "step": 25440
    },
    {
      "epoch": 76.43,
      "grad_norm": 16.72188949584961,
      "learning_rate": 7.452452452452453e-06,
      "loss": 0.2956,
      "step": 25450
    },
    {
      "epoch": 76.46,
      "grad_norm": 10.703816413879395,
      "learning_rate": 7.451451451451452e-06,
      "loss": 0.2556,
      "step": 25460
    },
    {
      "epoch": 76.49,
      "grad_norm": 8.552774429321289,
      "learning_rate": 7.450450450450452e-06,
      "loss": 0.3436,
      "step": 25470
    },
    {
      "epoch": 76.52,
      "grad_norm": 10.77054500579834,
      "learning_rate": 7.4494494494494496e-06,
      "loss": 0.3273,
      "step": 25480
    },
    {
      "epoch": 76.55,
      "grad_norm": 6.165707111358643,
      "learning_rate": 7.448448448448449e-06,
      "loss": 0.2926,
      "step": 25490
    },
    {
      "epoch": 76.58,
      "grad_norm": 8.086569786071777,
      "learning_rate": 7.447447447447448e-06,
      "loss": 0.3147,
      "step": 25500
    },
    {
      "epoch": 76.61,
      "grad_norm": 7.311812877655029,
      "learning_rate": 7.446446446446447e-06,
      "loss": 0.2596,
      "step": 25510
    },
    {
      "epoch": 76.64,
      "grad_norm": 7.7728705406188965,
      "learning_rate": 7.4454454454454465e-06,
      "loss": 0.267,
      "step": 25520
    },
    {
      "epoch": 76.67,
      "grad_norm": 9.495137214660645,
      "learning_rate": 7.444444444444445e-06,
      "loss": 0.285,
      "step": 25530
    },
    {
      "epoch": 76.7,
      "grad_norm": 11.292508125305176,
      "learning_rate": 7.4434434434434445e-06,
      "loss": 0.3028,
      "step": 25540
    },
    {
      "epoch": 76.73,
      "grad_norm": 8.03183650970459,
      "learning_rate": 7.442442442442443e-06,
      "loss": 0.3005,
      "step": 25550
    },
    {
      "epoch": 76.76,
      "grad_norm": 10.629243850708008,
      "learning_rate": 7.441441441441442e-06,
      "loss": 0.2979,
      "step": 25560
    },
    {
      "epoch": 76.79,
      "grad_norm": 17.501981735229492,
      "learning_rate": 7.44044044044044e-06,
      "loss": 0.2935,
      "step": 25570
    },
    {
      "epoch": 76.82,
      "grad_norm": 10.302696228027344,
      "learning_rate": 7.43943943943944e-06,
      "loss": 0.2885,
      "step": 25580
    },
    {
      "epoch": 76.85,
      "grad_norm": 8.07674503326416,
      "learning_rate": 7.438438438438439e-06,
      "loss": 0.3111,
      "step": 25590
    },
    {
      "epoch": 76.88,
      "grad_norm": 13.944626808166504,
      "learning_rate": 7.437437437437438e-06,
      "loss": 0.2994,
      "step": 25600
    },
    {
      "epoch": 76.91,
      "grad_norm": 10.235363006591797,
      "learning_rate": 7.436436436436437e-06,
      "loss": 0.3308,
      "step": 25610
    },
    {
      "epoch": 76.94,
      "grad_norm": 9.160021781921387,
      "learning_rate": 7.435435435435437e-06,
      "loss": 0.3474,
      "step": 25620
    },
    {
      "epoch": 76.97,
      "grad_norm": 12.820472717285156,
      "learning_rate": 7.434434434434435e-06,
      "loss": 0.2809,
      "step": 25630
    },
    {
      "epoch": 77.0,
      "grad_norm": 14.16905403137207,
      "learning_rate": 7.433433433433434e-06,
      "loss": 0.2863,
      "step": 25640
    },
    {
      "epoch": 77.0,
      "eval_accuracy": 0.9281,
      "eval_loss": 0.29000940918922424,
      "eval_runtime": 30.441,
      "eval_samples_per_second": 328.504,
      "eval_steps_per_second": 1.314,
      "step": 25641
    },
    {
      "epoch": 77.03,
      "grad_norm": 5.341837406158447,
      "learning_rate": 7.4324324324324324e-06,
      "loss": 0.2714,
      "step": 25650
    },
    {
      "epoch": 77.06,
      "grad_norm": 10.109475135803223,
      "learning_rate": 7.431431431431432e-06,
      "loss": 0.3309,
      "step": 25660
    },
    {
      "epoch": 77.09,
      "grad_norm": 8.118635177612305,
      "learning_rate": 7.4304304304304305e-06,
      "loss": 0.3287,
      "step": 25670
    },
    {
      "epoch": 77.12,
      "grad_norm": 9.928831100463867,
      "learning_rate": 7.42942942942943e-06,
      "loss": 0.3345,
      "step": 25680
    },
    {
      "epoch": 77.15,
      "grad_norm": 6.558286190032959,
      "learning_rate": 7.428428428428429e-06,
      "loss": 0.3085,
      "step": 25690
    },
    {
      "epoch": 77.18,
      "grad_norm": 10.067700386047363,
      "learning_rate": 7.427427427427428e-06,
      "loss": 0.2902,
      "step": 25700
    },
    {
      "epoch": 77.21,
      "grad_norm": 11.522305488586426,
      "learning_rate": 7.426426426426427e-06,
      "loss": 0.2839,
      "step": 25710
    },
    {
      "epoch": 77.24,
      "grad_norm": 9.111302375793457,
      "learning_rate": 7.425425425425427e-06,
      "loss": 0.288,
      "step": 25720
    },
    {
      "epoch": 77.27,
      "grad_norm": 4.338326454162598,
      "learning_rate": 7.424424424424425e-06,
      "loss": 0.2795,
      "step": 25730
    },
    {
      "epoch": 77.3,
      "grad_norm": 10.511844635009766,
      "learning_rate": 7.423423423423424e-06,
      "loss": 0.3351,
      "step": 25740
    },
    {
      "epoch": 77.33,
      "grad_norm": 9.067512512207031,
      "learning_rate": 7.422422422422423e-06,
      "loss": 0.2605,
      "step": 25750
    },
    {
      "epoch": 77.36,
      "grad_norm": 9.863814353942871,
      "learning_rate": 7.421421421421422e-06,
      "loss": 0.2514,
      "step": 25760
    },
    {
      "epoch": 77.39,
      "grad_norm": 10.875737190246582,
      "learning_rate": 7.4204204204204215e-06,
      "loss": 0.3045,
      "step": 25770
    },
    {
      "epoch": 77.42,
      "grad_norm": 7.592850685119629,
      "learning_rate": 7.41941941941942e-06,
      "loss": 0.2974,
      "step": 25780
    },
    {
      "epoch": 77.45,
      "grad_norm": 6.513065338134766,
      "learning_rate": 7.4184184184184195e-06,
      "loss": 0.289,
      "step": 25790
    },
    {
      "epoch": 77.48,
      "grad_norm": 12.718445777893066,
      "learning_rate": 7.417417417417418e-06,
      "loss": 0.3338,
      "step": 25800
    },
    {
      "epoch": 77.51,
      "grad_norm": 8.345598220825195,
      "learning_rate": 7.416416416416417e-06,
      "loss": 0.2544,
      "step": 25810
    },
    {
      "epoch": 77.54,
      "grad_norm": 9.903687477111816,
      "learning_rate": 7.415415415415415e-06,
      "loss": 0.2442,
      "step": 25820
    },
    {
      "epoch": 77.57,
      "grad_norm": 12.24393081665039,
      "learning_rate": 7.414414414414415e-06,
      "loss": 0.3051,
      "step": 25830
    },
    {
      "epoch": 77.6,
      "grad_norm": 6.779712677001953,
      "learning_rate": 7.413413413413414e-06,
      "loss": 0.3079,
      "step": 25840
    },
    {
      "epoch": 77.63,
      "grad_norm": 15.841277122497559,
      "learning_rate": 7.412412412412413e-06,
      "loss": 0.3055,
      "step": 25850
    },
    {
      "epoch": 77.66,
      "grad_norm": 8.456998825073242,
      "learning_rate": 7.411411411411412e-06,
      "loss": 0.3313,
      "step": 25860
    },
    {
      "epoch": 77.69,
      "grad_norm": 9.055587768554688,
      "learning_rate": 7.410410410410412e-06,
      "loss": 0.3288,
      "step": 25870
    },
    {
      "epoch": 77.72,
      "grad_norm": 13.7631254196167,
      "learning_rate": 7.40940940940941e-06,
      "loss": 0.2835,
      "step": 25880
    },
    {
      "epoch": 77.75,
      "grad_norm": 13.054932594299316,
      "learning_rate": 7.408408408408409e-06,
      "loss": 0.302,
      "step": 25890
    },
    {
      "epoch": 77.78,
      "grad_norm": 36.064979553222656,
      "learning_rate": 7.4074074074074075e-06,
      "loss": 0.2605,
      "step": 25900
    },
    {
      "epoch": 77.81,
      "grad_norm": 11.452004432678223,
      "learning_rate": 7.406406406406407e-06,
      "loss": 0.326,
      "step": 25910
    },
    {
      "epoch": 77.84,
      "grad_norm": 9.15996265411377,
      "learning_rate": 7.4054054054054055e-06,
      "loss": 0.2868,
      "step": 25920
    },
    {
      "epoch": 77.87,
      "grad_norm": 13.65489673614502,
      "learning_rate": 7.404404404404405e-06,
      "loss": 0.3124,
      "step": 25930
    },
    {
      "epoch": 77.9,
      "grad_norm": 9.829038619995117,
      "learning_rate": 7.403403403403404e-06,
      "loss": 0.2936,
      "step": 25940
    },
    {
      "epoch": 77.93,
      "grad_norm": 11.994723320007324,
      "learning_rate": 7.402402402402403e-06,
      "loss": 0.304,
      "step": 25950
    },
    {
      "epoch": 77.96,
      "grad_norm": 13.548896789550781,
      "learning_rate": 7.4014014014014024e-06,
      "loss": 0.2934,
      "step": 25960
    },
    {
      "epoch": 77.99,
      "grad_norm": 10.853764533996582,
      "learning_rate": 7.400400400400402e-06,
      "loss": 0.3417,
      "step": 25970
    },
    {
      "epoch": 78.0,
      "eval_accuracy": 0.927,
      "eval_loss": 0.29023274779319763,
      "eval_runtime": 29.9486,
      "eval_samples_per_second": 333.905,
      "eval_steps_per_second": 1.336,
      "step": 25974
    },
    {
      "epoch": 78.02,
      "grad_norm": 42.83219528198242,
      "learning_rate": 7.3993993993994e-06,
      "loss": 0.2715,
      "step": 25980
    },
    {
      "epoch": 78.05,
      "grad_norm": 7.009955883026123,
      "learning_rate": 7.398398398398399e-06,
      "loss": 0.3302,
      "step": 25990
    },
    {
      "epoch": 78.08,
      "grad_norm": 5.725772857666016,
      "learning_rate": 7.397397397397398e-06,
      "loss": 0.3218,
      "step": 26000
    },
    {
      "epoch": 78.11,
      "grad_norm": 9.869536399841309,
      "learning_rate": 7.396396396396397e-06,
      "loss": 0.279,
      "step": 26010
    },
    {
      "epoch": 78.14,
      "grad_norm": 7.667207717895508,
      "learning_rate": 7.395395395395396e-06,
      "loss": 0.2574,
      "step": 26020
    },
    {
      "epoch": 78.17,
      "grad_norm": 13.867158889770508,
      "learning_rate": 7.394394394394395e-06,
      "loss": 0.3134,
      "step": 26030
    },
    {
      "epoch": 78.2,
      "grad_norm": 10.050043106079102,
      "learning_rate": 7.393393393393395e-06,
      "loss": 0.2761,
      "step": 26040
    },
    {
      "epoch": 78.23,
      "grad_norm": 12.66734504699707,
      "learning_rate": 7.392392392392393e-06,
      "loss": 0.2565,
      "step": 26050
    },
    {
      "epoch": 78.26,
      "grad_norm": 6.713244438171387,
      "learning_rate": 7.391391391391392e-06,
      "loss": 0.3064,
      "step": 26060
    },
    {
      "epoch": 78.29,
      "grad_norm": 7.873671531677246,
      "learning_rate": 7.39039039039039e-06,
      "loss": 0.3086,
      "step": 26070
    },
    {
      "epoch": 78.32,
      "grad_norm": 11.539046287536621,
      "learning_rate": 7.38938938938939e-06,
      "loss": 0.3279,
      "step": 26080
    },
    {
      "epoch": 78.35,
      "grad_norm": 9.936339378356934,
      "learning_rate": 7.388388388388389e-06,
      "loss": 0.2906,
      "step": 26090
    },
    {
      "epoch": 78.38,
      "grad_norm": 11.362252235412598,
      "learning_rate": 7.387387387387388e-06,
      "loss": 0.2951,
      "step": 26100
    },
    {
      "epoch": 78.41,
      "grad_norm": 10.413354873657227,
      "learning_rate": 7.386386386386387e-06,
      "loss": 0.3328,
      "step": 26110
    },
    {
      "epoch": 78.44,
      "grad_norm": 6.827181339263916,
      "learning_rate": 7.385385385385386e-06,
      "loss": 0.2658,
      "step": 26120
    },
    {
      "epoch": 78.47,
      "grad_norm": 10.02496337890625,
      "learning_rate": 7.384384384384385e-06,
      "loss": 0.308,
      "step": 26130
    },
    {
      "epoch": 78.5,
      "grad_norm": 8.308564186096191,
      "learning_rate": 7.383383383383383e-06,
      "loss": 0.3363,
      "step": 26140
    },
    {
      "epoch": 78.53,
      "grad_norm": 9.091976165771484,
      "learning_rate": 7.3823823823823825e-06,
      "loss": 0.3294,
      "step": 26150
    },
    {
      "epoch": 78.56,
      "grad_norm": 6.880093574523926,
      "learning_rate": 7.381381381381382e-06,
      "loss": 0.3141,
      "step": 26160
    },
    {
      "epoch": 78.59,
      "grad_norm": 8.609142303466797,
      "learning_rate": 7.3803803803803806e-06,
      "loss": 0.2722,
      "step": 26170
    },
    {
      "epoch": 78.62,
      "grad_norm": 15.249781608581543,
      "learning_rate": 7.37937937937938e-06,
      "loss": 0.2984,
      "step": 26180
    },
    {
      "epoch": 78.65,
      "grad_norm": 6.6565775871276855,
      "learning_rate": 7.3783783783783794e-06,
      "loss": 0.2721,
      "step": 26190
    },
    {
      "epoch": 78.68,
      "grad_norm": 11.429220199584961,
      "learning_rate": 7.377377377377378e-06,
      "loss": 0.3229,
      "step": 26200
    },
    {
      "epoch": 78.71,
      "grad_norm": 8.404367446899414,
      "learning_rate": 7.3763763763763775e-06,
      "loss": 0.2892,
      "step": 26210
    },
    {
      "epoch": 78.74,
      "grad_norm": 8.5759916305542,
      "learning_rate": 7.375375375375377e-06,
      "loss": 0.319,
      "step": 26220
    },
    {
      "epoch": 78.77,
      "grad_norm": 14.003337860107422,
      "learning_rate": 7.374374374374375e-06,
      "loss": 0.3116,
      "step": 26230
    },
    {
      "epoch": 78.8,
      "grad_norm": 5.614511013031006,
      "learning_rate": 7.373373373373373e-06,
      "loss": 0.3266,
      "step": 26240
    },
    {
      "epoch": 78.83,
      "grad_norm": 20.751596450805664,
      "learning_rate": 7.372372372372373e-06,
      "loss": 0.3052,
      "step": 26250
    },
    {
      "epoch": 78.86,
      "grad_norm": 6.200272560119629,
      "learning_rate": 7.371371371371372e-06,
      "loss": 0.3126,
      "step": 26260
    },
    {
      "epoch": 78.89,
      "grad_norm": 9.790013313293457,
      "learning_rate": 7.370370370370371e-06,
      "loss": 0.2933,
      "step": 26270
    },
    {
      "epoch": 78.92,
      "grad_norm": 28.215110778808594,
      "learning_rate": 7.36936936936937e-06,
      "loss": 0.3296,
      "step": 26280
    },
    {
      "epoch": 78.95,
      "grad_norm": 12.338025093078613,
      "learning_rate": 7.36836836836837e-06,
      "loss": 0.3075,
      "step": 26290
    },
    {
      "epoch": 78.98,
      "grad_norm": 7.227609157562256,
      "learning_rate": 7.367367367367368e-06,
      "loss": 0.3053,
      "step": 26300
    },
    {
      "epoch": 79.0,
      "eval_accuracy": 0.9261,
      "eval_loss": 0.2924526333808899,
      "eval_runtime": 30.8184,
      "eval_samples_per_second": 324.482,
      "eval_steps_per_second": 1.298,
      "step": 26307
    },
    {
      "epoch": 79.01,
      "grad_norm": 9.645841598510742,
      "learning_rate": 7.366366366366367e-06,
      "loss": 0.2655,
      "step": 26310
    },
    {
      "epoch": 79.04,
      "grad_norm": 15.818194389343262,
      "learning_rate": 7.365365365365365e-06,
      "loss": 0.2884,
      "step": 26320
    },
    {
      "epoch": 79.07,
      "grad_norm": 16.222623825073242,
      "learning_rate": 7.364364364364365e-06,
      "loss": 0.3027,
      "step": 26330
    },
    {
      "epoch": 79.1,
      "grad_norm": 13.139893531799316,
      "learning_rate": 7.363363363363364e-06,
      "loss": 0.3154,
      "step": 26340
    },
    {
      "epoch": 79.13,
      "grad_norm": 5.564443111419678,
      "learning_rate": 7.362362362362363e-06,
      "loss": 0.2607,
      "step": 26350
    },
    {
      "epoch": 79.16,
      "grad_norm": 9.139181137084961,
      "learning_rate": 7.361361361361362e-06,
      "loss": 0.3491,
      "step": 26360
    },
    {
      "epoch": 79.19,
      "grad_norm": 10.221882820129395,
      "learning_rate": 7.360360360360361e-06,
      "loss": 0.2635,
      "step": 26370
    },
    {
      "epoch": 79.22,
      "grad_norm": 14.944685935974121,
      "learning_rate": 7.35935935935936e-06,
      "loss": 0.3088,
      "step": 26380
    },
    {
      "epoch": 79.25,
      "grad_norm": 13.538537979125977,
      "learning_rate": 7.358358358358358e-06,
      "loss": 0.2914,
      "step": 26390
    },
    {
      "epoch": 79.28,
      "grad_norm": 22.716920852661133,
      "learning_rate": 7.3573573573573575e-06,
      "loss": 0.3236,
      "step": 26400
    },
    {
      "epoch": 79.31,
      "grad_norm": 32.9311637878418,
      "learning_rate": 7.356356356356357e-06,
      "loss": 0.2895,
      "step": 26410
    },
    {
      "epoch": 79.34,
      "grad_norm": 9.955382347106934,
      "learning_rate": 7.355355355355356e-06,
      "loss": 0.3124,
      "step": 26420
    },
    {
      "epoch": 79.37,
      "grad_norm": 21.152822494506836,
      "learning_rate": 7.354354354354355e-06,
      "loss": 0.3098,
      "step": 26430
    },
    {
      "epoch": 79.4,
      "grad_norm": 6.504014015197754,
      "learning_rate": 7.3533533533533545e-06,
      "loss": 0.3155,
      "step": 26440
    },
    {
      "epoch": 79.43,
      "grad_norm": 14.5971040725708,
      "learning_rate": 7.352352352352353e-06,
      "loss": 0.2735,
      "step": 26450
    },
    {
      "epoch": 79.46,
      "grad_norm": 14.10580825805664,
      "learning_rate": 7.3513513513513525e-06,
      "loss": 0.2586,
      "step": 26460
    },
    {
      "epoch": 79.49,
      "grad_norm": 7.262406826019287,
      "learning_rate": 7.350350350350351e-06,
      "loss": 0.3322,
      "step": 26470
    },
    {
      "epoch": 79.52,
      "grad_norm": 8.412172317504883,
      "learning_rate": 7.34934934934935e-06,
      "loss": 0.2765,
      "step": 26480
    },
    {
      "epoch": 79.55,
      "grad_norm": 9.76326847076416,
      "learning_rate": 7.348348348348348e-06,
      "loss": 0.2897,
      "step": 26490
    },
    {
      "epoch": 79.58,
      "grad_norm": 12.168304443359375,
      "learning_rate": 7.347347347347348e-06,
      "loss": 0.2887,
      "step": 26500
    },
    {
      "epoch": 79.61,
      "grad_norm": 9.180903434753418,
      "learning_rate": 7.346346346346347e-06,
      "loss": 0.2949,
      "step": 26510
    },
    {
      "epoch": 79.64,
      "grad_norm": 9.365410804748535,
      "learning_rate": 7.345345345345346e-06,
      "loss": 0.3115,
      "step": 26520
    },
    {
      "epoch": 79.67,
      "grad_norm": 9.18213939666748,
      "learning_rate": 7.344344344344345e-06,
      "loss": 0.2885,
      "step": 26530
    },
    {
      "epoch": 79.7,
      "grad_norm": 10.00181770324707,
      "learning_rate": 7.343343343343345e-06,
      "loss": 0.318,
      "step": 26540
    },
    {
      "epoch": 79.73,
      "grad_norm": 30.755783081054688,
      "learning_rate": 7.342342342342343e-06,
      "loss": 0.3161,
      "step": 26550
    },
    {
      "epoch": 79.76,
      "grad_norm": 8.556061744689941,
      "learning_rate": 7.341341341341342e-06,
      "loss": 0.2791,
      "step": 26560
    },
    {
      "epoch": 79.79,
      "grad_norm": 7.7362380027771,
      "learning_rate": 7.3403403403403404e-06,
      "loss": 0.2681,
      "step": 26570
    },
    {
      "epoch": 79.82,
      "grad_norm": 8.998066902160645,
      "learning_rate": 7.33933933933934e-06,
      "loss": 0.3539,
      "step": 26580
    },
    {
      "epoch": 79.85,
      "grad_norm": 13.887422561645508,
      "learning_rate": 7.3383383383383385e-06,
      "loss": 0.3112,
      "step": 26590
    },
    {
      "epoch": 79.88,
      "grad_norm": 13.448328971862793,
      "learning_rate": 7.337337337337338e-06,
      "loss": 0.3377,
      "step": 26600
    },
    {
      "epoch": 79.91,
      "grad_norm": 11.029250144958496,
      "learning_rate": 7.336336336336337e-06,
      "loss": 0.2664,
      "step": 26610
    },
    {
      "epoch": 79.94,
      "grad_norm": 8.994755744934082,
      "learning_rate": 7.335335335335336e-06,
      "loss": 0.2752,
      "step": 26620
    },
    {
      "epoch": 79.97,
      "grad_norm": 8.945582389831543,
      "learning_rate": 7.334334334334335e-06,
      "loss": 0.309,
      "step": 26630
    },
    {
      "epoch": 80.0,
      "grad_norm": 22.66009521484375,
      "learning_rate": 7.333333333333333e-06,
      "loss": 0.2954,
      "step": 26640
    },
    {
      "epoch": 80.0,
      "eval_accuracy": 0.9257,
      "eval_loss": 0.29418402910232544,
      "eval_runtime": 30.2232,
      "eval_samples_per_second": 330.871,
      "eval_steps_per_second": 1.323,
      "step": 26640
    },
    {
      "epoch": 80.03,
      "grad_norm": 11.480300903320312,
      "learning_rate": 7.332332332332333e-06,
      "loss": 0.2882,
      "step": 26650
    },
    {
      "epoch": 80.06,
      "grad_norm": 8.611720085144043,
      "learning_rate": 7.331331331331332e-06,
      "loss": 0.2662,
      "step": 26660
    },
    {
      "epoch": 80.09,
      "grad_norm": 17.384756088256836,
      "learning_rate": 7.330330330330331e-06,
      "loss": 0.2815,
      "step": 26670
    },
    {
      "epoch": 80.12,
      "grad_norm": 12.247952461242676,
      "learning_rate": 7.32932932932933e-06,
      "loss": 0.3091,
      "step": 26680
    },
    {
      "epoch": 80.15,
      "grad_norm": 8.626712799072266,
      "learning_rate": 7.328328328328329e-06,
      "loss": 0.3079,
      "step": 26690
    },
    {
      "epoch": 80.18,
      "grad_norm": 8.431877136230469,
      "learning_rate": 7.327327327327328e-06,
      "loss": 0.2963,
      "step": 26700
    },
    {
      "epoch": 80.21,
      "grad_norm": 9.328495979309082,
      "learning_rate": 7.3263263263263275e-06,
      "loss": 0.3281,
      "step": 26710
    },
    {
      "epoch": 80.24,
      "grad_norm": 13.510149955749512,
      "learning_rate": 7.325325325325326e-06,
      "loss": 0.3013,
      "step": 26720
    },
    {
      "epoch": 80.27,
      "grad_norm": 12.874258995056152,
      "learning_rate": 7.324324324324325e-06,
      "loss": 0.3219,
      "step": 26730
    },
    {
      "epoch": 80.3,
      "grad_norm": 10.772368431091309,
      "learning_rate": 7.323323323323323e-06,
      "loss": 0.2842,
      "step": 26740
    },
    {
      "epoch": 80.33,
      "grad_norm": 6.161299705505371,
      "learning_rate": 7.322322322322323e-06,
      "loss": 0.2681,
      "step": 26750
    },
    {
      "epoch": 80.36,
      "grad_norm": 9.800402641296387,
      "learning_rate": 7.321321321321322e-06,
      "loss": 0.3108,
      "step": 26760
    },
    {
      "epoch": 80.39,
      "grad_norm": 7.039729595184326,
      "learning_rate": 7.320320320320321e-06,
      "loss": 0.2807,
      "step": 26770
    },
    {
      "epoch": 80.42,
      "grad_norm": 9.369364738464355,
      "learning_rate": 7.31931931931932e-06,
      "loss": 0.3236,
      "step": 26780
    },
    {
      "epoch": 80.45,
      "grad_norm": 17.48461151123047,
      "learning_rate": 7.31831831831832e-06,
      "loss": 0.3295,
      "step": 26790
    },
    {
      "epoch": 80.48,
      "grad_norm": 6.890653133392334,
      "learning_rate": 7.317317317317318e-06,
      "loss": 0.2827,
      "step": 26800
    },
    {
      "epoch": 80.51,
      "grad_norm": 8.72517204284668,
      "learning_rate": 7.316316316316316e-06,
      "loss": 0.2757,
      "step": 26810
    },
    {
      "epoch": 80.54,
      "grad_norm": 30.55694580078125,
      "learning_rate": 7.3153153153153155e-06,
      "loss": 0.3227,
      "step": 26820
    },
    {
      "epoch": 80.57,
      "grad_norm": 6.863607883453369,
      "learning_rate": 7.314314314314315e-06,
      "loss": 0.2959,
      "step": 26830
    },
    {
      "epoch": 80.6,
      "grad_norm": 8.139904022216797,
      "learning_rate": 7.3133133133133135e-06,
      "loss": 0.3331,
      "step": 26840
    },
    {
      "epoch": 80.63,
      "grad_norm": 7.438877105712891,
      "learning_rate": 7.312312312312313e-06,
      "loss": 0.2976,
      "step": 26850
    },
    {
      "epoch": 80.66,
      "grad_norm": 7.9227519035339355,
      "learning_rate": 7.311311311311312e-06,
      "loss": 0.3478,
      "step": 26860
    },
    {
      "epoch": 80.69,
      "grad_norm": 219.19346618652344,
      "learning_rate": 7.310310310310311e-06,
      "loss": 0.3145,
      "step": 26870
    },
    {
      "epoch": 80.72,
      "grad_norm": 8.627273559570312,
      "learning_rate": 7.3093093093093104e-06,
      "loss": 0.3296,
      "step": 26880
    },
    {
      "epoch": 80.75,
      "grad_norm": 6.543148517608643,
      "learning_rate": 7.308308308308308e-06,
      "loss": 0.2564,
      "step": 26890
    },
    {
      "epoch": 80.78,
      "grad_norm": 5.283452987670898,
      "learning_rate": 7.307307307307308e-06,
      "loss": 0.2995,
      "step": 26900
    },
    {
      "epoch": 80.81,
      "grad_norm": 9.285528182983398,
      "learning_rate": 7.306306306306307e-06,
      "loss": 0.2828,
      "step": 26910
    },
    {
      "epoch": 80.84,
      "grad_norm": 14.341130256652832,
      "learning_rate": 7.305305305305306e-06,
      "loss": 0.2763,
      "step": 26920
    },
    {
      "epoch": 80.87,
      "grad_norm": 13.928281784057617,
      "learning_rate": 7.304304304304305e-06,
      "loss": 0.2696,
      "step": 26930
    },
    {
      "epoch": 80.9,
      "grad_norm": 12.811273574829102,
      "learning_rate": 7.303303303303304e-06,
      "loss": 0.2921,
      "step": 26940
    },
    {
      "epoch": 80.93,
      "grad_norm": 9.974210739135742,
      "learning_rate": 7.302302302302303e-06,
      "loss": 0.298,
      "step": 26950
    },
    {
      "epoch": 80.96,
      "grad_norm": 12.99575424194336,
      "learning_rate": 7.3013013013013026e-06,
      "loss": 0.269,
      "step": 26960
    },
    {
      "epoch": 80.99,
      "grad_norm": 19.966533660888672,
      "learning_rate": 7.3003003003003e-06,
      "loss": 0.2539,
      "step": 26970
    },
    {
      "epoch": 81.0,
      "eval_accuracy": 0.9252,
      "eval_loss": 0.3003641963005066,
      "eval_runtime": 30.3982,
      "eval_samples_per_second": 328.967,
      "eval_steps_per_second": 1.316,
      "step": 26973
    },
    {
      "epoch": 81.02,
      "grad_norm": 10.901251792907715,
      "learning_rate": 7.2992992992993e-06,
      "loss": 0.303,
      "step": 26980
    },
    {
      "epoch": 81.05,
      "grad_norm": 8.914926528930664,
      "learning_rate": 7.298298298298298e-06,
      "loss": 0.2706,
      "step": 26990
    },
    {
      "epoch": 81.08,
      "grad_norm": 6.76442289352417,
      "learning_rate": 7.297297297297298e-06,
      "loss": 0.2717,
      "step": 27000
    },
    {
      "epoch": 81.11,
      "grad_norm": 37.34442901611328,
      "learning_rate": 7.296296296296297e-06,
      "loss": 0.3055,
      "step": 27010
    },
    {
      "epoch": 81.14,
      "grad_norm": 9.8020601272583,
      "learning_rate": 7.295295295295296e-06,
      "loss": 0.3093,
      "step": 27020
    },
    {
      "epoch": 81.17,
      "grad_norm": 12.030776023864746,
      "learning_rate": 7.294294294294295e-06,
      "loss": 0.3275,
      "step": 27030
    },
    {
      "epoch": 81.2,
      "grad_norm": 9.855073928833008,
      "learning_rate": 7.293293293293294e-06,
      "loss": 0.3076,
      "step": 27040
    },
    {
      "epoch": 81.23,
      "grad_norm": 17.103599548339844,
      "learning_rate": 7.292292292292293e-06,
      "loss": 0.3228,
      "step": 27050
    },
    {
      "epoch": 81.26,
      "grad_norm": 9.53991985321045,
      "learning_rate": 7.291291291291291e-06,
      "loss": 0.2907,
      "step": 27060
    },
    {
      "epoch": 81.29,
      "grad_norm": 8.926607131958008,
      "learning_rate": 7.2902902902902905e-06,
      "loss": 0.2802,
      "step": 27070
    },
    {
      "epoch": 81.32,
      "grad_norm": 6.32009220123291,
      "learning_rate": 7.28928928928929e-06,
      "loss": 0.2816,
      "step": 27080
    },
    {
      "epoch": 81.35,
      "grad_norm": 27.60784912109375,
      "learning_rate": 7.2882882882882885e-06,
      "loss": 0.2811,
      "step": 27090
    },
    {
      "epoch": 81.38,
      "grad_norm": 12.334404945373535,
      "learning_rate": 7.287287287287288e-06,
      "loss": 0.2672,
      "step": 27100
    },
    {
      "epoch": 81.41,
      "grad_norm": 8.628144264221191,
      "learning_rate": 7.2862862862862874e-06,
      "loss": 0.292,
      "step": 27110
    },
    {
      "epoch": 81.44,
      "grad_norm": 8.089194297790527,
      "learning_rate": 7.285285285285286e-06,
      "loss": 0.2894,
      "step": 27120
    },
    {
      "epoch": 81.47,
      "grad_norm": 11.088380813598633,
      "learning_rate": 7.2842842842842855e-06,
      "loss": 0.3135,
      "step": 27130
    },
    {
      "epoch": 81.5,
      "grad_norm": 8.3833646774292,
      "learning_rate": 7.283283283283283e-06,
      "loss": 0.2903,
      "step": 27140
    },
    {
      "epoch": 81.53,
      "grad_norm": 8.896328926086426,
      "learning_rate": 7.282282282282283e-06,
      "loss": 0.2707,
      "step": 27150
    },
    {
      "epoch": 81.56,
      "grad_norm": 7.986223220825195,
      "learning_rate": 7.281281281281281e-06,
      "loss": 0.2933,
      "step": 27160
    },
    {
      "epoch": 81.59,
      "grad_norm": 7.746074676513672,
      "learning_rate": 7.280280280280281e-06,
      "loss": 0.3154,
      "step": 27170
    },
    {
      "epoch": 81.62,
      "grad_norm": 8.172622680664062,
      "learning_rate": 7.27927927927928e-06,
      "loss": 0.3027,
      "step": 27180
    },
    {
      "epoch": 81.65,
      "grad_norm": 9.19118595123291,
      "learning_rate": 7.278278278278279e-06,
      "loss": 0.2637,
      "step": 27190
    },
    {
      "epoch": 81.68,
      "grad_norm": 11.13641357421875,
      "learning_rate": 7.277277277277278e-06,
      "loss": 0.3095,
      "step": 27200
    },
    {
      "epoch": 81.71,
      "grad_norm": 13.560955047607422,
      "learning_rate": 7.276276276276278e-06,
      "loss": 0.3153,
      "step": 27210
    },
    {
      "epoch": 81.74,
      "grad_norm": 6.896263122558594,
      "learning_rate": 7.275275275275275e-06,
      "loss": 0.257,
      "step": 27220
    },
    {
      "epoch": 81.77,
      "grad_norm": 6.891140460968018,
      "learning_rate": 7.274274274274275e-06,
      "loss": 0.3415,
      "step": 27230
    },
    {
      "epoch": 81.8,
      "grad_norm": 11.882068634033203,
      "learning_rate": 7.273273273273273e-06,
      "loss": 0.324,
      "step": 27240
    },
    {
      "epoch": 81.83,
      "grad_norm": 9.53926944732666,
      "learning_rate": 7.272272272272273e-06,
      "loss": 0.2901,
      "step": 27250
    },
    {
      "epoch": 81.86,
      "grad_norm": 16.090530395507812,
      "learning_rate": 7.271271271271272e-06,
      "loss": 0.2807,
      "step": 27260
    },
    {
      "epoch": 81.89,
      "grad_norm": 9.222565650939941,
      "learning_rate": 7.270270270270271e-06,
      "loss": 0.3417,
      "step": 27270
    },
    {
      "epoch": 81.92,
      "grad_norm": 8.862815856933594,
      "learning_rate": 7.26926926926927e-06,
      "loss": 0.2911,
      "step": 27280
    },
    {
      "epoch": 81.95,
      "grad_norm": 15.581215858459473,
      "learning_rate": 7.268268268268269e-06,
      "loss": 0.3326,
      "step": 27290
    },
    {
      "epoch": 81.98,
      "grad_norm": 14.999635696411133,
      "learning_rate": 7.267267267267268e-06,
      "loss": 0.2728,
      "step": 27300
    },
    {
      "epoch": 82.0,
      "eval_accuracy": 0.9252,
      "eval_loss": 0.29429349303245544,
      "eval_runtime": 30.0876,
      "eval_samples_per_second": 332.363,
      "eval_steps_per_second": 1.329,
      "step": 27306
    },
    {
      "epoch": 82.01,
      "grad_norm": 12.694634437561035,
      "learning_rate": 7.266266266266266e-06,
      "loss": 0.3409,
      "step": 27310
    },
    {
      "epoch": 82.04,
      "grad_norm": 7.2845072746276855,
      "learning_rate": 7.2652652652652655e-06,
      "loss": 0.2912,
      "step": 27320
    },
    {
      "epoch": 82.07,
      "grad_norm": 8.705058097839355,
      "learning_rate": 7.264264264264265e-06,
      "loss": 0.2552,
      "step": 27330
    },
    {
      "epoch": 82.1,
      "grad_norm": 15.071159362792969,
      "learning_rate": 7.263263263263264e-06,
      "loss": 0.3235,
      "step": 27340
    },
    {
      "epoch": 82.13,
      "grad_norm": 9.902262687683105,
      "learning_rate": 7.262262262262263e-06,
      "loss": 0.2938,
      "step": 27350
    },
    {
      "epoch": 82.16,
      "grad_norm": 12.302770614624023,
      "learning_rate": 7.2612612612612625e-06,
      "loss": 0.2948,
      "step": 27360
    },
    {
      "epoch": 82.19,
      "grad_norm": 10.296414375305176,
      "learning_rate": 7.260260260260261e-06,
      "loss": 0.2995,
      "step": 27370
    },
    {
      "epoch": 82.22,
      "grad_norm": 10.03219985961914,
      "learning_rate": 7.2592592592592605e-06,
      "loss": 0.2972,
      "step": 27380
    },
    {
      "epoch": 82.25,
      "grad_norm": 13.62802791595459,
      "learning_rate": 7.258258258258258e-06,
      "loss": 0.2855,
      "step": 27390
    },
    {
      "epoch": 82.28,
      "grad_norm": 6.880139350891113,
      "learning_rate": 7.257257257257258e-06,
      "loss": 0.3178,
      "step": 27400
    },
    {
      "epoch": 82.31,
      "grad_norm": 7.656834125518799,
      "learning_rate": 7.256256256256256e-06,
      "loss": 0.3254,
      "step": 27410
    },
    {
      "epoch": 82.34,
      "grad_norm": 6.245570182800293,
      "learning_rate": 7.255255255255256e-06,
      "loss": 0.3092,
      "step": 27420
    },
    {
      "epoch": 82.37,
      "grad_norm": 10.663616180419922,
      "learning_rate": 7.254254254254255e-06,
      "loss": 0.3236,
      "step": 27430
    },
    {
      "epoch": 82.4,
      "grad_norm": 6.805488109588623,
      "learning_rate": 7.253253253253254e-06,
      "loss": 0.3438,
      "step": 27440
    },
    {
      "epoch": 82.43,
      "grad_norm": 6.780258655548096,
      "learning_rate": 7.252252252252253e-06,
      "loss": 0.2936,
      "step": 27450
    },
    {
      "epoch": 82.46,
      "grad_norm": 9.752508163452148,
      "learning_rate": 7.251251251251253e-06,
      "loss": 0.2631,
      "step": 27460
    },
    {
      "epoch": 82.49,
      "grad_norm": 14.624298095703125,
      "learning_rate": 7.25025025025025e-06,
      "loss": 0.2319,
      "step": 27470
    },
    {
      "epoch": 82.52,
      "grad_norm": 17.38148307800293,
      "learning_rate": 7.24924924924925e-06,
      "loss": 0.271,
      "step": 27480
    },
    {
      "epoch": 82.55,
      "grad_norm": 38.592655181884766,
      "learning_rate": 7.2482482482482484e-06,
      "loss": 0.3486,
      "step": 27490
    },
    {
      "epoch": 82.58,
      "grad_norm": 7.233981132507324,
      "learning_rate": 7.247247247247248e-06,
      "loss": 0.3377,
      "step": 27500
    },
    {
      "epoch": 82.61,
      "grad_norm": 21.241090774536133,
      "learning_rate": 7.2462462462462465e-06,
      "loss": 0.3013,
      "step": 27510
    },
    {
      "epoch": 82.64,
      "grad_norm": 8.39799976348877,
      "learning_rate": 7.245245245245246e-06,
      "loss": 0.34,
      "step": 27520
    },
    {
      "epoch": 82.67,
      "grad_norm": 6.944133281707764,
      "learning_rate": 7.244244244244245e-06,
      "loss": 0.3078,
      "step": 27530
    },
    {
      "epoch": 82.7,
      "grad_norm": 10.005699157714844,
      "learning_rate": 7.243243243243244e-06,
      "loss": 0.2986,
      "step": 27540
    },
    {
      "epoch": 82.73,
      "grad_norm": 8.388346672058105,
      "learning_rate": 7.242242242242243e-06,
      "loss": 0.3151,
      "step": 27550
    },
    {
      "epoch": 82.76,
      "grad_norm": 7.2757744789123535,
      "learning_rate": 7.241241241241241e-06,
      "loss": 0.3173,
      "step": 27560
    },
    {
      "epoch": 82.79,
      "grad_norm": 9.244422912597656,
      "learning_rate": 7.240240240240241e-06,
      "loss": 0.3376,
      "step": 27570
    },
    {
      "epoch": 82.82,
      "grad_norm": 27.179750442504883,
      "learning_rate": 7.23923923923924e-06,
      "loss": 0.2995,
      "step": 27580
    },
    {
      "epoch": 82.85,
      "grad_norm": 6.613595485687256,
      "learning_rate": 7.238238238238239e-06,
      "loss": 0.3402,
      "step": 27590
    },
    {
      "epoch": 82.88,
      "grad_norm": 7.379167079925537,
      "learning_rate": 7.237237237237238e-06,
      "loss": 0.2744,
      "step": 27600
    },
    {
      "epoch": 82.91,
      "grad_norm": 10.301562309265137,
      "learning_rate": 7.236236236236237e-06,
      "loss": 0.2793,
      "step": 27610
    },
    {
      "epoch": 82.94,
      "grad_norm": 6.335416793823242,
      "learning_rate": 7.235235235235236e-06,
      "loss": 0.2806,
      "step": 27620
    },
    {
      "epoch": 82.97,
      "grad_norm": 8.69294548034668,
      "learning_rate": 7.2342342342342355e-06,
      "loss": 0.3537,
      "step": 27630
    },
    {
      "epoch": 83.0,
      "eval_accuracy": 0.9243,
      "eval_loss": 0.3001709580421448,
      "eval_runtime": 30.3907,
      "eval_samples_per_second": 329.048,
      "eval_steps_per_second": 1.316,
      "step": 27639
    },
    {
      "epoch": 83.0,
      "grad_norm": 7.845828056335449,
      "learning_rate": 7.233233233233233e-06,
      "loss": 0.2242,
      "step": 27640
    },
    {
      "epoch": 83.03,
      "grad_norm": 9.509461402893066,
      "learning_rate": 7.232232232232233e-06,
      "loss": 0.329,
      "step": 27650
    },
    {
      "epoch": 83.06,
      "grad_norm": 108.83116149902344,
      "learning_rate": 7.231231231231231e-06,
      "loss": 0.262,
      "step": 27660
    },
    {
      "epoch": 83.09,
      "grad_norm": 8.102056503295898,
      "learning_rate": 7.230230230230231e-06,
      "loss": 0.2735,
      "step": 27670
    },
    {
      "epoch": 83.12,
      "grad_norm": 10.87539291381836,
      "learning_rate": 7.22922922922923e-06,
      "loss": 0.3018,
      "step": 27680
    },
    {
      "epoch": 83.15,
      "grad_norm": 11.96666145324707,
      "learning_rate": 7.228228228228229e-06,
      "loss": 0.2851,
      "step": 27690
    },
    {
      "epoch": 83.18,
      "grad_norm": 13.283472061157227,
      "learning_rate": 7.227227227227228e-06,
      "loss": 0.2911,
      "step": 27700
    },
    {
      "epoch": 83.21,
      "grad_norm": 9.27159309387207,
      "learning_rate": 7.226226226226228e-06,
      "loss": 0.3112,
      "step": 27710
    },
    {
      "epoch": 83.24,
      "grad_norm": 7.854355812072754,
      "learning_rate": 7.2252252252252254e-06,
      "loss": 0.2814,
      "step": 27720
    },
    {
      "epoch": 83.27,
      "grad_norm": 15.956609725952148,
      "learning_rate": 7.224224224224224e-06,
      "loss": 0.3017,
      "step": 27730
    },
    {
      "epoch": 83.3,
      "grad_norm": 14.1936674118042,
      "learning_rate": 7.2232232232232235e-06,
      "loss": 0.3029,
      "step": 27740
    },
    {
      "epoch": 83.33,
      "grad_norm": 5.382256507873535,
      "learning_rate": 7.222222222222223e-06,
      "loss": 0.309,
      "step": 27750
    },
    {
      "epoch": 83.36,
      "grad_norm": 6.515827655792236,
      "learning_rate": 7.2212212212212215e-06,
      "loss": 0.2969,
      "step": 27760
    },
    {
      "epoch": 83.39,
      "grad_norm": 6.132240295410156,
      "learning_rate": 7.220220220220221e-06,
      "loss": 0.3142,
      "step": 27770
    },
    {
      "epoch": 83.42,
      "grad_norm": 11.70675277709961,
      "learning_rate": 7.21921921921922e-06,
      "loss": 0.2941,
      "step": 27780
    },
    {
      "epoch": 83.45,
      "grad_norm": 11.140359878540039,
      "learning_rate": 7.218218218218219e-06,
      "loss": 0.2671,
      "step": 27790
    },
    {
      "epoch": 83.48,
      "grad_norm": 11.692485809326172,
      "learning_rate": 7.217217217217218e-06,
      "loss": 0.2961,
      "step": 27800
    },
    {
      "epoch": 83.51,
      "grad_norm": 8.353963851928711,
      "learning_rate": 7.216216216216216e-06,
      "loss": 0.289,
      "step": 27810
    },
    {
      "epoch": 83.54,
      "grad_norm": 5.555064678192139,
      "learning_rate": 7.215215215215216e-06,
      "loss": 0.2744,
      "step": 27820
    },
    {
      "epoch": 83.57,
      "grad_norm": 17.33306121826172,
      "learning_rate": 7.214214214214215e-06,
      "loss": 0.3314,
      "step": 27830
    },
    {
      "epoch": 83.6,
      "grad_norm": 6.67274284362793,
      "learning_rate": 7.213213213213214e-06,
      "loss": 0.2728,
      "step": 27840
    },
    {
      "epoch": 83.63,
      "grad_norm": 12.355216026306152,
      "learning_rate": 7.212212212212213e-06,
      "loss": 0.2659,
      "step": 27850
    },
    {
      "epoch": 83.66,
      "grad_norm": 9.236814498901367,
      "learning_rate": 7.211211211211212e-06,
      "loss": 0.2984,
      "step": 27860
    },
    {
      "epoch": 83.69,
      "grad_norm": 7.480185031890869,
      "learning_rate": 7.210210210210211e-06,
      "loss": 0.2671,
      "step": 27870
    },
    {
      "epoch": 83.72,
      "grad_norm": 7.809084415435791,
      "learning_rate": 7.2092092092092106e-06,
      "loss": 0.2719,
      "step": 27880
    },
    {
      "epoch": 83.75,
      "grad_norm": 14.48155689239502,
      "learning_rate": 7.208208208208208e-06,
      "loss": 0.3126,
      "step": 27890
    },
    {
      "epoch": 83.78,
      "grad_norm": 12.51624584197998,
      "learning_rate": 7.207207207207208e-06,
      "loss": 0.2776,
      "step": 27900
    },
    {
      "epoch": 83.81,
      "grad_norm": 13.532564163208008,
      "learning_rate": 7.206206206206206e-06,
      "loss": 0.3099,
      "step": 27910
    },
    {
      "epoch": 83.84,
      "grad_norm": 11.693519592285156,
      "learning_rate": 7.205205205205206e-06,
      "loss": 0.3032,
      "step": 27920
    },
    {
      "epoch": 83.87,
      "grad_norm": 8.313236236572266,
      "learning_rate": 7.204204204204205e-06,
      "loss": 0.2837,
      "step": 27930
    },
    {
      "epoch": 83.9,
      "grad_norm": 6.393245220184326,
      "learning_rate": 7.203203203203204e-06,
      "loss": 0.3012,
      "step": 27940
    },
    {
      "epoch": 83.93,
      "grad_norm": 13.234593391418457,
      "learning_rate": 7.202202202202203e-06,
      "loss": 0.3322,
      "step": 27950
    },
    {
      "epoch": 83.96,
      "grad_norm": 17.15028953552246,
      "learning_rate": 7.201201201201202e-06,
      "loss": 0.24,
      "step": 27960
    },
    {
      "epoch": 83.99,
      "grad_norm": 9.032596588134766,
      "learning_rate": 7.2002002002002005e-06,
      "loss": 0.2917,
      "step": 27970
    },
    {
      "epoch": 84.0,
      "eval_accuracy": 0.9259,
      "eval_loss": 0.29068198800086975,
      "eval_runtime": 30.5133,
      "eval_samples_per_second": 327.726,
      "eval_steps_per_second": 1.311,
      "step": 27972
    },
    {
      "epoch": 84.02,
      "grad_norm": 9.670671463012695,
      "learning_rate": 7.199199199199199e-06,
      "loss": 0.269,
      "step": 27980
    },
    {
      "epoch": 84.05,
      "grad_norm": 8.880677223205566,
      "learning_rate": 7.1981981981981985e-06,
      "loss": 0.2558,
      "step": 27990
    },
    {
      "epoch": 84.08,
      "grad_norm": 12.26557731628418,
      "learning_rate": 7.197197197197198e-06,
      "loss": 0.2707,
      "step": 28000
    },
    {
      "epoch": 84.11,
      "grad_norm": 6.374432563781738,
      "learning_rate": 7.1961961961961965e-06,
      "loss": 0.3138,
      "step": 28010
    },
    {
      "epoch": 84.14,
      "grad_norm": 14.15295696258545,
      "learning_rate": 7.195195195195196e-06,
      "loss": 0.307,
      "step": 28020
    },
    {
      "epoch": 84.17,
      "grad_norm": 15.056781768798828,
      "learning_rate": 7.194194194194195e-06,
      "loss": 0.2823,
      "step": 28030
    },
    {
      "epoch": 84.2,
      "grad_norm": 10.194563865661621,
      "learning_rate": 7.193193193193194e-06,
      "loss": 0.2756,
      "step": 28040
    },
    {
      "epoch": 84.23,
      "grad_norm": 10.495965957641602,
      "learning_rate": 7.1921921921921935e-06,
      "loss": 0.3017,
      "step": 28050
    },
    {
      "epoch": 84.26,
      "grad_norm": 8.493156433105469,
      "learning_rate": 7.191191191191191e-06,
      "loss": 0.3055,
      "step": 28060
    },
    {
      "epoch": 84.29,
      "grad_norm": 9.841859817504883,
      "learning_rate": 7.190190190190191e-06,
      "loss": 0.2829,
      "step": 28070
    },
    {
      "epoch": 84.32,
      "grad_norm": 19.03350257873535,
      "learning_rate": 7.189189189189189e-06,
      "loss": 0.3226,
      "step": 28080
    },
    {
      "epoch": 84.35,
      "grad_norm": 8.846536636352539,
      "learning_rate": 7.188188188188189e-06,
      "loss": 0.318,
      "step": 28090
    },
    {
      "epoch": 84.38,
      "grad_norm": 8.474943161010742,
      "learning_rate": 7.187187187187188e-06,
      "loss": 0.3283,
      "step": 28100
    },
    {
      "epoch": 84.41,
      "grad_norm": 16.96185874938965,
      "learning_rate": 7.186186186186187e-06,
      "loss": 0.26,
      "step": 28110
    },
    {
      "epoch": 84.44,
      "grad_norm": 14.721213340759277,
      "learning_rate": 7.185185185185186e-06,
      "loss": 0.2985,
      "step": 28120
    },
    {
      "epoch": 84.47,
      "grad_norm": 10.672587394714355,
      "learning_rate": 7.184184184184186e-06,
      "loss": 0.266,
      "step": 28130
    },
    {
      "epoch": 84.5,
      "grad_norm": 20.473649978637695,
      "learning_rate": 7.183183183183183e-06,
      "loss": 0.2964,
      "step": 28140
    },
    {
      "epoch": 84.53,
      "grad_norm": 6.8922882080078125,
      "learning_rate": 7.182182182182183e-06,
      "loss": 0.2815,
      "step": 28150
    },
    {
      "epoch": 84.56,
      "grad_norm": 8.606745719909668,
      "learning_rate": 7.181181181181181e-06,
      "loss": 0.2824,
      "step": 28160
    },
    {
      "epoch": 84.59,
      "grad_norm": 8.165632247924805,
      "learning_rate": 7.180180180180181e-06,
      "loss": 0.3019,
      "step": 28170
    },
    {
      "epoch": 84.62,
      "grad_norm": 9.1893949508667,
      "learning_rate": 7.1791791791791794e-06,
      "loss": 0.3161,
      "step": 28180
    },
    {
      "epoch": 84.65,
      "grad_norm": 17.917783737182617,
      "learning_rate": 7.178178178178179e-06,
      "loss": 0.3018,
      "step": 28190
    },
    {
      "epoch": 84.68,
      "grad_norm": 11.91406536102295,
      "learning_rate": 7.177177177177178e-06,
      "loss": 0.3337,
      "step": 28200
    },
    {
      "epoch": 84.71,
      "grad_norm": 14.837333679199219,
      "learning_rate": 7.176176176176177e-06,
      "loss": 0.3036,
      "step": 28210
    },
    {
      "epoch": 84.74,
      "grad_norm": 11.18201732635498,
      "learning_rate": 7.1751751751751755e-06,
      "loss": 0.3026,
      "step": 28220
    },
    {
      "epoch": 84.77,
      "grad_norm": 8.717573165893555,
      "learning_rate": 7.174174174174174e-06,
      "loss": 0.2961,
      "step": 28230
    },
    {
      "epoch": 84.8,
      "grad_norm": 10.547808647155762,
      "learning_rate": 7.1731731731731735e-06,
      "loss": 0.3002,
      "step": 28240
    },
    {
      "epoch": 84.83,
      "grad_norm": 7.682621002197266,
      "learning_rate": 7.172172172172173e-06,
      "loss": 0.3289,
      "step": 28250
    },
    {
      "epoch": 84.86,
      "grad_norm": 16.758419036865234,
      "learning_rate": 7.1711711711711716e-06,
      "loss": 0.3133,
      "step": 28260
    },
    {
      "epoch": 84.89,
      "grad_norm": 9.294572830200195,
      "learning_rate": 7.170170170170171e-06,
      "loss": 0.2448,
      "step": 28270
    },
    {
      "epoch": 84.92,
      "grad_norm": 14.06871223449707,
      "learning_rate": 7.1691691691691705e-06,
      "loss": 0.3059,
      "step": 28280
    },
    {
      "epoch": 84.95,
      "grad_norm": 9.065227508544922,
      "learning_rate": 7.168168168168169e-06,
      "loss": 0.2351,
      "step": 28290
    },
    {
      "epoch": 84.98,
      "grad_norm": 14.96493148803711,
      "learning_rate": 7.167167167167167e-06,
      "loss": 0.26,
      "step": 28300
    },
    {
      "epoch": 85.0,
      "eval_accuracy": 0.9265,
      "eval_loss": 0.29467448592185974,
      "eval_runtime": 30.1688,
      "eval_samples_per_second": 331.469,
      "eval_steps_per_second": 1.326,
      "step": 28305
    },
    {
      "epoch": 85.02,
      "grad_norm": 12.974924087524414,
      "learning_rate": 7.166166166166166e-06,
      "loss": 0.3592,
      "step": 28310
    },
    {
      "epoch": 85.05,
      "grad_norm": 10.77220344543457,
      "learning_rate": 7.165165165165166e-06,
      "loss": 0.2998,
      "step": 28320
    },
    {
      "epoch": 85.08,
      "grad_norm": 15.71430778503418,
      "learning_rate": 7.164164164164164e-06,
      "loss": 0.2907,
      "step": 28330
    },
    {
      "epoch": 85.11,
      "grad_norm": 37.743717193603516,
      "learning_rate": 7.163163163163164e-06,
      "loss": 0.3007,
      "step": 28340
    },
    {
      "epoch": 85.14,
      "grad_norm": 9.445643424987793,
      "learning_rate": 7.162162162162163e-06,
      "loss": 0.3242,
      "step": 28350
    },
    {
      "epoch": 85.17,
      "grad_norm": 10.246770858764648,
      "learning_rate": 7.161161161161162e-06,
      "loss": 0.2919,
      "step": 28360
    },
    {
      "epoch": 85.2,
      "grad_norm": 11.181761741638184,
      "learning_rate": 7.160160160160161e-06,
      "loss": 0.2783,
      "step": 28370
    },
    {
      "epoch": 85.23,
      "grad_norm": 19.352380752563477,
      "learning_rate": 7.159159159159161e-06,
      "loss": 0.2876,
      "step": 28380
    },
    {
      "epoch": 85.26,
      "grad_norm": 8.152127265930176,
      "learning_rate": 7.158158158158158e-06,
      "loss": 0.3531,
      "step": 28390
    },
    {
      "epoch": 85.29,
      "grad_norm": 19.3756046295166,
      "learning_rate": 7.157157157157158e-06,
      "loss": 0.2761,
      "step": 28400
    },
    {
      "epoch": 85.32,
      "grad_norm": 18.131790161132812,
      "learning_rate": 7.156156156156156e-06,
      "loss": 0.2873,
      "step": 28410
    },
    {
      "epoch": 85.35,
      "grad_norm": 21.568172454833984,
      "learning_rate": 7.155155155155156e-06,
      "loss": 0.3119,
      "step": 28420
    },
    {
      "epoch": 85.38,
      "grad_norm": 9.13001537322998,
      "learning_rate": 7.1541541541541545e-06,
      "loss": 0.305,
      "step": 28430
    },
    {
      "epoch": 85.41,
      "grad_norm": 10.144412994384766,
      "learning_rate": 7.153153153153154e-06,
      "loss": 0.2406,
      "step": 28440
    },
    {
      "epoch": 85.44,
      "grad_norm": 12.374094009399414,
      "learning_rate": 7.152152152152153e-06,
      "loss": 0.3413,
      "step": 28450
    },
    {
      "epoch": 85.47,
      "grad_norm": 12.647130012512207,
      "learning_rate": 7.151151151151152e-06,
      "loss": 0.2884,
      "step": 28460
    },
    {
      "epoch": 85.5,
      "grad_norm": 12.356242179870605,
      "learning_rate": 7.1501501501501505e-06,
      "loss": 0.2811,
      "step": 28470
    },
    {
      "epoch": 85.53,
      "grad_norm": 6.7338409423828125,
      "learning_rate": 7.149149149149149e-06,
      "loss": 0.2886,
      "step": 28480
    },
    {
      "epoch": 85.56,
      "grad_norm": 7.867846488952637,
      "learning_rate": 7.1481481481481486e-06,
      "loss": 0.3481,
      "step": 28490
    },
    {
      "epoch": 85.59,
      "grad_norm": 122.45582580566406,
      "learning_rate": 7.147147147147148e-06,
      "loss": 0.3173,
      "step": 28500
    },
    {
      "epoch": 85.62,
      "grad_norm": 14.446922302246094,
      "learning_rate": 7.146146146146147e-06,
      "loss": 0.2715,
      "step": 28510
    },
    {
      "epoch": 85.65,
      "grad_norm": 10.396017074584961,
      "learning_rate": 7.145145145145146e-06,
      "loss": 0.2821,
      "step": 28520
    },
    {
      "epoch": 85.68,
      "grad_norm": 59.00685501098633,
      "learning_rate": 7.144144144144145e-06,
      "loss": 0.3074,
      "step": 28530
    },
    {
      "epoch": 85.71,
      "grad_norm": 13.762097358703613,
      "learning_rate": 7.143143143143144e-06,
      "loss": 0.2656,
      "step": 28540
    },
    {
      "epoch": 85.74,
      "grad_norm": 8.662734031677246,
      "learning_rate": 7.142142142142142e-06,
      "loss": 0.2936,
      "step": 28550
    },
    {
      "epoch": 85.77,
      "grad_norm": 8.635159492492676,
      "learning_rate": 7.141141141141141e-06,
      "loss": 0.2677,
      "step": 28560
    },
    {
      "epoch": 85.8,
      "grad_norm": 10.251562118530273,
      "learning_rate": 7.140140140140141e-06,
      "loss": 0.3323,
      "step": 28570
    },
    {
      "epoch": 85.83,
      "grad_norm": 14.08121395111084,
      "learning_rate": 7.139139139139139e-06,
      "loss": 0.2561,
      "step": 28580
    },
    {
      "epoch": 85.86,
      "grad_norm": 24.782039642333984,
      "learning_rate": 7.138138138138139e-06,
      "loss": 0.2918,
      "step": 28590
    },
    {
      "epoch": 85.89,
      "grad_norm": 9.088736534118652,
      "learning_rate": 7.137137137137138e-06,
      "loss": 0.2904,
      "step": 28600
    },
    {
      "epoch": 85.92,
      "grad_norm": 6.0007147789001465,
      "learning_rate": 7.136136136136137e-06,
      "loss": 0.3061,
      "step": 28610
    },
    {
      "epoch": 85.95,
      "grad_norm": 8.561457633972168,
      "learning_rate": 7.135135135135136e-06,
      "loss": 0.2619,
      "step": 28620
    },
    {
      "epoch": 85.98,
      "grad_norm": 6.140302658081055,
      "learning_rate": 7.134134134134135e-06,
      "loss": 0.2604,
      "step": 28630
    },
    {
      "epoch": 86.0,
      "eval_accuracy": 0.9261,
      "eval_loss": 0.30042600631713867,
      "eval_runtime": 30.5284,
      "eval_samples_per_second": 327.563,
      "eval_steps_per_second": 1.31,
      "step": 28638
    },
    {
      "epoch": 86.01,
      "grad_norm": 9.784933090209961,
      "learning_rate": 7.133133133133133e-06,
      "loss": 0.2903,
      "step": 28640
    },
    {
      "epoch": 86.04,
      "grad_norm": 8.905136108398438,
      "learning_rate": 7.132132132132132e-06,
      "loss": 0.2859,
      "step": 28650
    },
    {
      "epoch": 86.07,
      "grad_norm": 8.799027442932129,
      "learning_rate": 7.1311311311311315e-06,
      "loss": 0.2906,
      "step": 28660
    },
    {
      "epoch": 86.1,
      "grad_norm": 6.735307216644287,
      "learning_rate": 7.130130130130131e-06,
      "loss": 0.2495,
      "step": 28670
    },
    {
      "epoch": 86.13,
      "grad_norm": 14.475642204284668,
      "learning_rate": 7.1291291291291295e-06,
      "loss": 0.2876,
      "step": 28680
    },
    {
      "epoch": 86.16,
      "grad_norm": 10.24792766571045,
      "learning_rate": 7.128128128128129e-06,
      "loss": 0.2901,
      "step": 28690
    },
    {
      "epoch": 86.19,
      "grad_norm": 7.19622802734375,
      "learning_rate": 7.127127127127128e-06,
      "loss": 0.2661,
      "step": 28700
    },
    {
      "epoch": 86.22,
      "grad_norm": 7.377612590789795,
      "learning_rate": 7.126126126126127e-06,
      "loss": 0.3206,
      "step": 28710
    },
    {
      "epoch": 86.25,
      "grad_norm": 8.202073097229004,
      "learning_rate": 7.1251251251251256e-06,
      "loss": 0.2983,
      "step": 28720
    },
    {
      "epoch": 86.28,
      "grad_norm": 9.79664134979248,
      "learning_rate": 7.124124124124124e-06,
      "loss": 0.2537,
      "step": 28730
    },
    {
      "epoch": 86.31,
      "grad_norm": 7.684638977050781,
      "learning_rate": 7.123123123123124e-06,
      "loss": 0.287,
      "step": 28740
    },
    {
      "epoch": 86.34,
      "grad_norm": 8.077546119689941,
      "learning_rate": 7.122122122122122e-06,
      "loss": 0.3218,
      "step": 28750
    },
    {
      "epoch": 86.37,
      "grad_norm": 7.732881546020508,
      "learning_rate": 7.121121121121122e-06,
      "loss": 0.2511,
      "step": 28760
    },
    {
      "epoch": 86.4,
      "grad_norm": 9.891357421875,
      "learning_rate": 7.120120120120121e-06,
      "loss": 0.3186,
      "step": 28770
    },
    {
      "epoch": 86.43,
      "grad_norm": 7.475708484649658,
      "learning_rate": 7.11911911911912e-06,
      "loss": 0.3228,
      "step": 28780
    },
    {
      "epoch": 86.46,
      "grad_norm": 9.813971519470215,
      "learning_rate": 7.118118118118119e-06,
      "loss": 0.3159,
      "step": 28790
    },
    {
      "epoch": 86.49,
      "grad_norm": 8.095967292785645,
      "learning_rate": 7.117117117117117e-06,
      "loss": 0.3074,
      "step": 28800
    },
    {
      "epoch": 86.52,
      "grad_norm": 17.22477149963379,
      "learning_rate": 7.116116116116116e-06,
      "loss": 0.279,
      "step": 28810
    },
    {
      "epoch": 86.55,
      "grad_norm": 15.769278526306152,
      "learning_rate": 7.115115115115116e-06,
      "loss": 0.3286,
      "step": 28820
    },
    {
      "epoch": 86.58,
      "grad_norm": 18.95981788635254,
      "learning_rate": 7.114114114114114e-06,
      "loss": 0.3369,
      "step": 28830
    },
    {
      "epoch": 86.61,
      "grad_norm": 18.074329376220703,
      "learning_rate": 7.113113113113114e-06,
      "loss": 0.331,
      "step": 28840
    },
    {
      "epoch": 86.64,
      "grad_norm": 14.02779769897461,
      "learning_rate": 7.112112112112113e-06,
      "loss": 0.3298,
      "step": 28850
    },
    {
      "epoch": 86.67,
      "grad_norm": 9.945918083190918,
      "learning_rate": 7.111111111111112e-06,
      "loss": 0.2866,
      "step": 28860
    },
    {
      "epoch": 86.7,
      "grad_norm": 12.348076820373535,
      "learning_rate": 7.110110110110111e-06,
      "loss": 0.2855,
      "step": 28870
    },
    {
      "epoch": 86.73,
      "grad_norm": 8.411255836486816,
      "learning_rate": 7.10910910910911e-06,
      "loss": 0.2405,
      "step": 28880
    },
    {
      "epoch": 86.76,
      "grad_norm": 8.618349075317383,
      "learning_rate": 7.1081081081081085e-06,
      "loss": 0.3326,
      "step": 28890
    },
    {
      "epoch": 86.79,
      "grad_norm": 8.680502891540527,
      "learning_rate": 7.107107107107107e-06,
      "loss": 0.3129,
      "step": 28900
    },
    {
      "epoch": 86.82,
      "grad_norm": 7.5645575523376465,
      "learning_rate": 7.1061061061061065e-06,
      "loss": 0.2718,
      "step": 28910
    },
    {
      "epoch": 86.85,
      "grad_norm": 16.300708770751953,
      "learning_rate": 7.105105105105106e-06,
      "loss": 0.246,
      "step": 28920
    },
    {
      "epoch": 86.88,
      "grad_norm": 10.891782760620117,
      "learning_rate": 7.1041041041041045e-06,
      "loss": 0.2892,
      "step": 28930
    },
    {
      "epoch": 86.91,
      "grad_norm": 9.826558113098145,
      "learning_rate": 7.103103103103104e-06,
      "loss": 0.2562,
      "step": 28940
    },
    {
      "epoch": 86.94,
      "grad_norm": 10.250933647155762,
      "learning_rate": 7.102102102102103e-06,
      "loss": 0.2736,
      "step": 28950
    },
    {
      "epoch": 86.97,
      "grad_norm": 11.491052627563477,
      "learning_rate": 7.101101101101102e-06,
      "loss": 0.2342,
      "step": 28960
    },
    {
      "epoch": 87.0,
      "grad_norm": 11.536721229553223,
      "learning_rate": 7.100100100100101e-06,
      "loss": 0.3287,
      "step": 28970
    },
    {
      "epoch": 87.0,
      "eval_accuracy": 0.9266,
      "eval_loss": 0.3002987802028656,
      "eval_runtime": 30.9169,
      "eval_samples_per_second": 323.448,
      "eval_steps_per_second": 1.294,
      "step": 28971
    },
    {
      "epoch": 87.03,
      "grad_norm": 6.500946998596191,
      "learning_rate": 7.099099099099099e-06,
      "loss": 0.2485,
      "step": 28980
    },
    {
      "epoch": 87.06,
      "grad_norm": 6.26481819152832,
      "learning_rate": 7.098098098098099e-06,
      "loss": 0.3378,
      "step": 28990
    },
    {
      "epoch": 87.09,
      "grad_norm": 5.91065788269043,
      "learning_rate": 7.097097097097097e-06,
      "loss": 0.2565,
      "step": 29000
    },
    {
      "epoch": 87.12,
      "grad_norm": 8.952195167541504,
      "learning_rate": 7.096096096096097e-06,
      "loss": 0.2617,
      "step": 29010
    },
    {
      "epoch": 87.15,
      "grad_norm": 7.297741889953613,
      "learning_rate": 7.095095095095096e-06,
      "loss": 0.3247,
      "step": 29020
    },
    {
      "epoch": 87.18,
      "grad_norm": 7.321237564086914,
      "learning_rate": 7.094094094094095e-06,
      "loss": 0.2451,
      "step": 29030
    },
    {
      "epoch": 87.21,
      "grad_norm": 15.131948471069336,
      "learning_rate": 7.093093093093094e-06,
      "loss": 0.2885,
      "step": 29040
    },
    {
      "epoch": 87.24,
      "grad_norm": 21.344871520996094,
      "learning_rate": 7.092092092092092e-06,
      "loss": 0.2908,
      "step": 29050
    },
    {
      "epoch": 87.27,
      "grad_norm": 12.868663787841797,
      "learning_rate": 7.091091091091091e-06,
      "loss": 0.2771,
      "step": 29060
    },
    {
      "epoch": 87.3,
      "grad_norm": 9.727673530578613,
      "learning_rate": 7.090090090090091e-06,
      "loss": 0.2833,
      "step": 29070
    },
    {
      "epoch": 87.33,
      "grad_norm": 9.284957885742188,
      "learning_rate": 7.089089089089089e-06,
      "loss": 0.2759,
      "step": 29080
    },
    {
      "epoch": 87.36,
      "grad_norm": 9.222702980041504,
      "learning_rate": 7.088088088088089e-06,
      "loss": 0.2758,
      "step": 29090
    },
    {
      "epoch": 87.39,
      "grad_norm": 35.191810607910156,
      "learning_rate": 7.087087087087087e-06,
      "loss": 0.2638,
      "step": 29100
    },
    {
      "epoch": 87.42,
      "grad_norm": 9.609633445739746,
      "learning_rate": 7.086086086086087e-06,
      "loss": 0.2591,
      "step": 29110
    },
    {
      "epoch": 87.45,
      "grad_norm": 9.782835006713867,
      "learning_rate": 7.085085085085086e-06,
      "loss": 0.3525,
      "step": 29120
    },
    {
      "epoch": 87.48,
      "grad_norm": 5.736701965332031,
      "learning_rate": 7.084084084084085e-06,
      "loss": 0.297,
      "step": 29130
    },
    {
      "epoch": 87.51,
      "grad_norm": 8.390459060668945,
      "learning_rate": 7.0830830830830835e-06,
      "loss": 0.3012,
      "step": 29140
    },
    {
      "epoch": 87.54,
      "grad_norm": 11.734288215637207,
      "learning_rate": 7.082082082082082e-06,
      "loss": 0.3207,
      "step": 29150
    },
    {
      "epoch": 87.57,
      "grad_norm": 26.962360382080078,
      "learning_rate": 7.0810810810810815e-06,
      "loss": 0.2812,
      "step": 29160
    },
    {
      "epoch": 87.6,
      "grad_norm": 5.585750102996826,
      "learning_rate": 7.080080080080081e-06,
      "loss": 0.3029,
      "step": 29170
    },
    {
      "epoch": 87.63,
      "grad_norm": 8.923055648803711,
      "learning_rate": 7.0790790790790796e-06,
      "loss": 0.2949,
      "step": 29180
    },
    {
      "epoch": 87.66,
      "grad_norm": 22.127901077270508,
      "learning_rate": 7.078078078078079e-06,
      "loss": 0.3274,
      "step": 29190
    },
    {
      "epoch": 87.69,
      "grad_norm": 10.564157485961914,
      "learning_rate": 7.0770770770770784e-06,
      "loss": 0.263,
      "step": 29200
    },
    {
      "epoch": 87.72,
      "grad_norm": 7.4355926513671875,
      "learning_rate": 7.076076076076077e-06,
      "loss": 0.2703,
      "step": 29210
    },
    {
      "epoch": 87.75,
      "grad_norm": 9.199353218078613,
      "learning_rate": 7.075075075075075e-06,
      "loss": 0.3313,
      "step": 29220
    },
    {
      "epoch": 87.78,
      "grad_norm": 8.930550575256348,
      "learning_rate": 7.074074074074074e-06,
      "loss": 0.317,
      "step": 29230
    },
    {
      "epoch": 87.81,
      "grad_norm": 12.43084716796875,
      "learning_rate": 7.073073073073074e-06,
      "loss": 0.339,
      "step": 29240
    },
    {
      "epoch": 87.84,
      "grad_norm": 9.698212623596191,
      "learning_rate": 7.072072072072072e-06,
      "loss": 0.303,
      "step": 29250
    },
    {
      "epoch": 87.87,
      "grad_norm": 8.141523361206055,
      "learning_rate": 7.071071071071072e-06,
      "loss": 0.2872,
      "step": 29260
    },
    {
      "epoch": 87.9,
      "grad_norm": 5.143004894256592,
      "learning_rate": 7.070070070070071e-06,
      "loss": 0.2388,
      "step": 29270
    },
    {
      "epoch": 87.93,
      "grad_norm": 10.673596382141113,
      "learning_rate": 7.06906906906907e-06,
      "loss": 0.2691,
      "step": 29280
    },
    {
      "epoch": 87.96,
      "grad_norm": 15.381929397583008,
      "learning_rate": 7.068068068068069e-06,
      "loss": 0.3322,
      "step": 29290
    },
    {
      "epoch": 87.99,
      "grad_norm": 10.597864151000977,
      "learning_rate": 7.067067067067067e-06,
      "loss": 0.3101,
      "step": 29300
    },
    {
      "epoch": 88.0,
      "eval_accuracy": 0.9277,
      "eval_loss": 0.300062894821167,
      "eval_runtime": 30.1324,
      "eval_samples_per_second": 331.869,
      "eval_steps_per_second": 1.327,
      "step": 29304
    },
    {
      "epoch": 88.02,
      "grad_norm": 8.634299278259277,
      "learning_rate": 7.066066066066066e-06,
      "loss": 0.2667,
      "step": 29310
    },
    {
      "epoch": 88.05,
      "grad_norm": 9.06824779510498,
      "learning_rate": 7.065065065065066e-06,
      "loss": 0.3085,
      "step": 29320
    },
    {
      "epoch": 88.08,
      "grad_norm": 6.699244022369385,
      "learning_rate": 7.064064064064064e-06,
      "loss": 0.2812,
      "step": 29330
    },
    {
      "epoch": 88.11,
      "grad_norm": 11.107563972473145,
      "learning_rate": 7.063063063063064e-06,
      "loss": 0.2794,
      "step": 29340
    },
    {
      "epoch": 88.14,
      "grad_norm": 8.84051513671875,
      "learning_rate": 7.0620620620620625e-06,
      "loss": 0.2664,
      "step": 29350
    },
    {
      "epoch": 88.17,
      "grad_norm": 10.723832130432129,
      "learning_rate": 7.061061061061062e-06,
      "loss": 0.2116,
      "step": 29360
    },
    {
      "epoch": 88.2,
      "grad_norm": 7.627597332000732,
      "learning_rate": 7.060060060060061e-06,
      "loss": 0.2638,
      "step": 29370
    },
    {
      "epoch": 88.23,
      "grad_norm": 10.529794692993164,
      "learning_rate": 7.059059059059059e-06,
      "loss": 0.3342,
      "step": 29380
    },
    {
      "epoch": 88.26,
      "grad_norm": 7.4625935554504395,
      "learning_rate": 7.0580580580580585e-06,
      "loss": 0.2683,
      "step": 29390
    },
    {
      "epoch": 88.29,
      "grad_norm": 10.652615547180176,
      "learning_rate": 7.057057057057057e-06,
      "loss": 0.2476,
      "step": 29400
    },
    {
      "epoch": 88.32,
      "grad_norm": 13.747520446777344,
      "learning_rate": 7.0560560560560566e-06,
      "loss": 0.2688,
      "step": 29410
    },
    {
      "epoch": 88.35,
      "grad_norm": 14.304877281188965,
      "learning_rate": 7.055055055055056e-06,
      "loss": 0.3002,
      "step": 29420
    },
    {
      "epoch": 88.38,
      "grad_norm": 8.990193367004395,
      "learning_rate": 7.054054054054055e-06,
      "loss": 0.2612,
      "step": 29430
    },
    {
      "epoch": 88.41,
      "grad_norm": 11.855331420898438,
      "learning_rate": 7.053053053053054e-06,
      "loss": 0.2733,
      "step": 29440
    },
    {
      "epoch": 88.44,
      "grad_norm": 4.98137903213501,
      "learning_rate": 7.052052052052053e-06,
      "loss": 0.3117,
      "step": 29450
    },
    {
      "epoch": 88.47,
      "grad_norm": 29.941574096679688,
      "learning_rate": 7.051051051051052e-06,
      "loss": 0.3218,
      "step": 29460
    },
    {
      "epoch": 88.5,
      "grad_norm": 11.09192943572998,
      "learning_rate": 7.05005005005005e-06,
      "loss": 0.2883,
      "step": 29470
    },
    {
      "epoch": 88.53,
      "grad_norm": 6.5553765296936035,
      "learning_rate": 7.049049049049049e-06,
      "loss": 0.2982,
      "step": 29480
    },
    {
      "epoch": 88.56,
      "grad_norm": 9.140303611755371,
      "learning_rate": 7.048048048048049e-06,
      "loss": 0.2928,
      "step": 29490
    },
    {
      "epoch": 88.59,
      "grad_norm": 9.46389102935791,
      "learning_rate": 7.047047047047047e-06,
      "loss": 0.2816,
      "step": 29500
    },
    {
      "epoch": 88.62,
      "grad_norm": 11.448698043823242,
      "learning_rate": 7.046046046046047e-06,
      "loss": 0.2946,
      "step": 29510
    },
    {
      "epoch": 88.65,
      "grad_norm": 10.794591903686523,
      "learning_rate": 7.045045045045046e-06,
      "loss": 0.2936,
      "step": 29520
    },
    {
      "epoch": 88.68,
      "grad_norm": 12.189172744750977,
      "learning_rate": 7.044044044044045e-06,
      "loss": 0.2847,
      "step": 29530
    },
    {
      "epoch": 88.71,
      "grad_norm": 12.926974296569824,
      "learning_rate": 7.043043043043044e-06,
      "loss": 0.3169,
      "step": 29540
    },
    {
      "epoch": 88.74,
      "grad_norm": 8.834761619567871,
      "learning_rate": 7.042042042042042e-06,
      "loss": 0.3168,
      "step": 29550
    },
    {
      "epoch": 88.77,
      "grad_norm": 11.050698280334473,
      "learning_rate": 7.041041041041041e-06,
      "loss": 0.2912,
      "step": 29560
    },
    {
      "epoch": 88.8,
      "grad_norm": 6.671822547912598,
      "learning_rate": 7.04004004004004e-06,
      "loss": 0.2841,
      "step": 29570
    },
    {
      "epoch": 88.83,
      "grad_norm": 9.98063850402832,
      "learning_rate": 7.0390390390390395e-06,
      "loss": 0.2558,
      "step": 29580
    },
    {
      "epoch": 88.86,
      "grad_norm": 9.138448715209961,
      "learning_rate": 7.038038038038039e-06,
      "loss": 0.2895,
      "step": 29590
    },
    {
      "epoch": 88.89,
      "grad_norm": 12.977328300476074,
      "learning_rate": 7.0370370370370375e-06,
      "loss": 0.3342,
      "step": 29600
    },
    {
      "epoch": 88.92,
      "grad_norm": 13.30918025970459,
      "learning_rate": 7.036036036036037e-06,
      "loss": 0.3322,
      "step": 29610
    },
    {
      "epoch": 88.95,
      "grad_norm": 11.403301239013672,
      "learning_rate": 7.035035035035036e-06,
      "loss": 0.2729,
      "step": 29620
    },
    {
      "epoch": 88.98,
      "grad_norm": 7.307366847991943,
      "learning_rate": 7.034034034034034e-06,
      "loss": 0.3029,
      "step": 29630
    },
    {
      "epoch": 89.0,
      "eval_accuracy": 0.9275,
      "eval_loss": 0.2987031936645508,
      "eval_runtime": 30.4174,
      "eval_samples_per_second": 328.759,
      "eval_steps_per_second": 1.315,
      "step": 29637
    },
    {
      "epoch": 89.01,
      "grad_norm": 12.190999984741211,
      "learning_rate": 7.0330330330330336e-06,
      "loss": 0.2154,
      "step": 29640
    },
    {
      "epoch": 89.04,
      "grad_norm": 10.135000228881836,
      "learning_rate": 7.032032032032032e-06,
      "loss": 0.3404,
      "step": 29650
    },
    {
      "epoch": 89.07,
      "grad_norm": 8.10612678527832,
      "learning_rate": 7.031031031031032e-06,
      "loss": 0.2521,
      "step": 29660
    },
    {
      "epoch": 89.1,
      "grad_norm": 7.471124172210693,
      "learning_rate": 7.03003003003003e-06,
      "loss": 0.3129,
      "step": 29670
    },
    {
      "epoch": 89.13,
      "grad_norm": 11.817010879516602,
      "learning_rate": 7.02902902902903e-06,
      "loss": 0.3069,
      "step": 29680
    },
    {
      "epoch": 89.16,
      "grad_norm": 8.054304122924805,
      "learning_rate": 7.028028028028029e-06,
      "loss": 0.2937,
      "step": 29690
    },
    {
      "epoch": 89.19,
      "grad_norm": 11.196502685546875,
      "learning_rate": 7.027027027027028e-06,
      "loss": 0.2864,
      "step": 29700
    },
    {
      "epoch": 89.22,
      "grad_norm": 9.79306697845459,
      "learning_rate": 7.026026026026027e-06,
      "loss": 0.2713,
      "step": 29710
    },
    {
      "epoch": 89.25,
      "grad_norm": 11.620367050170898,
      "learning_rate": 7.025025025025025e-06,
      "loss": 0.2809,
      "step": 29720
    },
    {
      "epoch": 89.28,
      "grad_norm": 8.36703109741211,
      "learning_rate": 7.024024024024024e-06,
      "loss": 0.3252,
      "step": 29730
    },
    {
      "epoch": 89.31,
      "grad_norm": 9.065302848815918,
      "learning_rate": 7.023023023023024e-06,
      "loss": 0.2712,
      "step": 29740
    },
    {
      "epoch": 89.34,
      "grad_norm": 10.900071144104004,
      "learning_rate": 7.022022022022022e-06,
      "loss": 0.2828,
      "step": 29750
    },
    {
      "epoch": 89.37,
      "grad_norm": 6.83170747756958,
      "learning_rate": 7.021021021021022e-06,
      "loss": 0.2132,
      "step": 29760
    },
    {
      "epoch": 89.4,
      "grad_norm": 10.87137508392334,
      "learning_rate": 7.020020020020021e-06,
      "loss": 0.3529,
      "step": 29770
    },
    {
      "epoch": 89.43,
      "grad_norm": 9.650304794311523,
      "learning_rate": 7.01901901901902e-06,
      "loss": 0.3105,
      "step": 29780
    },
    {
      "epoch": 89.46,
      "grad_norm": 8.772736549377441,
      "learning_rate": 7.018018018018019e-06,
      "loss": 0.2785,
      "step": 29790
    },
    {
      "epoch": 89.49,
      "grad_norm": 9.998861312866211,
      "learning_rate": 7.017017017017017e-06,
      "loss": 0.3132,
      "step": 29800
    },
    {
      "epoch": 89.52,
      "grad_norm": 10.021339416503906,
      "learning_rate": 7.0160160160160164e-06,
      "loss": 0.3247,
      "step": 29810
    },
    {
      "epoch": 89.55,
      "grad_norm": 20.3895263671875,
      "learning_rate": 7.015015015015015e-06,
      "loss": 0.3035,
      "step": 29820
    },
    {
      "epoch": 89.58,
      "grad_norm": 6.067256927490234,
      "learning_rate": 7.0140140140140145e-06,
      "loss": 0.2527,
      "step": 29830
    },
    {
      "epoch": 89.61,
      "grad_norm": 8.502212524414062,
      "learning_rate": 7.013013013013014e-06,
      "loss": 0.278,
      "step": 29840
    },
    {
      "epoch": 89.64,
      "grad_norm": 11.242568016052246,
      "learning_rate": 7.0120120120120125e-06,
      "loss": 0.2728,
      "step": 29850
    },
    {
      "epoch": 89.67,
      "grad_norm": 14.380844116210938,
      "learning_rate": 7.011011011011012e-06,
      "loss": 0.2751,
      "step": 29860
    },
    {
      "epoch": 89.7,
      "grad_norm": 6.143453598022461,
      "learning_rate": 7.010010010010011e-06,
      "loss": 0.2977,
      "step": 29870
    },
    {
      "epoch": 89.73,
      "grad_norm": 13.226771354675293,
      "learning_rate": 7.009009009009009e-06,
      "loss": 0.3252,
      "step": 29880
    },
    {
      "epoch": 89.76,
      "grad_norm": 10.803184509277344,
      "learning_rate": 7.008008008008009e-06,
      "loss": 0.2607,
      "step": 29890
    },
    {
      "epoch": 89.79,
      "grad_norm": 17.22111701965332,
      "learning_rate": 7.007007007007007e-06,
      "loss": 0.2854,
      "step": 29900
    },
    {
      "epoch": 89.82,
      "grad_norm": 10.099568367004395,
      "learning_rate": 7.006006006006007e-06,
      "loss": 0.2708,
      "step": 29910
    },
    {
      "epoch": 89.85,
      "grad_norm": 6.2415690422058105,
      "learning_rate": 7.005005005005005e-06,
      "loss": 0.2692,
      "step": 29920
    },
    {
      "epoch": 89.88,
      "grad_norm": 22.18634796142578,
      "learning_rate": 7.004004004004005e-06,
      "loss": 0.3053,
      "step": 29930
    },
    {
      "epoch": 89.91,
      "grad_norm": 4.689823627471924,
      "learning_rate": 7.003003003003004e-06,
      "loss": 0.2636,
      "step": 29940
    },
    {
      "epoch": 89.94,
      "grad_norm": 12.848011016845703,
      "learning_rate": 7.002002002002003e-06,
      "loss": 0.2537,
      "step": 29950
    },
    {
      "epoch": 89.97,
      "grad_norm": 7.329112529754639,
      "learning_rate": 7.001001001001002e-06,
      "loss": 0.2948,
      "step": 29960
    },
    {
      "epoch": 90.0,
      "grad_norm": 71.33290100097656,
      "learning_rate": 7e-06,
      "loss": 0.2824,
      "step": 29970
    },
    {
      "epoch": 90.0,
      "eval_accuracy": 0.9266,
      "eval_loss": 0.30013662576675415,
      "eval_runtime": 30.1126,
      "eval_samples_per_second": 332.087,
      "eval_steps_per_second": 1.328,
      "step": 29970
    },
    {
      "epoch": 90.03,
      "grad_norm": 8.068912506103516,
      "learning_rate": 6.998998998998999e-06,
      "loss": 0.273,
      "step": 29980
    },
    {
      "epoch": 90.06,
      "grad_norm": 11.949800491333008,
      "learning_rate": 6.997997997997999e-06,
      "loss": 0.2954,
      "step": 29990
    },
    {
      "epoch": 90.09,
      "grad_norm": 9.636696815490723,
      "learning_rate": 6.996996996996997e-06,
      "loss": 0.3672,
      "step": 30000
    },
    {
      "epoch": 90.12,
      "grad_norm": 8.978914260864258,
      "learning_rate": 6.995995995995997e-06,
      "loss": 0.3312,
      "step": 30010
    },
    {
      "epoch": 90.15,
      "grad_norm": 25.254981994628906,
      "learning_rate": 6.994994994994995e-06,
      "loss": 0.2588,
      "step": 30020
    },
    {
      "epoch": 90.18,
      "grad_norm": 9.49367618560791,
      "learning_rate": 6.993993993993995e-06,
      "loss": 0.2829,
      "step": 30030
    },
    {
      "epoch": 90.21,
      "grad_norm": 8.209656715393066,
      "learning_rate": 6.992992992992994e-06,
      "loss": 0.3093,
      "step": 30040
    },
    {
      "epoch": 90.24,
      "grad_norm": 11.603708267211914,
      "learning_rate": 6.991991991991992e-06,
      "loss": 0.2735,
      "step": 30050
    },
    {
      "epoch": 90.27,
      "grad_norm": 4.515967845916748,
      "learning_rate": 6.9909909909909915e-06,
      "loss": 0.3053,
      "step": 30060
    },
    {
      "epoch": 90.3,
      "grad_norm": 25.136272430419922,
      "learning_rate": 6.98998998998999e-06,
      "loss": 0.2717,
      "step": 30070
    },
    {
      "epoch": 90.33,
      "grad_norm": 6.0635905265808105,
      "learning_rate": 6.9889889889889895e-06,
      "loss": 0.2953,
      "step": 30080
    },
    {
      "epoch": 90.36,
      "grad_norm": 8.261253356933594,
      "learning_rate": 6.987987987987989e-06,
      "loss": 0.3099,
      "step": 30090
    },
    {
      "epoch": 90.39,
      "grad_norm": 6.424069881439209,
      "learning_rate": 6.9869869869869876e-06,
      "loss": 0.2526,
      "step": 30100
    },
    {
      "epoch": 90.42,
      "grad_norm": 12.956831932067871,
      "learning_rate": 6.985985985985987e-06,
      "loss": 0.2991,
      "step": 30110
    },
    {
      "epoch": 90.45,
      "grad_norm": 6.644471645355225,
      "learning_rate": 6.984984984984986e-06,
      "loss": 0.2637,
      "step": 30120
    },
    {
      "epoch": 90.48,
      "grad_norm": 17.697816848754883,
      "learning_rate": 6.983983983983984e-06,
      "loss": 0.2657,
      "step": 30130
    },
    {
      "epoch": 90.51,
      "grad_norm": 86.36605834960938,
      "learning_rate": 6.982982982982983e-06,
      "loss": 0.2791,
      "step": 30140
    },
    {
      "epoch": 90.54,
      "grad_norm": 9.793549537658691,
      "learning_rate": 6.981981981981982e-06,
      "loss": 0.3521,
      "step": 30150
    },
    {
      "epoch": 90.57,
      "grad_norm": 7.923374652862549,
      "learning_rate": 6.980980980980982e-06,
      "loss": 0.2971,
      "step": 30160
    },
    {
      "epoch": 90.6,
      "grad_norm": 9.556580543518066,
      "learning_rate": 6.97997997997998e-06,
      "loss": 0.2629,
      "step": 30170
    },
    {
      "epoch": 90.63,
      "grad_norm": 8.561888694763184,
      "learning_rate": 6.97897897897898e-06,
      "loss": 0.2703,
      "step": 30180
    },
    {
      "epoch": 90.66,
      "grad_norm": 7.277593612670898,
      "learning_rate": 6.977977977977979e-06,
      "loss": 0.3144,
      "step": 30190
    },
    {
      "epoch": 90.69,
      "grad_norm": 10.599038124084473,
      "learning_rate": 6.976976976976978e-06,
      "loss": 0.2748,
      "step": 30200
    },
    {
      "epoch": 90.72,
      "grad_norm": 7.776122570037842,
      "learning_rate": 6.975975975975977e-06,
      "loss": 0.2642,
      "step": 30210
    },
    {
      "epoch": 90.75,
      "grad_norm": 10.75730037689209,
      "learning_rate": 6.974974974974975e-06,
      "loss": 0.3028,
      "step": 30220
    },
    {
      "epoch": 90.78,
      "grad_norm": 7.443554401397705,
      "learning_rate": 6.973973973973974e-06,
      "loss": 0.3222,
      "step": 30230
    },
    {
      "epoch": 90.81,
      "grad_norm": 8.884283065795898,
      "learning_rate": 6.972972972972973e-06,
      "loss": 0.2925,
      "step": 30240
    },
    {
      "epoch": 90.84,
      "grad_norm": 6.917208671569824,
      "learning_rate": 6.971971971971972e-06,
      "loss": 0.3096,
      "step": 30250
    },
    {
      "epoch": 90.87,
      "grad_norm": 21.176616668701172,
      "learning_rate": 6.970970970970972e-06,
      "loss": 0.233,
      "step": 30260
    },
    {
      "epoch": 90.9,
      "grad_norm": 14.437288284301758,
      "learning_rate": 6.9699699699699704e-06,
      "loss": 0.3085,
      "step": 30270
    },
    {
      "epoch": 90.93,
      "grad_norm": 11.070788383483887,
      "learning_rate": 6.96896896896897e-06,
      "loss": 0.3071,
      "step": 30280
    },
    {
      "epoch": 90.96,
      "grad_norm": 9.158970832824707,
      "learning_rate": 6.967967967967969e-06,
      "loss": 0.2951,
      "step": 30290
    },
    {
      "epoch": 90.99,
      "grad_norm": 31.45600128173828,
      "learning_rate": 6.966966966966967e-06,
      "loss": 0.2771,
      "step": 30300
    },
    {
      "epoch": 91.0,
      "eval_accuracy": 0.9261,
      "eval_loss": 0.30255126953125,
      "eval_runtime": 30.0855,
      "eval_samples_per_second": 332.386,
      "eval_steps_per_second": 1.33,
      "step": 30303
    },
    {
      "epoch": 91.02,
      "grad_norm": 15.893773078918457,
      "learning_rate": 6.9659659659659665e-06,
      "loss": 0.2711,
      "step": 30310
    },
    {
      "epoch": 91.05,
      "grad_norm": 9.631701469421387,
      "learning_rate": 6.964964964964965e-06,
      "loss": 0.3627,
      "step": 30320
    },
    {
      "epoch": 91.08,
      "grad_norm": 7.366189002990723,
      "learning_rate": 6.9639639639639646e-06,
      "loss": 0.2533,
      "step": 30330
    },
    {
      "epoch": 91.11,
      "grad_norm": 9.620949745178223,
      "learning_rate": 6.962962962962964e-06,
      "loss": 0.2556,
      "step": 30340
    },
    {
      "epoch": 91.14,
      "grad_norm": 6.847657203674316,
      "learning_rate": 6.961961961961963e-06,
      "loss": 0.2663,
      "step": 30350
    },
    {
      "epoch": 91.17,
      "grad_norm": 8.400248527526855,
      "learning_rate": 6.960960960960962e-06,
      "loss": 0.267,
      "step": 30360
    },
    {
      "epoch": 91.2,
      "grad_norm": 13.98305606842041,
      "learning_rate": 6.959959959959961e-06,
      "loss": 0.2616,
      "step": 30370
    },
    {
      "epoch": 91.23,
      "grad_norm": 6.7363409996032715,
      "learning_rate": 6.958958958958959e-06,
      "loss": 0.2621,
      "step": 30380
    },
    {
      "epoch": 91.26,
      "grad_norm": 8.319665908813477,
      "learning_rate": 6.957957957957958e-06,
      "loss": 0.2813,
      "step": 30390
    },
    {
      "epoch": 91.29,
      "grad_norm": 6.610774040222168,
      "learning_rate": 6.956956956956957e-06,
      "loss": 0.3003,
      "step": 30400
    },
    {
      "epoch": 91.32,
      "grad_norm": 6.5186591148376465,
      "learning_rate": 6.955955955955957e-06,
      "loss": 0.2534,
      "step": 30410
    },
    {
      "epoch": 91.35,
      "grad_norm": 11.418011665344238,
      "learning_rate": 6.954954954954955e-06,
      "loss": 0.344,
      "step": 30420
    },
    {
      "epoch": 91.38,
      "grad_norm": 7.956821918487549,
      "learning_rate": 6.953953953953955e-06,
      "loss": 0.3138,
      "step": 30430
    },
    {
      "epoch": 91.41,
      "grad_norm": 15.757438659667969,
      "learning_rate": 6.952952952952954e-06,
      "loss": 0.3003,
      "step": 30440
    },
    {
      "epoch": 91.44,
      "grad_norm": 9.013151168823242,
      "learning_rate": 6.951951951951953e-06,
      "loss": 0.2169,
      "step": 30450
    },
    {
      "epoch": 91.47,
      "grad_norm": 7.713266372680664,
      "learning_rate": 6.950950950950952e-06,
      "loss": 0.3201,
      "step": 30460
    },
    {
      "epoch": 91.5,
      "grad_norm": 7.176088333129883,
      "learning_rate": 6.94994994994995e-06,
      "loss": 0.2959,
      "step": 30470
    },
    {
      "epoch": 91.53,
      "grad_norm": 6.8665547370910645,
      "learning_rate": 6.948948948948949e-06,
      "loss": 0.2916,
      "step": 30480
    },
    {
      "epoch": 91.56,
      "grad_norm": 8.723104476928711,
      "learning_rate": 6.947947947947948e-06,
      "loss": 0.3563,
      "step": 30490
    },
    {
      "epoch": 91.59,
      "grad_norm": 9.111759185791016,
      "learning_rate": 6.9469469469469474e-06,
      "loss": 0.2882,
      "step": 30500
    },
    {
      "epoch": 91.62,
      "grad_norm": 9.61408519744873,
      "learning_rate": 6.945945945945947e-06,
      "loss": 0.2774,
      "step": 30510
    },
    {
      "epoch": 91.65,
      "grad_norm": 27.754512786865234,
      "learning_rate": 6.9449449449449455e-06,
      "loss": 0.3092,
      "step": 30520
    },
    {
      "epoch": 91.68,
      "grad_norm": 11.079262733459473,
      "learning_rate": 6.943943943943945e-06,
      "loss": 0.3076,
      "step": 30530
    },
    {
      "epoch": 91.71,
      "grad_norm": 8.84855842590332,
      "learning_rate": 6.942942942942944e-06,
      "loss": 0.2501,
      "step": 30540
    },
    {
      "epoch": 91.74,
      "grad_norm": 15.667255401611328,
      "learning_rate": 6.941941941941942e-06,
      "loss": 0.3101,
      "step": 30550
    },
    {
      "epoch": 91.77,
      "grad_norm": 7.347929954528809,
      "learning_rate": 6.9409409409409416e-06,
      "loss": 0.2555,
      "step": 30560
    },
    {
      "epoch": 91.8,
      "grad_norm": 6.374351978302002,
      "learning_rate": 6.93993993993994e-06,
      "loss": 0.2773,
      "step": 30570
    },
    {
      "epoch": 91.83,
      "grad_norm": 13.59241771697998,
      "learning_rate": 6.93893893893894e-06,
      "loss": 0.2895,
      "step": 30580
    },
    {
      "epoch": 91.86,
      "grad_norm": 6.793631553649902,
      "learning_rate": 6.937937937937938e-06,
      "loss": 0.2811,
      "step": 30590
    },
    {
      "epoch": 91.89,
      "grad_norm": 8.553747177124023,
      "learning_rate": 6.936936936936938e-06,
      "loss": 0.2717,
      "step": 30600
    },
    {
      "epoch": 91.92,
      "grad_norm": 10.6891508102417,
      "learning_rate": 6.935935935935937e-06,
      "loss": 0.2838,
      "step": 30610
    },
    {
      "epoch": 91.95,
      "grad_norm": 7.197050094604492,
      "learning_rate": 6.934934934934936e-06,
      "loss": 0.235,
      "step": 30620
    },
    {
      "epoch": 91.98,
      "grad_norm": 18.38099479675293,
      "learning_rate": 6.933933933933934e-06,
      "loss": 0.2428,
      "step": 30630
    },
    {
      "epoch": 92.0,
      "eval_accuracy": 0.9256,
      "eval_loss": 0.30522796511650085,
      "eval_runtime": 30.2233,
      "eval_samples_per_second": 330.87,
      "eval_steps_per_second": 1.323,
      "step": 30636
    },
    {
      "epoch": 92.01,
      "grad_norm": 8.0303373336792,
      "learning_rate": 6.932932932932933e-06,
      "loss": 0.271,
      "step": 30640
    },
    {
      "epoch": 92.04,
      "grad_norm": 9.588861465454102,
      "learning_rate": 6.931931931931932e-06,
      "loss": 0.2723,
      "step": 30650
    },
    {
      "epoch": 92.07,
      "grad_norm": 8.93665885925293,
      "learning_rate": 6.930930930930932e-06,
      "loss": 0.315,
      "step": 30660
    },
    {
      "epoch": 92.1,
      "grad_norm": 7.469074249267578,
      "learning_rate": 6.92992992992993e-06,
      "loss": 0.2498,
      "step": 30670
    },
    {
      "epoch": 92.13,
      "grad_norm": 7.462866306304932,
      "learning_rate": 6.92892892892893e-06,
      "loss": 0.2449,
      "step": 30680
    },
    {
      "epoch": 92.16,
      "grad_norm": 10.99606990814209,
      "learning_rate": 6.927927927927928e-06,
      "loss": 0.3222,
      "step": 30690
    },
    {
      "epoch": 92.19,
      "grad_norm": 7.0242085456848145,
      "learning_rate": 6.926926926926928e-06,
      "loss": 0.2592,
      "step": 30700
    },
    {
      "epoch": 92.22,
      "grad_norm": 11.912219047546387,
      "learning_rate": 6.9259259259259256e-06,
      "loss": 0.3055,
      "step": 30710
    },
    {
      "epoch": 92.25,
      "grad_norm": 8.399483680725098,
      "learning_rate": 6.924924924924925e-06,
      "loss": 0.2819,
      "step": 30720
    },
    {
      "epoch": 92.28,
      "grad_norm": 9.302955627441406,
      "learning_rate": 6.9239239239239244e-06,
      "loss": 0.2769,
      "step": 30730
    },
    {
      "epoch": 92.31,
      "grad_norm": 7.306400299072266,
      "learning_rate": 6.922922922922923e-06,
      "loss": 0.2531,
      "step": 30740
    },
    {
      "epoch": 92.34,
      "grad_norm": 9.257245063781738,
      "learning_rate": 6.9219219219219225e-06,
      "loss": 0.2893,
      "step": 30750
    },
    {
      "epoch": 92.37,
      "grad_norm": 9.316096305847168,
      "learning_rate": 6.920920920920922e-06,
      "loss": 0.2807,
      "step": 30760
    },
    {
      "epoch": 92.4,
      "grad_norm": 10.110206604003906,
      "learning_rate": 6.9199199199199205e-06,
      "loss": 0.2664,
      "step": 30770
    },
    {
      "epoch": 92.43,
      "grad_norm": 9.103623390197754,
      "learning_rate": 6.91891891891892e-06,
      "loss": 0.3156,
      "step": 30780
    },
    {
      "epoch": 92.46,
      "grad_norm": 7.3321452140808105,
      "learning_rate": 6.917917917917919e-06,
      "loss": 0.2892,
      "step": 30790
    },
    {
      "epoch": 92.49,
      "grad_norm": 8.637866020202637,
      "learning_rate": 6.916916916916917e-06,
      "loss": 0.2938,
      "step": 30800
    },
    {
      "epoch": 92.52,
      "grad_norm": 12.488765716552734,
      "learning_rate": 6.915915915915916e-06,
      "loss": 0.2464,
      "step": 30810
    },
    {
      "epoch": 92.55,
      "grad_norm": 10.727287292480469,
      "learning_rate": 6.914914914914915e-06,
      "loss": 0.2826,
      "step": 30820
    },
    {
      "epoch": 92.58,
      "grad_norm": 5.9482035636901855,
      "learning_rate": 6.913913913913915e-06,
      "loss": 0.2551,
      "step": 30830
    },
    {
      "epoch": 92.61,
      "grad_norm": 31.636600494384766,
      "learning_rate": 6.912912912912913e-06,
      "loss": 0.2994,
      "step": 30840
    },
    {
      "epoch": 92.64,
      "grad_norm": 7.922688007354736,
      "learning_rate": 6.911911911911913e-06,
      "loss": 0.2476,
      "step": 30850
    },
    {
      "epoch": 92.67,
      "grad_norm": 11.354991912841797,
      "learning_rate": 6.910910910910912e-06,
      "loss": 0.2665,
      "step": 30860
    },
    {
      "epoch": 92.7,
      "grad_norm": 8.500253677368164,
      "learning_rate": 6.909909909909911e-06,
      "loss": 0.3012,
      "step": 30870
    },
    {
      "epoch": 92.73,
      "grad_norm": 25.929216384887695,
      "learning_rate": 6.908908908908909e-06,
      "loss": 0.2855,
      "step": 30880
    },
    {
      "epoch": 92.76,
      "grad_norm": 10.42201042175293,
      "learning_rate": 6.907907907907908e-06,
      "loss": 0.2878,
      "step": 30890
    },
    {
      "epoch": 92.79,
      "grad_norm": 12.286577224731445,
      "learning_rate": 6.906906906906907e-06,
      "loss": 0.2899,
      "step": 30900
    },
    {
      "epoch": 92.82,
      "grad_norm": 7.791528701782227,
      "learning_rate": 6.905905905905907e-06,
      "loss": 0.2484,
      "step": 30910
    },
    {
      "epoch": 92.85,
      "grad_norm": 12.248088836669922,
      "learning_rate": 6.904904904904905e-06,
      "loss": 0.2916,
      "step": 30920
    },
    {
      "epoch": 92.88,
      "grad_norm": 7.236509323120117,
      "learning_rate": 6.903903903903905e-06,
      "loss": 0.2887,
      "step": 30930
    },
    {
      "epoch": 92.91,
      "grad_norm": 6.952904224395752,
      "learning_rate": 6.902902902902903e-06,
      "loss": 0.256,
      "step": 30940
    },
    {
      "epoch": 92.94,
      "grad_norm": 10.365106582641602,
      "learning_rate": 6.901901901901903e-06,
      "loss": 0.3306,
      "step": 30950
    },
    {
      "epoch": 92.97,
      "grad_norm": 6.497699737548828,
      "learning_rate": 6.900900900900901e-06,
      "loss": 0.2784,
      "step": 30960
    },
    {
      "epoch": 93.0,
      "eval_accuracy": 0.9273,
      "eval_loss": 0.3011516332626343,
      "eval_runtime": 31.1087,
      "eval_samples_per_second": 321.454,
      "eval_steps_per_second": 1.286,
      "step": 30969
    },
    {
      "epoch": 93.0,
      "grad_norm": 8.13181209564209,
      "learning_rate": 6.8998998998999e-06,
      "loss": 0.2617,
      "step": 30970
    },
    {
      "epoch": 93.03,
      "grad_norm": 11.391140937805176,
      "learning_rate": 6.8988988988988995e-06,
      "loss": 0.2291,
      "step": 30980
    },
    {
      "epoch": 93.06,
      "grad_norm": 13.557493209838867,
      "learning_rate": 6.897897897897898e-06,
      "loss": 0.2559,
      "step": 30990
    },
    {
      "epoch": 93.09,
      "grad_norm": 36.308719635009766,
      "learning_rate": 6.8968968968968975e-06,
      "loss": 0.3107,
      "step": 31000
    },
    {
      "epoch": 93.12,
      "grad_norm": 9.09756851196289,
      "learning_rate": 6.895895895895897e-06,
      "loss": 0.2799,
      "step": 31010
    },
    {
      "epoch": 93.15,
      "grad_norm": 8.696488380432129,
      "learning_rate": 6.8948948948948955e-06,
      "loss": 0.234,
      "step": 31020
    },
    {
      "epoch": 93.18,
      "grad_norm": 6.447506904602051,
      "learning_rate": 6.893893893893895e-06,
      "loss": 0.2809,
      "step": 31030
    },
    {
      "epoch": 93.21,
      "grad_norm": 16.823759078979492,
      "learning_rate": 6.892892892892894e-06,
      "loss": 0.2758,
      "step": 31040
    },
    {
      "epoch": 93.24,
      "grad_norm": 7.406863212585449,
      "learning_rate": 6.891891891891892e-06,
      "loss": 0.2917,
      "step": 31050
    },
    {
      "epoch": 93.27,
      "grad_norm": 13.512945175170898,
      "learning_rate": 6.890890890890891e-06,
      "loss": 0.2889,
      "step": 31060
    },
    {
      "epoch": 93.3,
      "grad_norm": 7.168666362762451,
      "learning_rate": 6.88988988988989e-06,
      "loss": 0.2557,
      "step": 31070
    },
    {
      "epoch": 93.33,
      "grad_norm": 11.11386489868164,
      "learning_rate": 6.88888888888889e-06,
      "loss": 0.2783,
      "step": 31080
    },
    {
      "epoch": 93.36,
      "grad_norm": 6.674493312835693,
      "learning_rate": 6.887887887887888e-06,
      "loss": 0.2832,
      "step": 31090
    },
    {
      "epoch": 93.39,
      "grad_norm": 12.5983304977417,
      "learning_rate": 6.886886886886888e-06,
      "loss": 0.2864,
      "step": 31100
    },
    {
      "epoch": 93.42,
      "grad_norm": 8.410176277160645,
      "learning_rate": 6.885885885885887e-06,
      "loss": 0.3214,
      "step": 31110
    },
    {
      "epoch": 93.45,
      "grad_norm": 5.772496223449707,
      "learning_rate": 6.884884884884886e-06,
      "loss": 0.2657,
      "step": 31120
    },
    {
      "epoch": 93.48,
      "grad_norm": 10.196081161499023,
      "learning_rate": 6.883883883883884e-06,
      "loss": 0.2662,
      "step": 31130
    },
    {
      "epoch": 93.51,
      "grad_norm": 8.689815521240234,
      "learning_rate": 6.882882882882883e-06,
      "loss": 0.2848,
      "step": 31140
    },
    {
      "epoch": 93.54,
      "grad_norm": 14.203797340393066,
      "learning_rate": 6.881881881881882e-06,
      "loss": 0.2912,
      "step": 31150
    },
    {
      "epoch": 93.57,
      "grad_norm": 15.588343620300293,
      "learning_rate": 6.880880880880881e-06,
      "loss": 0.3487,
      "step": 31160
    },
    {
      "epoch": 93.6,
      "grad_norm": 6.798870086669922,
      "learning_rate": 6.87987987987988e-06,
      "loss": 0.2914,
      "step": 31170
    },
    {
      "epoch": 93.63,
      "grad_norm": 8.310786247253418,
      "learning_rate": 6.87887887887888e-06,
      "loss": 0.3143,
      "step": 31180
    },
    {
      "epoch": 93.66,
      "grad_norm": 7.888229846954346,
      "learning_rate": 6.8778778778778784e-06,
      "loss": 0.3114,
      "step": 31190
    },
    {
      "epoch": 93.69,
      "grad_norm": 10.32701301574707,
      "learning_rate": 6.876876876876878e-06,
      "loss": 0.3026,
      "step": 31200
    },
    {
      "epoch": 93.72,
      "grad_norm": 16.000547409057617,
      "learning_rate": 6.875875875875876e-06,
      "loss": 0.2461,
      "step": 31210
    },
    {
      "epoch": 93.75,
      "grad_norm": 10.27876091003418,
      "learning_rate": 6.874874874874875e-06,
      "loss": 0.3071,
      "step": 31220
    },
    {
      "epoch": 93.78,
      "grad_norm": 9.99016284942627,
      "learning_rate": 6.8738738738738745e-06,
      "loss": 0.3009,
      "step": 31230
    },
    {
      "epoch": 93.81,
      "grad_norm": 7.436997890472412,
      "learning_rate": 6.872872872872873e-06,
      "loss": 0.2942,
      "step": 31240
    },
    {
      "epoch": 93.84,
      "grad_norm": 8.05577564239502,
      "learning_rate": 6.8718718718718725e-06,
      "loss": 0.316,
      "step": 31250
    },
    {
      "epoch": 93.87,
      "grad_norm": 10.009994506835938,
      "learning_rate": 6.870870870870872e-06,
      "loss": 0.2838,
      "step": 31260
    },
    {
      "epoch": 93.9,
      "grad_norm": 7.771977424621582,
      "learning_rate": 6.869869869869871e-06,
      "loss": 0.275,
      "step": 31270
    },
    {
      "epoch": 93.93,
      "grad_norm": 9.326883316040039,
      "learning_rate": 6.86886886886887e-06,
      "loss": 0.2814,
      "step": 31280
    },
    {
      "epoch": 93.96,
      "grad_norm": 11.587516784667969,
      "learning_rate": 6.867867867867869e-06,
      "loss": 0.288,
      "step": 31290
    },
    {
      "epoch": 93.99,
      "grad_norm": 17.80978012084961,
      "learning_rate": 6.866866866866867e-06,
      "loss": 0.2397,
      "step": 31300
    },
    {
      "epoch": 94.0,
      "eval_accuracy": 0.9275,
      "eval_loss": 0.2990071177482605,
      "eval_runtime": 30.5116,
      "eval_samples_per_second": 327.744,
      "eval_steps_per_second": 1.311,
      "step": 31302
    },
    {
      "epoch": 94.02,
      "grad_norm": 44.93727493286133,
      "learning_rate": 6.865865865865866e-06,
      "loss": 0.2632,
      "step": 31310
    },
    {
      "epoch": 94.05,
      "grad_norm": 10.342436790466309,
      "learning_rate": 6.864864864864865e-06,
      "loss": 0.2827,
      "step": 31320
    },
    {
      "epoch": 94.08,
      "grad_norm": 8.093668937683105,
      "learning_rate": 6.863863863863865e-06,
      "loss": 0.3047,
      "step": 31330
    },
    {
      "epoch": 94.11,
      "grad_norm": 9.819478034973145,
      "learning_rate": 6.862862862862863e-06,
      "loss": 0.3374,
      "step": 31340
    },
    {
      "epoch": 94.14,
      "grad_norm": 7.377626419067383,
      "learning_rate": 6.861861861861863e-06,
      "loss": 0.2705,
      "step": 31350
    },
    {
      "epoch": 94.17,
      "grad_norm": 9.301630973815918,
      "learning_rate": 6.860860860860862e-06,
      "loss": 0.2736,
      "step": 31360
    },
    {
      "epoch": 94.2,
      "grad_norm": 7.818398475646973,
      "learning_rate": 6.859859859859861e-06,
      "loss": 0.3013,
      "step": 31370
    },
    {
      "epoch": 94.23,
      "grad_norm": 8.01862621307373,
      "learning_rate": 6.8588588588588585e-06,
      "loss": 0.2611,
      "step": 31380
    },
    {
      "epoch": 94.26,
      "grad_norm": 13.315115928649902,
      "learning_rate": 6.857857857857858e-06,
      "loss": 0.2822,
      "step": 31390
    },
    {
      "epoch": 94.29,
      "grad_norm": 6.994703769683838,
      "learning_rate": 6.856856856856857e-06,
      "loss": 0.3226,
      "step": 31400
    },
    {
      "epoch": 94.32,
      "grad_norm": 6.335434913635254,
      "learning_rate": 6.855855855855856e-06,
      "loss": 0.2601,
      "step": 31410
    },
    {
      "epoch": 94.35,
      "grad_norm": 7.086188793182373,
      "learning_rate": 6.8548548548548554e-06,
      "loss": 0.2836,
      "step": 31420
    },
    {
      "epoch": 94.38,
      "grad_norm": 7.510226726531982,
      "learning_rate": 6.853853853853855e-06,
      "loss": 0.2806,
      "step": 31430
    },
    {
      "epoch": 94.41,
      "grad_norm": 13.960613250732422,
      "learning_rate": 6.8528528528528535e-06,
      "loss": 0.2515,
      "step": 31440
    },
    {
      "epoch": 94.44,
      "grad_norm": 10.930683135986328,
      "learning_rate": 6.851851851851853e-06,
      "loss": 0.3113,
      "step": 31450
    },
    {
      "epoch": 94.47,
      "grad_norm": 10.78978157043457,
      "learning_rate": 6.850850850850851e-06,
      "loss": 0.2674,
      "step": 31460
    },
    {
      "epoch": 94.5,
      "grad_norm": 7.306484699249268,
      "learning_rate": 6.84984984984985e-06,
      "loss": 0.3176,
      "step": 31470
    },
    {
      "epoch": 94.53,
      "grad_norm": 10.434814453125,
      "learning_rate": 6.8488488488488495e-06,
      "loss": 0.298,
      "step": 31480
    },
    {
      "epoch": 94.56,
      "grad_norm": 6.851132869720459,
      "learning_rate": 6.847847847847848e-06,
      "loss": 0.2684,
      "step": 31490
    },
    {
      "epoch": 94.59,
      "grad_norm": 12.857870101928711,
      "learning_rate": 6.846846846846848e-06,
      "loss": 0.2799,
      "step": 31500
    },
    {
      "epoch": 94.62,
      "grad_norm": 6.579387664794922,
      "learning_rate": 6.845845845845846e-06,
      "loss": 0.2569,
      "step": 31510
    },
    {
      "epoch": 94.65,
      "grad_norm": 8.868461608886719,
      "learning_rate": 6.844844844844846e-06,
      "loss": 0.2998,
      "step": 31520
    },
    {
      "epoch": 94.68,
      "grad_norm": 12.92994213104248,
      "learning_rate": 6.843843843843845e-06,
      "loss": 0.2865,
      "step": 31530
    },
    {
      "epoch": 94.71,
      "grad_norm": 11.20040512084961,
      "learning_rate": 6.842842842842844e-06,
      "loss": 0.2695,
      "step": 31540
    },
    {
      "epoch": 94.74,
      "grad_norm": 9.780352592468262,
      "learning_rate": 6.841841841841842e-06,
      "loss": 0.3033,
      "step": 31550
    },
    {
      "epoch": 94.77,
      "grad_norm": 9.140556335449219,
      "learning_rate": 6.840840840840841e-06,
      "loss": 0.2336,
      "step": 31560
    },
    {
      "epoch": 94.8,
      "grad_norm": 9.282447814941406,
      "learning_rate": 6.83983983983984e-06,
      "loss": 0.2621,
      "step": 31570
    },
    {
      "epoch": 94.83,
      "grad_norm": 10.349772453308105,
      "learning_rate": 6.83883883883884e-06,
      "loss": 0.2816,
      "step": 31580
    },
    {
      "epoch": 94.86,
      "grad_norm": 26.986980438232422,
      "learning_rate": 6.837837837837838e-06,
      "loss": 0.2985,
      "step": 31590
    },
    {
      "epoch": 94.89,
      "grad_norm": 6.997070789337158,
      "learning_rate": 6.836836836836838e-06,
      "loss": 0.2942,
      "step": 31600
    },
    {
      "epoch": 94.92,
      "grad_norm": 8.59882926940918,
      "learning_rate": 6.835835835835836e-06,
      "loss": 0.3003,
      "step": 31610
    },
    {
      "epoch": 94.95,
      "grad_norm": 10.77139949798584,
      "learning_rate": 6.834834834834836e-06,
      "loss": 0.2781,
      "step": 31620
    },
    {
      "epoch": 94.98,
      "grad_norm": 17.627429962158203,
      "learning_rate": 6.8338338338338336e-06,
      "loss": 0.2789,
      "step": 31630
    },
    {
      "epoch": 95.0,
      "eval_accuracy": 0.9257,
      "eval_loss": 0.3008793592453003,
      "eval_runtime": 30.3135,
      "eval_samples_per_second": 329.886,
      "eval_steps_per_second": 1.32,
      "step": 31635
    },
    {
      "epoch": 95.02,
      "grad_norm": 11.384157180786133,
      "learning_rate": 6.832832832832833e-06,
      "loss": 0.2683,
      "step": 31640
    },
    {
      "epoch": 95.05,
      "grad_norm": 12.668841361999512,
      "learning_rate": 6.8318318318318324e-06,
      "loss": 0.256,
      "step": 31650
    },
    {
      "epoch": 95.08,
      "grad_norm": 9.939920425415039,
      "learning_rate": 6.830830830830831e-06,
      "loss": 0.307,
      "step": 31660
    },
    {
      "epoch": 95.11,
      "grad_norm": 6.073023319244385,
      "learning_rate": 6.8298298298298305e-06,
      "loss": 0.246,
      "step": 31670
    },
    {
      "epoch": 95.14,
      "grad_norm": 8.655766487121582,
      "learning_rate": 6.82882882882883e-06,
      "loss": 0.3151,
      "step": 31680
    },
    {
      "epoch": 95.17,
      "grad_norm": 11.446720123291016,
      "learning_rate": 6.8278278278278285e-06,
      "loss": 0.29,
      "step": 31690
    },
    {
      "epoch": 95.2,
      "grad_norm": 9.134337425231934,
      "learning_rate": 6.826826826826828e-06,
      "loss": 0.3171,
      "step": 31700
    },
    {
      "epoch": 95.23,
      "grad_norm": 14.500204086303711,
      "learning_rate": 6.825825825825826e-06,
      "loss": 0.3285,
      "step": 31710
    },
    {
      "epoch": 95.26,
      "grad_norm": 20.994029998779297,
      "learning_rate": 6.824824824824825e-06,
      "loss": 0.2977,
      "step": 31720
    },
    {
      "epoch": 95.29,
      "grad_norm": 15.590896606445312,
      "learning_rate": 6.823823823823824e-06,
      "loss": 0.3077,
      "step": 31730
    },
    {
      "epoch": 95.32,
      "grad_norm": 16.696096420288086,
      "learning_rate": 6.822822822822823e-06,
      "loss": 0.3004,
      "step": 31740
    },
    {
      "epoch": 95.35,
      "grad_norm": 9.41129207611084,
      "learning_rate": 6.821821821821823e-06,
      "loss": 0.2468,
      "step": 31750
    },
    {
      "epoch": 95.38,
      "grad_norm": 14.188141822814941,
      "learning_rate": 6.820820820820821e-06,
      "loss": 0.3183,
      "step": 31760
    },
    {
      "epoch": 95.41,
      "grad_norm": 6.2552900314331055,
      "learning_rate": 6.819819819819821e-06,
      "loss": 0.2425,
      "step": 31770
    },
    {
      "epoch": 95.44,
      "grad_norm": 12.261460304260254,
      "learning_rate": 6.81881881881882e-06,
      "loss": 0.2743,
      "step": 31780
    },
    {
      "epoch": 95.47,
      "grad_norm": 11.901776313781738,
      "learning_rate": 6.817817817817819e-06,
      "loss": 0.2645,
      "step": 31790
    },
    {
      "epoch": 95.5,
      "grad_norm": 9.6975679397583,
      "learning_rate": 6.816816816816817e-06,
      "loss": 0.302,
      "step": 31800
    },
    {
      "epoch": 95.53,
      "grad_norm": 14.111804008483887,
      "learning_rate": 6.815815815815816e-06,
      "loss": 0.2981,
      "step": 31810
    },
    {
      "epoch": 95.56,
      "grad_norm": 13.735102653503418,
      "learning_rate": 6.814814814814815e-06,
      "loss": 0.2768,
      "step": 31820
    },
    {
      "epoch": 95.59,
      "grad_norm": 10.596446990966797,
      "learning_rate": 6.813813813813815e-06,
      "loss": 0.2851,
      "step": 31830
    },
    {
      "epoch": 95.62,
      "grad_norm": 9.902466773986816,
      "learning_rate": 6.812812812812813e-06,
      "loss": 0.2897,
      "step": 31840
    },
    {
      "epoch": 95.65,
      "grad_norm": 6.794235706329346,
      "learning_rate": 6.811811811811813e-06,
      "loss": 0.2655,
      "step": 31850
    },
    {
      "epoch": 95.68,
      "grad_norm": 7.49955415725708,
      "learning_rate": 6.810810810810811e-06,
      "loss": 0.2984,
      "step": 31860
    },
    {
      "epoch": 95.71,
      "grad_norm": 11.329473495483398,
      "learning_rate": 6.809809809809811e-06,
      "loss": 0.314,
      "step": 31870
    },
    {
      "epoch": 95.74,
      "grad_norm": 12.554167747497559,
      "learning_rate": 6.808808808808809e-06,
      "loss": 0.2553,
      "step": 31880
    },
    {
      "epoch": 95.77,
      "grad_norm": 5.9691948890686035,
      "learning_rate": 6.807807807807808e-06,
      "loss": 0.2174,
      "step": 31890
    },
    {
      "epoch": 95.8,
      "grad_norm": 13.263482093811035,
      "learning_rate": 6.8068068068068075e-06,
      "loss": 0.2922,
      "step": 31900
    },
    {
      "epoch": 95.83,
      "grad_norm": 15.111215591430664,
      "learning_rate": 6.805805805805806e-06,
      "loss": 0.2621,
      "step": 31910
    },
    {
      "epoch": 95.86,
      "grad_norm": 5.736842632293701,
      "learning_rate": 6.8048048048048055e-06,
      "loss": 0.2835,
      "step": 31920
    },
    {
      "epoch": 95.89,
      "grad_norm": 11.550211906433105,
      "learning_rate": 6.803803803803805e-06,
      "loss": 0.3458,
      "step": 31930
    },
    {
      "epoch": 95.92,
      "grad_norm": 7.49605131149292,
      "learning_rate": 6.8028028028028035e-06,
      "loss": 0.2713,
      "step": 31940
    },
    {
      "epoch": 95.95,
      "grad_norm": 10.239762306213379,
      "learning_rate": 6.801801801801803e-06,
      "loss": 0.2981,
      "step": 31950
    },
    {
      "epoch": 95.98,
      "grad_norm": 12.222084999084473,
      "learning_rate": 6.800800800800801e-06,
      "loss": 0.3029,
      "step": 31960
    },
    {
      "epoch": 96.0,
      "eval_accuracy": 0.9257,
      "eval_loss": 0.3023281693458557,
      "eval_runtime": 30.2874,
      "eval_samples_per_second": 330.171,
      "eval_steps_per_second": 1.321,
      "step": 31968
    },
    {
      "epoch": 96.01,
      "grad_norm": 11.418628692626953,
      "learning_rate": 6.7997997997998e-06,
      "loss": 0.2402,
      "step": 31970
    },
    {
      "epoch": 96.04,
      "grad_norm": 6.8892741203308105,
      "learning_rate": 6.798798798798799e-06,
      "loss": 0.2162,
      "step": 31980
    },
    {
      "epoch": 96.07,
      "grad_norm": 9.98530101776123,
      "learning_rate": 6.797797797797798e-06,
      "loss": 0.2817,
      "step": 31990
    },
    {
      "epoch": 96.1,
      "grad_norm": 16.632781982421875,
      "learning_rate": 6.796796796796798e-06,
      "loss": 0.2486,
      "step": 32000
    },
    {
      "epoch": 96.13,
      "grad_norm": 12.279532432556152,
      "learning_rate": 6.795795795795796e-06,
      "loss": 0.2506,
      "step": 32010
    },
    {
      "epoch": 96.16,
      "grad_norm": 10.711172103881836,
      "learning_rate": 6.794794794794796e-06,
      "loss": 0.2545,
      "step": 32020
    },
    {
      "epoch": 96.19,
      "grad_norm": 7.9953083992004395,
      "learning_rate": 6.793793793793795e-06,
      "loss": 0.2696,
      "step": 32030
    },
    {
      "epoch": 96.22,
      "grad_norm": 11.604365348815918,
      "learning_rate": 6.792792792792793e-06,
      "loss": 0.3142,
      "step": 32040
    },
    {
      "epoch": 96.25,
      "grad_norm": 8.295256614685059,
      "learning_rate": 6.791791791791792e-06,
      "loss": 0.2846,
      "step": 32050
    },
    {
      "epoch": 96.28,
      "grad_norm": 10.377599716186523,
      "learning_rate": 6.790790790790791e-06,
      "loss": 0.2604,
      "step": 32060
    },
    {
      "epoch": 96.31,
      "grad_norm": 8.826902389526367,
      "learning_rate": 6.78978978978979e-06,
      "loss": 0.3148,
      "step": 32070
    },
    {
      "epoch": 96.34,
      "grad_norm": 15.091923713684082,
      "learning_rate": 6.788788788788789e-06,
      "loss": 0.3202,
      "step": 32080
    },
    {
      "epoch": 96.37,
      "grad_norm": 19.15727424621582,
      "learning_rate": 6.787787787787788e-06,
      "loss": 0.311,
      "step": 32090
    },
    {
      "epoch": 96.4,
      "grad_norm": 12.143660545349121,
      "learning_rate": 6.786786786786788e-06,
      "loss": 0.2476,
      "step": 32100
    },
    {
      "epoch": 96.43,
      "grad_norm": 10.453089714050293,
      "learning_rate": 6.7857857857857864e-06,
      "loss": 0.2686,
      "step": 32110
    },
    {
      "epoch": 96.46,
      "grad_norm": 10.847125053405762,
      "learning_rate": 6.784784784784786e-06,
      "loss": 0.2905,
      "step": 32120
    },
    {
      "epoch": 96.49,
      "grad_norm": 6.391953468322754,
      "learning_rate": 6.783783783783784e-06,
      "loss": 0.2546,
      "step": 32130
    },
    {
      "epoch": 96.52,
      "grad_norm": 10.588299751281738,
      "learning_rate": 6.782782782782783e-06,
      "loss": 0.2647,
      "step": 32140
    },
    {
      "epoch": 96.55,
      "grad_norm": 8.175736427307129,
      "learning_rate": 6.7817817817817825e-06,
      "loss": 0.2673,
      "step": 32150
    },
    {
      "epoch": 96.58,
      "grad_norm": 10.091588020324707,
      "learning_rate": 6.780780780780781e-06,
      "loss": 0.2305,
      "step": 32160
    },
    {
      "epoch": 96.61,
      "grad_norm": 15.170490264892578,
      "learning_rate": 6.7797797797797805e-06,
      "loss": 0.3087,
      "step": 32170
    },
    {
      "epoch": 96.64,
      "grad_norm": 44.962528228759766,
      "learning_rate": 6.778778778778779e-06,
      "loss": 0.3042,
      "step": 32180
    },
    {
      "epoch": 96.67,
      "grad_norm": 7.372290134429932,
      "learning_rate": 6.777777777777779e-06,
      "loss": 0.284,
      "step": 32190
    },
    {
      "epoch": 96.7,
      "grad_norm": 10.02160358428955,
      "learning_rate": 6.776776776776778e-06,
      "loss": 0.2758,
      "step": 32200
    },
    {
      "epoch": 96.73,
      "grad_norm": 5.951150894165039,
      "learning_rate": 6.775775775775776e-06,
      "loss": 0.3139,
      "step": 32210
    },
    {
      "epoch": 96.76,
      "grad_norm": 74.85804748535156,
      "learning_rate": 6.774774774774775e-06,
      "loss": 0.2747,
      "step": 32220
    },
    {
      "epoch": 96.79,
      "grad_norm": 9.541463851928711,
      "learning_rate": 6.773773773773774e-06,
      "loss": 0.2962,
      "step": 32230
    },
    {
      "epoch": 96.82,
      "grad_norm": 8.234984397888184,
      "learning_rate": 6.772772772772773e-06,
      "loss": 0.2865,
      "step": 32240
    },
    {
      "epoch": 96.85,
      "grad_norm": 8.810842514038086,
      "learning_rate": 6.771771771771773e-06,
      "loss": 0.2575,
      "step": 32250
    },
    {
      "epoch": 96.88,
      "grad_norm": 8.91469669342041,
      "learning_rate": 6.770770770770771e-06,
      "loss": 0.2655,
      "step": 32260
    },
    {
      "epoch": 96.91,
      "grad_norm": 6.032533645629883,
      "learning_rate": 6.769769769769771e-06,
      "loss": 0.2294,
      "step": 32270
    },
    {
      "epoch": 96.94,
      "grad_norm": 9.18570613861084,
      "learning_rate": 6.76876876876877e-06,
      "loss": 0.2724,
      "step": 32280
    },
    {
      "epoch": 96.97,
      "grad_norm": 11.718268394470215,
      "learning_rate": 6.767767767767768e-06,
      "loss": 0.2692,
      "step": 32290
    },
    {
      "epoch": 97.0,
      "grad_norm": 7.533609867095947,
      "learning_rate": 6.7667667667667665e-06,
      "loss": 0.2966,
      "step": 32300
    },
    {
      "epoch": 97.0,
      "eval_accuracy": 0.9273,
      "eval_loss": 0.30069881677627563,
      "eval_runtime": 30.5024,
      "eval_samples_per_second": 327.843,
      "eval_steps_per_second": 1.311,
      "step": 32301
    },
    {
      "epoch": 97.03,
      "grad_norm": 8.62232494354248,
      "learning_rate": 6.765765765765766e-06,
      "loss": 0.2504,
      "step": 32310
    },
    {
      "epoch": 97.06,
      "grad_norm": 8.463385581970215,
      "learning_rate": 6.764764764764765e-06,
      "loss": 0.2832,
      "step": 32320
    },
    {
      "epoch": 97.09,
      "grad_norm": 10.349716186523438,
      "learning_rate": 6.763763763763764e-06,
      "loss": 0.3018,
      "step": 32330
    },
    {
      "epoch": 97.12,
      "grad_norm": 4.514964580535889,
      "learning_rate": 6.7627627627627634e-06,
      "loss": 0.2862,
      "step": 32340
    },
    {
      "epoch": 97.15,
      "grad_norm": 5.379201412200928,
      "learning_rate": 6.761761761761763e-06,
      "loss": 0.257,
      "step": 32350
    },
    {
      "epoch": 97.18,
      "grad_norm": 22.537818908691406,
      "learning_rate": 6.7607607607607615e-06,
      "loss": 0.2689,
      "step": 32360
    },
    {
      "epoch": 97.21,
      "grad_norm": 9.949661254882812,
      "learning_rate": 6.759759759759761e-06,
      "loss": 0.2771,
      "step": 32370
    },
    {
      "epoch": 97.24,
      "grad_norm": 8.90918254852295,
      "learning_rate": 6.758758758758759e-06,
      "loss": 0.2903,
      "step": 32380
    },
    {
      "epoch": 97.27,
      "grad_norm": 15.413347244262695,
      "learning_rate": 6.757757757757758e-06,
      "loss": 0.2877,
      "step": 32390
    },
    {
      "epoch": 97.3,
      "grad_norm": 12.704748153686523,
      "learning_rate": 6.7567567567567575e-06,
      "loss": 0.2714,
      "step": 32400
    },
    {
      "epoch": 97.33,
      "grad_norm": 5.573267936706543,
      "learning_rate": 6.755755755755756e-06,
      "loss": 0.3095,
      "step": 32410
    },
    {
      "epoch": 97.36,
      "grad_norm": 11.858955383300781,
      "learning_rate": 6.7547547547547556e-06,
      "loss": 0.2737,
      "step": 32420
    },
    {
      "epoch": 97.39,
      "grad_norm": 7.781627655029297,
      "learning_rate": 6.753753753753754e-06,
      "loss": 0.2627,
      "step": 32430
    },
    {
      "epoch": 97.42,
      "grad_norm": 10.05484676361084,
      "learning_rate": 6.752752752752754e-06,
      "loss": 0.3193,
      "step": 32440
    },
    {
      "epoch": 97.45,
      "grad_norm": 6.619483947753906,
      "learning_rate": 6.751751751751753e-06,
      "loss": 0.3302,
      "step": 32450
    },
    {
      "epoch": 97.48,
      "grad_norm": 9.129146575927734,
      "learning_rate": 6.750750750750751e-06,
      "loss": 0.269,
      "step": 32460
    },
    {
      "epoch": 97.51,
      "grad_norm": 5.868916988372803,
      "learning_rate": 6.74974974974975e-06,
      "loss": 0.2664,
      "step": 32470
    },
    {
      "epoch": 97.54,
      "grad_norm": 7.557883262634277,
      "learning_rate": 6.748748748748749e-06,
      "loss": 0.2633,
      "step": 32480
    },
    {
      "epoch": 97.57,
      "grad_norm": 10.011571884155273,
      "learning_rate": 6.747747747747748e-06,
      "loss": 0.2475,
      "step": 32490
    },
    {
      "epoch": 97.6,
      "grad_norm": 7.663501739501953,
      "learning_rate": 6.746746746746748e-06,
      "loss": 0.2708,
      "step": 32500
    },
    {
      "epoch": 97.63,
      "grad_norm": 12.160679817199707,
      "learning_rate": 6.745745745745746e-06,
      "loss": 0.2899,
      "step": 32510
    },
    {
      "epoch": 97.66,
      "grad_norm": 7.529998302459717,
      "learning_rate": 6.744744744744746e-06,
      "loss": 0.2341,
      "step": 32520
    },
    {
      "epoch": 97.69,
      "grad_norm": 19.51910972595215,
      "learning_rate": 6.743743743743744e-06,
      "loss": 0.2532,
      "step": 32530
    },
    {
      "epoch": 97.72,
      "grad_norm": 8.872967720031738,
      "learning_rate": 6.742742742742743e-06,
      "loss": 0.2723,
      "step": 32540
    },
    {
      "epoch": 97.75,
      "grad_norm": 7.461001873016357,
      "learning_rate": 6.7417417417417415e-06,
      "loss": 0.2514,
      "step": 32550
    },
    {
      "epoch": 97.78,
      "grad_norm": 7.938113212585449,
      "learning_rate": 6.740740740740741e-06,
      "loss": 0.27,
      "step": 32560
    },
    {
      "epoch": 97.81,
      "grad_norm": 8.518202781677246,
      "learning_rate": 6.7397397397397404e-06,
      "loss": 0.2682,
      "step": 32570
    },
    {
      "epoch": 97.84,
      "grad_norm": 8.658758163452148,
      "learning_rate": 6.738738738738739e-06,
      "loss": 0.1886,
      "step": 32580
    },
    {
      "epoch": 97.87,
      "grad_norm": 8.755126953125,
      "learning_rate": 6.7377377377377385e-06,
      "loss": 0.2491,
      "step": 32590
    },
    {
      "epoch": 97.9,
      "grad_norm": 5.998520374298096,
      "learning_rate": 6.736736736736738e-06,
      "loss": 0.2552,
      "step": 32600
    },
    {
      "epoch": 97.93,
      "grad_norm": 9.903300285339355,
      "learning_rate": 6.7357357357357365e-06,
      "loss": 0.3088,
      "step": 32610
    },
    {
      "epoch": 97.96,
      "grad_norm": 9.938077926635742,
      "learning_rate": 6.734734734734736e-06,
      "loss": 0.2759,
      "step": 32620
    },
    {
      "epoch": 97.99,
      "grad_norm": 7.930237770080566,
      "learning_rate": 6.733733733733734e-06,
      "loss": 0.3114,
      "step": 32630
    },
    {
      "epoch": 98.0,
      "eval_accuracy": 0.9277,
      "eval_loss": 0.2944505512714386,
      "eval_runtime": 30.3474,
      "eval_samples_per_second": 329.517,
      "eval_steps_per_second": 1.318,
      "step": 32634
    },
    {
      "epoch": 98.02,
      "grad_norm": 14.570849418640137,
      "learning_rate": 6.732732732732733e-06,
      "loss": 0.3823,
      "step": 32640
    },
    {
      "epoch": 98.05,
      "grad_norm": 10.072640419006348,
      "learning_rate": 6.731731731731732e-06,
      "loss": 0.3036,
      "step": 32650
    },
    {
      "epoch": 98.08,
      "grad_norm": 10.958309173583984,
      "learning_rate": 6.730730730730731e-06,
      "loss": 0.2547,
      "step": 32660
    },
    {
      "epoch": 98.11,
      "grad_norm": 9.084903717041016,
      "learning_rate": 6.729729729729731e-06,
      "loss": 0.2344,
      "step": 32670
    },
    {
      "epoch": 98.14,
      "grad_norm": 12.275114059448242,
      "learning_rate": 6.728728728728729e-06,
      "loss": 0.321,
      "step": 32680
    },
    {
      "epoch": 98.17,
      "grad_norm": 13.930333137512207,
      "learning_rate": 6.727727727727729e-06,
      "loss": 0.283,
      "step": 32690
    },
    {
      "epoch": 98.2,
      "grad_norm": 7.053762435913086,
      "learning_rate": 6.726726726726728e-06,
      "loss": 0.2569,
      "step": 32700
    },
    {
      "epoch": 98.23,
      "grad_norm": 9.743502616882324,
      "learning_rate": 6.725725725725726e-06,
      "loss": 0.2781,
      "step": 32710
    },
    {
      "epoch": 98.26,
      "grad_norm": 8.969356536865234,
      "learning_rate": 6.724724724724725e-06,
      "loss": 0.3047,
      "step": 32720
    },
    {
      "epoch": 98.29,
      "grad_norm": 8.363102912902832,
      "learning_rate": 6.723723723723724e-06,
      "loss": 0.2763,
      "step": 32730
    },
    {
      "epoch": 98.32,
      "grad_norm": 6.385475158691406,
      "learning_rate": 6.722722722722723e-06,
      "loss": 0.2978,
      "step": 32740
    },
    {
      "epoch": 98.35,
      "grad_norm": 4.19107723236084,
      "learning_rate": 6.721721721721722e-06,
      "loss": 0.2397,
      "step": 32750
    },
    {
      "epoch": 98.38,
      "grad_norm": 12.520075798034668,
      "learning_rate": 6.720720720720721e-06,
      "loss": 0.2781,
      "step": 32760
    },
    {
      "epoch": 98.41,
      "grad_norm": 8.958023071289062,
      "learning_rate": 6.719719719719721e-06,
      "loss": 0.2784,
      "step": 32770
    },
    {
      "epoch": 98.44,
      "grad_norm": 6.202537536621094,
      "learning_rate": 6.718718718718719e-06,
      "loss": 0.2511,
      "step": 32780
    },
    {
      "epoch": 98.47,
      "grad_norm": 6.872201442718506,
      "learning_rate": 6.717717717717718e-06,
      "loss": 0.3201,
      "step": 32790
    },
    {
      "epoch": 98.5,
      "grad_norm": 12.404664993286133,
      "learning_rate": 6.716716716716717e-06,
      "loss": 0.2632,
      "step": 32800
    },
    {
      "epoch": 98.53,
      "grad_norm": 9.05828857421875,
      "learning_rate": 6.715715715715716e-06,
      "loss": 0.2625,
      "step": 32810
    },
    {
      "epoch": 98.56,
      "grad_norm": 6.393959045410156,
      "learning_rate": 6.7147147147147155e-06,
      "loss": 0.2503,
      "step": 32820
    },
    {
      "epoch": 98.59,
      "grad_norm": 8.885269165039062,
      "learning_rate": 6.713713713713714e-06,
      "loss": 0.2998,
      "step": 32830
    },
    {
      "epoch": 98.62,
      "grad_norm": 14.019634246826172,
      "learning_rate": 6.7127127127127135e-06,
      "loss": 0.3011,
      "step": 32840
    },
    {
      "epoch": 98.65,
      "grad_norm": 7.754277229309082,
      "learning_rate": 6.711711711711713e-06,
      "loss": 0.2618,
      "step": 32850
    },
    {
      "epoch": 98.68,
      "grad_norm": 8.421472549438477,
      "learning_rate": 6.7107107107107115e-06,
      "loss": 0.3399,
      "step": 32860
    },
    {
      "epoch": 98.71,
      "grad_norm": 7.4260783195495605,
      "learning_rate": 6.709709709709711e-06,
      "loss": 0.2748,
      "step": 32870
    },
    {
      "epoch": 98.74,
      "grad_norm": 29.457313537597656,
      "learning_rate": 6.708708708708709e-06,
      "loss": 0.2678,
      "step": 32880
    },
    {
      "epoch": 98.77,
      "grad_norm": 9.5269136428833,
      "learning_rate": 6.707707707707708e-06,
      "loss": 0.2914,
      "step": 32890
    },
    {
      "epoch": 98.8,
      "grad_norm": 9.380958557128906,
      "learning_rate": 6.706706706706707e-06,
      "loss": 0.2766,
      "step": 32900
    },
    {
      "epoch": 98.83,
      "grad_norm": 6.4506072998046875,
      "learning_rate": 6.705705705705706e-06,
      "loss": 0.2841,
      "step": 32910
    },
    {
      "epoch": 98.86,
      "grad_norm": 9.98419189453125,
      "learning_rate": 6.704704704704706e-06,
      "loss": 0.3072,
      "step": 32920
    },
    {
      "epoch": 98.89,
      "grad_norm": 10.854504585266113,
      "learning_rate": 6.703703703703704e-06,
      "loss": 0.2649,
      "step": 32930
    },
    {
      "epoch": 98.92,
      "grad_norm": 9.57335376739502,
      "learning_rate": 6.702702702702704e-06,
      "loss": 0.3227,
      "step": 32940
    },
    {
      "epoch": 98.95,
      "grad_norm": 8.38938045501709,
      "learning_rate": 6.701701701701703e-06,
      "loss": 0.2617,
      "step": 32950
    },
    {
      "epoch": 98.98,
      "grad_norm": 7.985697269439697,
      "learning_rate": 6.700700700700701e-06,
      "loss": 0.2892,
      "step": 32960
    },
    {
      "epoch": 99.0,
      "eval_accuracy": 0.9257,
      "eval_loss": 0.30279433727264404,
      "eval_runtime": 30.0667,
      "eval_samples_per_second": 332.593,
      "eval_steps_per_second": 1.33,
      "step": 32967
    },
    {
      "epoch": 99.01,
      "grad_norm": 5.74517822265625,
      "learning_rate": 6.6996996996997e-06,
      "loss": 0.3121,
      "step": 32970
    },
    {
      "epoch": 99.04,
      "grad_norm": 8.184067726135254,
      "learning_rate": 6.698698698698699e-06,
      "loss": 0.3293,
      "step": 32980
    },
    {
      "epoch": 99.07,
      "grad_norm": 12.17577838897705,
      "learning_rate": 6.697697697697698e-06,
      "loss": 0.2667,
      "step": 32990
    },
    {
      "epoch": 99.1,
      "grad_norm": 7.8977251052856445,
      "learning_rate": 6.696696696696697e-06,
      "loss": 0.2521,
      "step": 33000
    },
    {
      "epoch": 99.13,
      "grad_norm": 20.997556686401367,
      "learning_rate": 6.695695695695696e-06,
      "loss": 0.2884,
      "step": 33010
    },
    {
      "epoch": 99.16,
      "grad_norm": 23.104536056518555,
      "learning_rate": 6.694694694694696e-06,
      "loss": 0.2547,
      "step": 33020
    },
    {
      "epoch": 99.19,
      "grad_norm": 7.558016777038574,
      "learning_rate": 6.693693693693694e-06,
      "loss": 0.2418,
      "step": 33030
    },
    {
      "epoch": 99.22,
      "grad_norm": 7.423081398010254,
      "learning_rate": 6.692692692692693e-06,
      "loss": 0.2707,
      "step": 33040
    },
    {
      "epoch": 99.25,
      "grad_norm": 8.073999404907227,
      "learning_rate": 6.691691691691692e-06,
      "loss": 0.2329,
      "step": 33050
    },
    {
      "epoch": 99.28,
      "grad_norm": 19.52268409729004,
      "learning_rate": 6.690690690690691e-06,
      "loss": 0.2911,
      "step": 33060
    },
    {
      "epoch": 99.31,
      "grad_norm": 9.362861633300781,
      "learning_rate": 6.6896896896896905e-06,
      "loss": 0.2724,
      "step": 33070
    },
    {
      "epoch": 99.34,
      "grad_norm": 9.179414749145508,
      "learning_rate": 6.688688688688689e-06,
      "loss": 0.2647,
      "step": 33080
    },
    {
      "epoch": 99.37,
      "grad_norm": 9.351597785949707,
      "learning_rate": 6.6876876876876885e-06,
      "loss": 0.313,
      "step": 33090
    },
    {
      "epoch": 99.4,
      "grad_norm": 9.23470687866211,
      "learning_rate": 6.686686686686687e-06,
      "loss": 0.301,
      "step": 33100
    },
    {
      "epoch": 99.43,
      "grad_norm": 6.840019226074219,
      "learning_rate": 6.6856856856856866e-06,
      "loss": 0.236,
      "step": 33110
    },
    {
      "epoch": 99.46,
      "grad_norm": 13.408376693725586,
      "learning_rate": 6.684684684684684e-06,
      "loss": 0.2759,
      "step": 33120
    },
    {
      "epoch": 99.49,
      "grad_norm": 10.427532196044922,
      "learning_rate": 6.683683683683684e-06,
      "loss": 0.263,
      "step": 33130
    },
    {
      "epoch": 99.52,
      "grad_norm": 14.088022232055664,
      "learning_rate": 6.682682682682683e-06,
      "loss": 0.2621,
      "step": 33140
    },
    {
      "epoch": 99.55,
      "grad_norm": 8.568033218383789,
      "learning_rate": 6.681681681681682e-06,
      "loss": 0.3078,
      "step": 33150
    },
    {
      "epoch": 99.58,
      "grad_norm": 8.8381929397583,
      "learning_rate": 6.680680680680681e-06,
      "loss": 0.2903,
      "step": 33160
    },
    {
      "epoch": 99.61,
      "grad_norm": 16.63536262512207,
      "learning_rate": 6.679679679679681e-06,
      "loss": 0.2974,
      "step": 33170
    },
    {
      "epoch": 99.64,
      "grad_norm": 13.323402404785156,
      "learning_rate": 6.678678678678679e-06,
      "loss": 0.2868,
      "step": 33180
    },
    {
      "epoch": 99.67,
      "grad_norm": 6.976755619049072,
      "learning_rate": 6.677677677677679e-06,
      "loss": 0.2794,
      "step": 33190
    },
    {
      "epoch": 99.7,
      "grad_norm": 5.439624786376953,
      "learning_rate": 6.676676676676678e-06,
      "loss": 0.2679,
      "step": 33200
    },
    {
      "epoch": 99.73,
      "grad_norm": 10.858809471130371,
      "learning_rate": 6.675675675675676e-06,
      "loss": 0.2578,
      "step": 33210
    },
    {
      "epoch": 99.76,
      "grad_norm": 10.361727714538574,
      "learning_rate": 6.6746746746746745e-06,
      "loss": 0.2202,
      "step": 33220
    },
    {
      "epoch": 99.79,
      "grad_norm": 18.96086883544922,
      "learning_rate": 6.673673673673674e-06,
      "loss": 0.3198,
      "step": 33230
    },
    {
      "epoch": 99.82,
      "grad_norm": 13.408867835998535,
      "learning_rate": 6.672672672672673e-06,
      "loss": 0.2509,
      "step": 33240
    },
    {
      "epoch": 99.85,
      "grad_norm": 9.113380432128906,
      "learning_rate": 6.671671671671672e-06,
      "loss": 0.265,
      "step": 33250
    },
    {
      "epoch": 99.88,
      "grad_norm": 5.565963268280029,
      "learning_rate": 6.670670670670671e-06,
      "loss": 0.3009,
      "step": 33260
    },
    {
      "epoch": 99.91,
      "grad_norm": 6.729846000671387,
      "learning_rate": 6.669669669669671e-06,
      "loss": 0.2681,
      "step": 33270
    },
    {
      "epoch": 99.94,
      "grad_norm": 7.855889797210693,
      "learning_rate": 6.6686686686686695e-06,
      "loss": 0.291,
      "step": 33280
    },
    {
      "epoch": 99.97,
      "grad_norm": 8.215636253356934,
      "learning_rate": 6.667667667667668e-06,
      "loss": 0.2959,
      "step": 33290
    },
    {
      "epoch": 100.0,
      "grad_norm": 8.728350639343262,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.248,
      "step": 33300
    },
    {
      "epoch": 100.0,
      "eval_accuracy": 0.9284,
      "eval_loss": 0.29709747433662415,
      "eval_runtime": 30.2306,
      "eval_samples_per_second": 330.791,
      "eval_steps_per_second": 1.323,
      "step": 33300
    },
    {
      "epoch": 100.03,
      "grad_norm": 9.133115768432617,
      "learning_rate": 6.665665665665666e-06,
      "loss": 0.2625,
      "step": 33310
    },
    {
      "epoch": 100.06,
      "grad_norm": 10.710502624511719,
      "learning_rate": 6.664664664664665e-06,
      "loss": 0.2716,
      "step": 33320
    },
    {
      "epoch": 100.09,
      "grad_norm": 7.580682754516602,
      "learning_rate": 6.663663663663664e-06,
      "loss": 0.2743,
      "step": 33330
    },
    {
      "epoch": 100.12,
      "grad_norm": 10.505684852600098,
      "learning_rate": 6.6626626626626636e-06,
      "loss": 0.2777,
      "step": 33340
    },
    {
      "epoch": 100.15,
      "grad_norm": 26.07035255432129,
      "learning_rate": 6.661661661661662e-06,
      "loss": 0.269,
      "step": 33350
    },
    {
      "epoch": 100.18,
      "grad_norm": 6.539912223815918,
      "learning_rate": 6.660660660660662e-06,
      "loss": 0.2808,
      "step": 33360
    },
    {
      "epoch": 100.21,
      "grad_norm": 8.704121589660645,
      "learning_rate": 6.659659659659659e-06,
      "loss": 0.2946,
      "step": 33370
    },
    {
      "epoch": 100.24,
      "grad_norm": 9.868767738342285,
      "learning_rate": 6.658658658658659e-06,
      "loss": 0.2743,
      "step": 33380
    },
    {
      "epoch": 100.27,
      "grad_norm": 5.5747575759887695,
      "learning_rate": 6.657657657657658e-06,
      "loss": 0.2566,
      "step": 33390
    },
    {
      "epoch": 100.3,
      "grad_norm": 6.672573566436768,
      "learning_rate": 6.656656656656657e-06,
      "loss": 0.2264,
      "step": 33400
    },
    {
      "epoch": 100.33,
      "grad_norm": 14.07436752319336,
      "learning_rate": 6.655655655655656e-06,
      "loss": 0.2882,
      "step": 33410
    },
    {
      "epoch": 100.36,
      "grad_norm": 10.414012908935547,
      "learning_rate": 6.654654654654656e-06,
      "loss": 0.3028,
      "step": 33420
    },
    {
      "epoch": 100.39,
      "grad_norm": 8.566722869873047,
      "learning_rate": 6.653653653653654e-06,
      "loss": 0.2522,
      "step": 33430
    },
    {
      "epoch": 100.42,
      "grad_norm": 7.541955471038818,
      "learning_rate": 6.652652652652654e-06,
      "loss": 0.2567,
      "step": 33440
    },
    {
      "epoch": 100.45,
      "grad_norm": 7.703740119934082,
      "learning_rate": 6.651651651651652e-06,
      "loss": 0.2538,
      "step": 33450
    },
    {
      "epoch": 100.48,
      "grad_norm": 26.88205909729004,
      "learning_rate": 6.650650650650651e-06,
      "loss": 0.2897,
      "step": 33460
    },
    {
      "epoch": 100.51,
      "grad_norm": 10.565537452697754,
      "learning_rate": 6.6496496496496495e-06,
      "loss": 0.2451,
      "step": 33470
    },
    {
      "epoch": 100.54,
      "grad_norm": 8.705312728881836,
      "learning_rate": 6.648648648648649e-06,
      "loss": 0.2653,
      "step": 33480
    },
    {
      "epoch": 100.57,
      "grad_norm": 6.9941301345825195,
      "learning_rate": 6.647647647647648e-06,
      "loss": 0.2718,
      "step": 33490
    },
    {
      "epoch": 100.6,
      "grad_norm": 7.9401936531066895,
      "learning_rate": 6.646646646646647e-06,
      "loss": 0.2953,
      "step": 33500
    },
    {
      "epoch": 100.63,
      "grad_norm": 9.394766807556152,
      "learning_rate": 6.6456456456456465e-06,
      "loss": 0.2733,
      "step": 33510
    },
    {
      "epoch": 100.66,
      "grad_norm": 4.355703353881836,
      "learning_rate": 6.644644644644646e-06,
      "loss": 0.2397,
      "step": 33520
    },
    {
      "epoch": 100.69,
      "grad_norm": 18.95008659362793,
      "learning_rate": 6.6436436436436445e-06,
      "loss": 0.2753,
      "step": 33530
    },
    {
      "epoch": 100.72,
      "grad_norm": 7.305892467498779,
      "learning_rate": 6.642642642642643e-06,
      "loss": 0.2787,
      "step": 33540
    },
    {
      "epoch": 100.75,
      "grad_norm": 7.7178802490234375,
      "learning_rate": 6.641641641641642e-06,
      "loss": 0.2511,
      "step": 33550
    },
    {
      "epoch": 100.78,
      "grad_norm": 8.339197158813477,
      "learning_rate": 6.640640640640641e-06,
      "loss": 0.2502,
      "step": 33560
    },
    {
      "epoch": 100.81,
      "grad_norm": 11.612171173095703,
      "learning_rate": 6.63963963963964e-06,
      "loss": 0.2373,
      "step": 33570
    },
    {
      "epoch": 100.84,
      "grad_norm": 9.678927421569824,
      "learning_rate": 6.638638638638639e-06,
      "loss": 0.2832,
      "step": 33580
    },
    {
      "epoch": 100.87,
      "grad_norm": 10.104452133178711,
      "learning_rate": 6.637637637637639e-06,
      "loss": 0.25,
      "step": 33590
    },
    {
      "epoch": 100.9,
      "grad_norm": 16.960325241088867,
      "learning_rate": 6.636636636636637e-06,
      "loss": 0.3137,
      "step": 33600
    },
    {
      "epoch": 100.93,
      "grad_norm": 13.266034126281738,
      "learning_rate": 6.635635635635637e-06,
      "loss": 0.3095,
      "step": 33610
    },
    {
      "epoch": 100.96,
      "grad_norm": 12.649651527404785,
      "learning_rate": 6.634634634634634e-06,
      "loss": 0.258,
      "step": 33620
    },
    {
      "epoch": 100.99,
      "grad_norm": 9.057004928588867,
      "learning_rate": 6.633633633633634e-06,
      "loss": 0.3176,
      "step": 33630
    },
    {
      "epoch": 101.0,
      "eval_accuracy": 0.9264,
      "eval_loss": 0.3050178587436676,
      "eval_runtime": 30.3155,
      "eval_samples_per_second": 329.864,
      "eval_steps_per_second": 1.319,
      "step": 33633
    },
    {
      "epoch": 101.02,
      "grad_norm": 10.12387466430664,
      "learning_rate": 6.632632632632633e-06,
      "loss": 0.3396,
      "step": 33640
    },
    {
      "epoch": 101.05,
      "grad_norm": 5.312917709350586,
      "learning_rate": 6.631631631631632e-06,
      "loss": 0.2705,
      "step": 33650
    },
    {
      "epoch": 101.08,
      "grad_norm": 9.66663932800293,
      "learning_rate": 6.630630630630631e-06,
      "loss": 0.3032,
      "step": 33660
    },
    {
      "epoch": 101.11,
      "grad_norm": 7.7819695472717285,
      "learning_rate": 6.62962962962963e-06,
      "loss": 0.2689,
      "step": 33670
    },
    {
      "epoch": 101.14,
      "grad_norm": 7.772282600402832,
      "learning_rate": 6.628628628628629e-06,
      "loss": 0.2961,
      "step": 33680
    },
    {
      "epoch": 101.17,
      "grad_norm": 7.207189559936523,
      "learning_rate": 6.627627627627629e-06,
      "loss": 0.283,
      "step": 33690
    },
    {
      "epoch": 101.2,
      "grad_norm": 14.43197250366211,
      "learning_rate": 6.626626626626627e-06,
      "loss": 0.2796,
      "step": 33700
    },
    {
      "epoch": 101.23,
      "grad_norm": 13.574539184570312,
      "learning_rate": 6.625625625625626e-06,
      "loss": 0.2553,
      "step": 33710
    },
    {
      "epoch": 101.26,
      "grad_norm": 8.844590187072754,
      "learning_rate": 6.6246246246246246e-06,
      "loss": 0.2587,
      "step": 33720
    },
    {
      "epoch": 101.29,
      "grad_norm": 9.105852127075195,
      "learning_rate": 6.623623623623624e-06,
      "loss": 0.2492,
      "step": 33730
    },
    {
      "epoch": 101.32,
      "grad_norm": 9.189749717712402,
      "learning_rate": 6.6226226226226235e-06,
      "loss": 0.2861,
      "step": 33740
    },
    {
      "epoch": 101.35,
      "grad_norm": 10.341216087341309,
      "learning_rate": 6.621621621621622e-06,
      "loss": 0.3282,
      "step": 33750
    },
    {
      "epoch": 101.38,
      "grad_norm": 10.8491792678833,
      "learning_rate": 6.6206206206206215e-06,
      "loss": 0.2672,
      "step": 33760
    },
    {
      "epoch": 101.41,
      "grad_norm": 22.92510414123535,
      "learning_rate": 6.619619619619621e-06,
      "loss": 0.2349,
      "step": 33770
    },
    {
      "epoch": 101.44,
      "grad_norm": 11.606851577758789,
      "learning_rate": 6.6186186186186195e-06,
      "loss": 0.2885,
      "step": 33780
    },
    {
      "epoch": 101.47,
      "grad_norm": 7.744279384613037,
      "learning_rate": 6.617617617617617e-06,
      "loss": 0.2751,
      "step": 33790
    },
    {
      "epoch": 101.5,
      "grad_norm": 7.444315433502197,
      "learning_rate": 6.616616616616617e-06,
      "loss": 0.2554,
      "step": 33800
    },
    {
      "epoch": 101.53,
      "grad_norm": 8.118093490600586,
      "learning_rate": 6.615615615615616e-06,
      "loss": 0.2624,
      "step": 33810
    },
    {
      "epoch": 101.56,
      "grad_norm": 10.365460395812988,
      "learning_rate": 6.614614614614615e-06,
      "loss": 0.2674,
      "step": 33820
    },
    {
      "epoch": 101.59,
      "grad_norm": 6.6541361808776855,
      "learning_rate": 6.613613613613614e-06,
      "loss": 0.2635,
      "step": 33830
    },
    {
      "epoch": 101.62,
      "grad_norm": 13.592180252075195,
      "learning_rate": 6.612612612612614e-06,
      "loss": 0.2777,
      "step": 33840
    },
    {
      "epoch": 101.65,
      "grad_norm": 11.684847831726074,
      "learning_rate": 6.611611611611612e-06,
      "loss": 0.2913,
      "step": 33850
    },
    {
      "epoch": 101.68,
      "grad_norm": 9.323593139648438,
      "learning_rate": 6.610610610610612e-06,
      "loss": 0.246,
      "step": 33860
    },
    {
      "epoch": 101.71,
      "grad_norm": 14.526183128356934,
      "learning_rate": 6.609609609609609e-06,
      "loss": 0.2857,
      "step": 33870
    },
    {
      "epoch": 101.74,
      "grad_norm": 7.005405426025391,
      "learning_rate": 6.608608608608609e-06,
      "loss": 0.285,
      "step": 33880
    },
    {
      "epoch": 101.77,
      "grad_norm": 8.132481575012207,
      "learning_rate": 6.607607607607608e-06,
      "loss": 0.2799,
      "step": 33890
    },
    {
      "epoch": 101.8,
      "grad_norm": 7.157766342163086,
      "learning_rate": 6.606606606606607e-06,
      "loss": 0.2741,
      "step": 33900
    },
    {
      "epoch": 101.83,
      "grad_norm": 12.151853561401367,
      "learning_rate": 6.605605605605606e-06,
      "loss": 0.2496,
      "step": 33910
    },
    {
      "epoch": 101.86,
      "grad_norm": 8.53442096710205,
      "learning_rate": 6.604604604604605e-06,
      "loss": 0.2671,
      "step": 33920
    },
    {
      "epoch": 101.89,
      "grad_norm": 6.46816873550415,
      "learning_rate": 6.603603603603604e-06,
      "loss": 0.2478,
      "step": 33930
    },
    {
      "epoch": 101.92,
      "grad_norm": 9.947648048400879,
      "learning_rate": 6.602602602602604e-06,
      "loss": 0.2435,
      "step": 33940
    },
    {
      "epoch": 101.95,
      "grad_norm": 8.044540405273438,
      "learning_rate": 6.601601601601602e-06,
      "loss": 0.2357,
      "step": 33950
    },
    {
      "epoch": 101.98,
      "grad_norm": 8.386220932006836,
      "learning_rate": 6.600600600600601e-06,
      "loss": 0.3074,
      "step": 33960
    },
    {
      "epoch": 102.0,
      "eval_accuracy": 0.929,
      "eval_loss": 0.3066052198410034,
      "eval_runtime": 30.2769,
      "eval_samples_per_second": 330.285,
      "eval_steps_per_second": 1.321,
      "step": 33966
    },
    {
      "epoch": 102.01,
      "grad_norm": 7.036721706390381,
      "learning_rate": 6.5995995995996e-06,
      "loss": 0.2169,
      "step": 33970
    },
    {
      "epoch": 102.04,
      "grad_norm": 13.106322288513184,
      "learning_rate": 6.598598598598599e-06,
      "loss": 0.2573,
      "step": 33980
    },
    {
      "epoch": 102.07,
      "grad_norm": 7.228784561157227,
      "learning_rate": 6.5975975975975985e-06,
      "loss": 0.2453,
      "step": 33990
    },
    {
      "epoch": 102.1,
      "grad_norm": 7.014801025390625,
      "learning_rate": 6.596596596596597e-06,
      "loss": 0.2515,
      "step": 34000
    },
    {
      "epoch": 102.13,
      "grad_norm": 14.499686241149902,
      "learning_rate": 6.5955955955955965e-06,
      "loss": 0.2242,
      "step": 34010
    },
    {
      "epoch": 102.16,
      "grad_norm": 14.68889045715332,
      "learning_rate": 6.594594594594595e-06,
      "loss": 0.3023,
      "step": 34020
    },
    {
      "epoch": 102.19,
      "grad_norm": 42.10697555541992,
      "learning_rate": 6.5935935935935946e-06,
      "loss": 0.2908,
      "step": 34030
    },
    {
      "epoch": 102.22,
      "grad_norm": 8.210929870605469,
      "learning_rate": 6.592592592592592e-06,
      "loss": 0.2841,
      "step": 34040
    },
    {
      "epoch": 102.25,
      "grad_norm": 12.025544166564941,
      "learning_rate": 6.591591591591592e-06,
      "loss": 0.2475,
      "step": 34050
    },
    {
      "epoch": 102.28,
      "grad_norm": 8.27448844909668,
      "learning_rate": 6.590590590590591e-06,
      "loss": 0.284,
      "step": 34060
    },
    {
      "epoch": 102.31,
      "grad_norm": 8.365056037902832,
      "learning_rate": 6.58958958958959e-06,
      "loss": 0.2675,
      "step": 34070
    },
    {
      "epoch": 102.34,
      "grad_norm": 9.332198143005371,
      "learning_rate": 6.588588588588589e-06,
      "loss": 0.2451,
      "step": 34080
    },
    {
      "epoch": 102.37,
      "grad_norm": 12.193876266479492,
      "learning_rate": 6.587587587587589e-06,
      "loss": 0.2653,
      "step": 34090
    },
    {
      "epoch": 102.4,
      "grad_norm": 10.680607795715332,
      "learning_rate": 6.586586586586587e-06,
      "loss": 0.2924,
      "step": 34100
    },
    {
      "epoch": 102.43,
      "grad_norm": 10.515753746032715,
      "learning_rate": 6.585585585585587e-06,
      "loss": 0.236,
      "step": 34110
    },
    {
      "epoch": 102.46,
      "grad_norm": 8.639690399169922,
      "learning_rate": 6.5845845845845845e-06,
      "loss": 0.2814,
      "step": 34120
    },
    {
      "epoch": 102.49,
      "grad_norm": 8.181571006774902,
      "learning_rate": 6.583583583583584e-06,
      "loss": 0.2694,
      "step": 34130
    },
    {
      "epoch": 102.52,
      "grad_norm": 8.340597152709961,
      "learning_rate": 6.5825825825825825e-06,
      "loss": 0.2948,
      "step": 34140
    },
    {
      "epoch": 102.55,
      "grad_norm": 6.999747276306152,
      "learning_rate": 6.581581581581582e-06,
      "loss": 0.276,
      "step": 34150
    },
    {
      "epoch": 102.58,
      "grad_norm": 9.805461883544922,
      "learning_rate": 6.580580580580581e-06,
      "loss": 0.2986,
      "step": 34160
    },
    {
      "epoch": 102.61,
      "grad_norm": 9.11960220336914,
      "learning_rate": 6.57957957957958e-06,
      "loss": 0.2775,
      "step": 34170
    },
    {
      "epoch": 102.64,
      "grad_norm": 22.386348724365234,
      "learning_rate": 6.578578578578579e-06,
      "loss": 0.2801,
      "step": 34180
    },
    {
      "epoch": 102.67,
      "grad_norm": 80.8497314453125,
      "learning_rate": 6.577577577577579e-06,
      "loss": 0.2609,
      "step": 34190
    },
    {
      "epoch": 102.7,
      "grad_norm": 19.452957153320312,
      "learning_rate": 6.5765765765765775e-06,
      "loss": 0.2413,
      "step": 34200
    },
    {
      "epoch": 102.73,
      "grad_norm": 10.291293144226074,
      "learning_rate": 6.575575575575576e-06,
      "loss": 0.2761,
      "step": 34210
    },
    {
      "epoch": 102.76,
      "grad_norm": 147.09104919433594,
      "learning_rate": 6.574574574574575e-06,
      "loss": 0.2581,
      "step": 34220
    },
    {
      "epoch": 102.79,
      "grad_norm": 6.565243244171143,
      "learning_rate": 6.573573573573574e-06,
      "loss": 0.2627,
      "step": 34230
    },
    {
      "epoch": 102.82,
      "grad_norm": 6.203887462615967,
      "learning_rate": 6.572572572572573e-06,
      "loss": 0.2227,
      "step": 34240
    },
    {
      "epoch": 102.85,
      "grad_norm": 9.113268852233887,
      "learning_rate": 6.571571571571572e-06,
      "loss": 0.2997,
      "step": 34250
    },
    {
      "epoch": 102.88,
      "grad_norm": 16.351016998291016,
      "learning_rate": 6.5705705705705716e-06,
      "loss": 0.2354,
      "step": 34260
    },
    {
      "epoch": 102.91,
      "grad_norm": 6.0041704177856445,
      "learning_rate": 6.56956956956957e-06,
      "loss": 0.3007,
      "step": 34270
    },
    {
      "epoch": 102.94,
      "grad_norm": 13.255903244018555,
      "learning_rate": 6.56856856856857e-06,
      "loss": 0.3052,
      "step": 34280
    },
    {
      "epoch": 102.97,
      "grad_norm": 15.834759712219238,
      "learning_rate": 6.567567567567567e-06,
      "loss": 0.2901,
      "step": 34290
    },
    {
      "epoch": 103.0,
      "eval_accuracy": 0.9252,
      "eval_loss": 0.30368533730506897,
      "eval_runtime": 30.5203,
      "eval_samples_per_second": 327.651,
      "eval_steps_per_second": 1.311,
      "step": 34299
    },
    {
      "epoch": 103.0,
      "grad_norm": 9.32498836517334,
      "learning_rate": 6.566566566566567e-06,
      "loss": 0.2611,
      "step": 34300
    },
    {
      "epoch": 103.03,
      "grad_norm": 14.848755836486816,
      "learning_rate": 6.565565565565566e-06,
      "loss": 0.2781,
      "step": 34310
    },
    {
      "epoch": 103.06,
      "grad_norm": 17.2694034576416,
      "learning_rate": 6.564564564564565e-06,
      "loss": 0.2746,
      "step": 34320
    },
    {
      "epoch": 103.09,
      "grad_norm": 5.892505168914795,
      "learning_rate": 6.563563563563564e-06,
      "loss": 0.2934,
      "step": 34330
    },
    {
      "epoch": 103.12,
      "grad_norm": 17.312923431396484,
      "learning_rate": 6.562562562562564e-06,
      "loss": 0.2862,
      "step": 34340
    },
    {
      "epoch": 103.15,
      "grad_norm": 7.550317764282227,
      "learning_rate": 6.561561561561562e-06,
      "loss": 0.3039,
      "step": 34350
    },
    {
      "epoch": 103.18,
      "grad_norm": 9.697704315185547,
      "learning_rate": 6.560560560560562e-06,
      "loss": 0.2881,
      "step": 34360
    },
    {
      "epoch": 103.21,
      "grad_norm": 6.1663432121276855,
      "learning_rate": 6.5595595595595595e-06,
      "loss": 0.2722,
      "step": 34370
    },
    {
      "epoch": 103.24,
      "grad_norm": 10.087909698486328,
      "learning_rate": 6.558558558558559e-06,
      "loss": 0.2824,
      "step": 34380
    },
    {
      "epoch": 103.27,
      "grad_norm": 6.75860071182251,
      "learning_rate": 6.5575575575575575e-06,
      "loss": 0.24,
      "step": 34390
    },
    {
      "epoch": 103.3,
      "grad_norm": 7.222005367279053,
      "learning_rate": 6.556556556556557e-06,
      "loss": 0.2186,
      "step": 34400
    },
    {
      "epoch": 103.33,
      "grad_norm": 6.739204406738281,
      "learning_rate": 6.555555555555556e-06,
      "loss": 0.2915,
      "step": 34410
    },
    {
      "epoch": 103.36,
      "grad_norm": 9.655633926391602,
      "learning_rate": 6.554554554554555e-06,
      "loss": 0.2896,
      "step": 34420
    },
    {
      "epoch": 103.39,
      "grad_norm": 13.43899917602539,
      "learning_rate": 6.5535535535535544e-06,
      "loss": 0.2754,
      "step": 34430
    },
    {
      "epoch": 103.42,
      "grad_norm": 8.57459545135498,
      "learning_rate": 6.552552552552554e-06,
      "loss": 0.3165,
      "step": 34440
    },
    {
      "epoch": 103.45,
      "grad_norm": 10.76154613494873,
      "learning_rate": 6.551551551551552e-06,
      "loss": 0.2579,
      "step": 34450
    },
    {
      "epoch": 103.48,
      "grad_norm": 7.143580913543701,
      "learning_rate": 6.550550550550551e-06,
      "loss": 0.2817,
      "step": 34460
    },
    {
      "epoch": 103.51,
      "grad_norm": 7.108366012573242,
      "learning_rate": 6.54954954954955e-06,
      "loss": 0.268,
      "step": 34470
    },
    {
      "epoch": 103.54,
      "grad_norm": 9.682100296020508,
      "learning_rate": 6.548548548548549e-06,
      "loss": 0.2798,
      "step": 34480
    },
    {
      "epoch": 103.57,
      "grad_norm": 6.345441818237305,
      "learning_rate": 6.547547547547548e-06,
      "loss": 0.2566,
      "step": 34490
    },
    {
      "epoch": 103.6,
      "grad_norm": 10.730772018432617,
      "learning_rate": 6.546546546546547e-06,
      "loss": 0.242,
      "step": 34500
    },
    {
      "epoch": 103.63,
      "grad_norm": 10.8158597946167,
      "learning_rate": 6.545545545545547e-06,
      "loss": 0.319,
      "step": 34510
    },
    {
      "epoch": 103.66,
      "grad_norm": 8.104308128356934,
      "learning_rate": 6.544544544544545e-06,
      "loss": 0.3144,
      "step": 34520
    },
    {
      "epoch": 103.69,
      "grad_norm": 11.098527908325195,
      "learning_rate": 6.543543543543545e-06,
      "loss": 0.2726,
      "step": 34530
    },
    {
      "epoch": 103.72,
      "grad_norm": 7.728797912597656,
      "learning_rate": 6.542542542542542e-06,
      "loss": 0.2624,
      "step": 34540
    },
    {
      "epoch": 103.75,
      "grad_norm": 4.381239414215088,
      "learning_rate": 6.541541541541542e-06,
      "loss": 0.2737,
      "step": 34550
    },
    {
      "epoch": 103.78,
      "grad_norm": 7.534073829650879,
      "learning_rate": 6.540540540540541e-06,
      "loss": 0.2681,
      "step": 34560
    },
    {
      "epoch": 103.81,
      "grad_norm": 10.166553497314453,
      "learning_rate": 6.53953953953954e-06,
      "loss": 0.3402,
      "step": 34570
    },
    {
      "epoch": 103.84,
      "grad_norm": 13.805150985717773,
      "learning_rate": 6.538538538538539e-06,
      "loss": 0.2898,
      "step": 34580
    },
    {
      "epoch": 103.87,
      "grad_norm": 10.499022483825684,
      "learning_rate": 6.537537537537538e-06,
      "loss": 0.2899,
      "step": 34590
    },
    {
      "epoch": 103.9,
      "grad_norm": 7.733349323272705,
      "learning_rate": 6.536536536536537e-06,
      "loss": 0.2632,
      "step": 34600
    },
    {
      "epoch": 103.93,
      "grad_norm": 7.6425628662109375,
      "learning_rate": 6.535535535535537e-06,
      "loss": 0.2509,
      "step": 34610
    },
    {
      "epoch": 103.96,
      "grad_norm": 8.15743350982666,
      "learning_rate": 6.5345345345345345e-06,
      "loss": 0.2525,
      "step": 34620
    },
    {
      "epoch": 103.99,
      "grad_norm": 7.228348731994629,
      "learning_rate": 6.533533533533534e-06,
      "loss": 0.3027,
      "step": 34630
    },
    {
      "epoch": 104.0,
      "eval_accuracy": 0.9245,
      "eval_loss": 0.3039300739765167,
      "eval_runtime": 30.3399,
      "eval_samples_per_second": 329.599,
      "eval_steps_per_second": 1.318,
      "step": 34632
    },
    {
      "epoch": 104.02,
      "grad_norm": 10.839275360107422,
      "learning_rate": 6.5325325325325326e-06,
      "loss": 0.2756,
      "step": 34640
    },
    {
      "epoch": 104.05,
      "grad_norm": 8.539299964904785,
      "learning_rate": 6.531531531531532e-06,
      "loss": 0.2779,
      "step": 34650
    },
    {
      "epoch": 104.08,
      "grad_norm": 9.088879585266113,
      "learning_rate": 6.5305305305305314e-06,
      "loss": 0.2238,
      "step": 34660
    },
    {
      "epoch": 104.11,
      "grad_norm": 9.17037582397461,
      "learning_rate": 6.52952952952953e-06,
      "loss": 0.2879,
      "step": 34670
    },
    {
      "epoch": 104.14,
      "grad_norm": 7.568253040313721,
      "learning_rate": 6.5285285285285295e-06,
      "loss": 0.3078,
      "step": 34680
    },
    {
      "epoch": 104.17,
      "grad_norm": 8.752594947814941,
      "learning_rate": 6.527527527527528e-06,
      "loss": 0.3153,
      "step": 34690
    },
    {
      "epoch": 104.2,
      "grad_norm": 74.84830474853516,
      "learning_rate": 6.526526526526527e-06,
      "loss": 0.2533,
      "step": 34700
    },
    {
      "epoch": 104.23,
      "grad_norm": 6.056700229644775,
      "learning_rate": 6.525525525525525e-06,
      "loss": 0.2455,
      "step": 34710
    },
    {
      "epoch": 104.26,
      "grad_norm": 7.825743198394775,
      "learning_rate": 6.524524524524525e-06,
      "loss": 0.2756,
      "step": 34720
    },
    {
      "epoch": 104.29,
      "grad_norm": 12.763605117797852,
      "learning_rate": 6.523523523523524e-06,
      "loss": 0.261,
      "step": 34730
    },
    {
      "epoch": 104.32,
      "grad_norm": 6.954773426055908,
      "learning_rate": 6.522522522522523e-06,
      "loss": 0.249,
      "step": 34740
    },
    {
      "epoch": 104.35,
      "grad_norm": 11.493459701538086,
      "learning_rate": 6.521521521521522e-06,
      "loss": 0.2797,
      "step": 34750
    },
    {
      "epoch": 104.38,
      "grad_norm": 28.412919998168945,
      "learning_rate": 6.520520520520522e-06,
      "loss": 0.3111,
      "step": 34760
    },
    {
      "epoch": 104.41,
      "grad_norm": 11.885354995727539,
      "learning_rate": 6.51951951951952e-06,
      "loss": 0.2689,
      "step": 34770
    },
    {
      "epoch": 104.44,
      "grad_norm": 8.936657905578613,
      "learning_rate": 6.51851851851852e-06,
      "loss": 0.3037,
      "step": 34780
    },
    {
      "epoch": 104.47,
      "grad_norm": 8.514922142028809,
      "learning_rate": 6.517517517517517e-06,
      "loss": 0.2695,
      "step": 34790
    },
    {
      "epoch": 104.5,
      "grad_norm": 11.98552131652832,
      "learning_rate": 6.516516516516517e-06,
      "loss": 0.2747,
      "step": 34800
    },
    {
      "epoch": 104.53,
      "grad_norm": 9.62468147277832,
      "learning_rate": 6.5155155155155155e-06,
      "loss": 0.2899,
      "step": 34810
    },
    {
      "epoch": 104.56,
      "grad_norm": 10.763346672058105,
      "learning_rate": 6.514514514514515e-06,
      "loss": 0.2859,
      "step": 34820
    },
    {
      "epoch": 104.59,
      "grad_norm": 8.471110343933105,
      "learning_rate": 6.513513513513514e-06,
      "loss": 0.2542,
      "step": 34830
    },
    {
      "epoch": 104.62,
      "grad_norm": 18.47028350830078,
      "learning_rate": 6.512512512512513e-06,
      "loss": 0.2967,
      "step": 34840
    },
    {
      "epoch": 104.65,
      "grad_norm": 5.779874324798584,
      "learning_rate": 6.511511511511512e-06,
      "loss": 0.2809,
      "step": 34850
    },
    {
      "epoch": 104.68,
      "grad_norm": 9.269431114196777,
      "learning_rate": 6.510510510510512e-06,
      "loss": 0.2686,
      "step": 34860
    },
    {
      "epoch": 104.71,
      "grad_norm": 14.00922679901123,
      "learning_rate": 6.5095095095095096e-06,
      "loss": 0.2512,
      "step": 34870
    },
    {
      "epoch": 104.74,
      "grad_norm": 13.466287612915039,
      "learning_rate": 6.508508508508509e-06,
      "loss": 0.2714,
      "step": 34880
    },
    {
      "epoch": 104.77,
      "grad_norm": 9.76388931274414,
      "learning_rate": 6.507507507507508e-06,
      "loss": 0.2853,
      "step": 34890
    },
    {
      "epoch": 104.8,
      "grad_norm": 10.132883071899414,
      "learning_rate": 6.506506506506507e-06,
      "loss": 0.2578,
      "step": 34900
    },
    {
      "epoch": 104.83,
      "grad_norm": 6.709754943847656,
      "learning_rate": 6.5055055055055065e-06,
      "loss": 0.2619,
      "step": 34910
    },
    {
      "epoch": 104.86,
      "grad_norm": 9.05151081085205,
      "learning_rate": 6.504504504504505e-06,
      "loss": 0.2759,
      "step": 34920
    },
    {
      "epoch": 104.89,
      "grad_norm": 18.06028938293457,
      "learning_rate": 6.5035035035035045e-06,
      "loss": 0.3007,
      "step": 34930
    },
    {
      "epoch": 104.92,
      "grad_norm": 6.614147663116455,
      "learning_rate": 6.502502502502503e-06,
      "loss": 0.2739,
      "step": 34940
    },
    {
      "epoch": 104.95,
      "grad_norm": 13.732505798339844,
      "learning_rate": 6.501501501501502e-06,
      "loss": 0.2937,
      "step": 34950
    },
    {
      "epoch": 104.98,
      "grad_norm": 6.2821125984191895,
      "learning_rate": 6.5005005005005e-06,
      "loss": 0.3048,
      "step": 34960
    },
    {
      "epoch": 105.0,
      "eval_accuracy": 0.9287,
      "eval_loss": 0.2982769012451172,
      "eval_runtime": 30.3152,
      "eval_samples_per_second": 329.868,
      "eval_steps_per_second": 1.319,
      "step": 34965
    },
    {
      "epoch": 105.02,
      "grad_norm": 9.684813499450684,
      "learning_rate": 6.4994994994995e-06,
      "loss": 0.2453,
      "step": 34970
    },
    {
      "epoch": 105.05,
      "grad_norm": 8.511345863342285,
      "learning_rate": 6.498498498498499e-06,
      "loss": 0.2807,
      "step": 34980
    },
    {
      "epoch": 105.08,
      "grad_norm": 9.757181167602539,
      "learning_rate": 6.497497497497498e-06,
      "loss": 0.2246,
      "step": 34990
    },
    {
      "epoch": 105.11,
      "grad_norm": 13.535529136657715,
      "learning_rate": 6.496496496496497e-06,
      "loss": 0.286,
      "step": 35000
    },
    {
      "epoch": 105.14,
      "grad_norm": 33.40559005737305,
      "learning_rate": 6.495495495495497e-06,
      "loss": 0.3365,
      "step": 35010
    },
    {
      "epoch": 105.17,
      "grad_norm": 6.469178676605225,
      "learning_rate": 6.494494494494495e-06,
      "loss": 0.1869,
      "step": 35020
    },
    {
      "epoch": 105.2,
      "grad_norm": 5.173757076263428,
      "learning_rate": 6.493493493493495e-06,
      "loss": 0.2321,
      "step": 35030
    },
    {
      "epoch": 105.23,
      "grad_norm": 9.591053009033203,
      "learning_rate": 6.4924924924924924e-06,
      "loss": 0.2996,
      "step": 35040
    },
    {
      "epoch": 105.26,
      "grad_norm": 7.327498435974121,
      "learning_rate": 6.491491491491492e-06,
      "loss": 0.2679,
      "step": 35050
    },
    {
      "epoch": 105.29,
      "grad_norm": 7.273467063903809,
      "learning_rate": 6.4904904904904905e-06,
      "loss": 0.2463,
      "step": 35060
    },
    {
      "epoch": 105.32,
      "grad_norm": 7.546713352203369,
      "learning_rate": 6.48948948948949e-06,
      "loss": 0.2854,
      "step": 35070
    },
    {
      "epoch": 105.35,
      "grad_norm": 9.68294620513916,
      "learning_rate": 6.488488488488489e-06,
      "loss": 0.3299,
      "step": 35080
    },
    {
      "epoch": 105.38,
      "grad_norm": 8.180148124694824,
      "learning_rate": 6.487487487487488e-06,
      "loss": 0.2933,
      "step": 35090
    },
    {
      "epoch": 105.41,
      "grad_norm": 9.66993522644043,
      "learning_rate": 6.486486486486487e-06,
      "loss": 0.2731,
      "step": 35100
    },
    {
      "epoch": 105.44,
      "grad_norm": 9.717958450317383,
      "learning_rate": 6.485485485485487e-06,
      "loss": 0.2463,
      "step": 35110
    },
    {
      "epoch": 105.47,
      "grad_norm": 8.248550415039062,
      "learning_rate": 6.484484484484485e-06,
      "loss": 0.2045,
      "step": 35120
    },
    {
      "epoch": 105.5,
      "grad_norm": 14.79832649230957,
      "learning_rate": 6.483483483483484e-06,
      "loss": 0.2395,
      "step": 35130
    },
    {
      "epoch": 105.53,
      "grad_norm": 31.760353088378906,
      "learning_rate": 6.482482482482483e-06,
      "loss": 0.2475,
      "step": 35140
    },
    {
      "epoch": 105.56,
      "grad_norm": 7.3405303955078125,
      "learning_rate": 6.481481481481482e-06,
      "loss": 0.2667,
      "step": 35150
    },
    {
      "epoch": 105.59,
      "grad_norm": 8.566689491271973,
      "learning_rate": 6.480480480480481e-06,
      "loss": 0.2887,
      "step": 35160
    },
    {
      "epoch": 105.62,
      "grad_norm": 13.19359302520752,
      "learning_rate": 6.47947947947948e-06,
      "loss": 0.2487,
      "step": 35170
    },
    {
      "epoch": 105.65,
      "grad_norm": 11.063419342041016,
      "learning_rate": 6.4784784784784796e-06,
      "loss": 0.2987,
      "step": 35180
    },
    {
      "epoch": 105.68,
      "grad_norm": 6.550830364227295,
      "learning_rate": 6.477477477477478e-06,
      "loss": 0.2713,
      "step": 35190
    },
    {
      "epoch": 105.71,
      "grad_norm": 10.293651580810547,
      "learning_rate": 6.476476476476477e-06,
      "loss": 0.301,
      "step": 35200
    },
    {
      "epoch": 105.74,
      "grad_norm": 43.64778137207031,
      "learning_rate": 6.475475475475475e-06,
      "loss": 0.239,
      "step": 35210
    },
    {
      "epoch": 105.77,
      "grad_norm": 17.497909545898438,
      "learning_rate": 6.474474474474475e-06,
      "loss": 0.2887,
      "step": 35220
    },
    {
      "epoch": 105.8,
      "grad_norm": 7.982706546783447,
      "learning_rate": 6.473473473473474e-06,
      "loss": 0.2328,
      "step": 35230
    },
    {
      "epoch": 105.83,
      "grad_norm": 7.981947422027588,
      "learning_rate": 6.472472472472473e-06,
      "loss": 0.248,
      "step": 35240
    },
    {
      "epoch": 105.86,
      "grad_norm": 8.261198997497559,
      "learning_rate": 6.471471471471472e-06,
      "loss": 0.2412,
      "step": 35250
    },
    {
      "epoch": 105.89,
      "grad_norm": 8.579157829284668,
      "learning_rate": 6.470470470470472e-06,
      "loss": 0.2898,
      "step": 35260
    },
    {
      "epoch": 105.92,
      "grad_norm": 5.150553226470947,
      "learning_rate": 6.46946946946947e-06,
      "loss": 0.2395,
      "step": 35270
    },
    {
      "epoch": 105.95,
      "grad_norm": 12.488615036010742,
      "learning_rate": 6.46846846846847e-06,
      "loss": 0.2646,
      "step": 35280
    },
    {
      "epoch": 105.98,
      "grad_norm": 9.796416282653809,
      "learning_rate": 6.4674674674674675e-06,
      "loss": 0.2573,
      "step": 35290
    },
    {
      "epoch": 106.0,
      "eval_accuracy": 0.928,
      "eval_loss": 0.3004063665866852,
      "eval_runtime": 30.4141,
      "eval_samples_per_second": 328.795,
      "eval_steps_per_second": 1.315,
      "step": 35298
    },
    {
      "epoch": 106.01,
      "grad_norm": 22.477924346923828,
      "learning_rate": 6.466466466466467e-06,
      "loss": 0.2455,
      "step": 35300
    },
    {
      "epoch": 106.04,
      "grad_norm": 6.430484294891357,
      "learning_rate": 6.4654654654654655e-06,
      "loss": 0.2598,
      "step": 35310
    },
    {
      "epoch": 106.07,
      "grad_norm": 10.80440616607666,
      "learning_rate": 6.464464464464465e-06,
      "loss": 0.2733,
      "step": 35320
    },
    {
      "epoch": 106.1,
      "grad_norm": 8.089791297912598,
      "learning_rate": 6.463463463463464e-06,
      "loss": 0.2844,
      "step": 35330
    },
    {
      "epoch": 106.13,
      "grad_norm": 7.686271667480469,
      "learning_rate": 6.462462462462463e-06,
      "loss": 0.3049,
      "step": 35340
    },
    {
      "epoch": 106.16,
      "grad_norm": 10.718899726867676,
      "learning_rate": 6.4614614614614624e-06,
      "loss": 0.2783,
      "step": 35350
    },
    {
      "epoch": 106.19,
      "grad_norm": 5.594985008239746,
      "learning_rate": 6.460460460460462e-06,
      "loss": 0.2754,
      "step": 35360
    },
    {
      "epoch": 106.22,
      "grad_norm": 8.789316177368164,
      "learning_rate": 6.45945945945946e-06,
      "loss": 0.2469,
      "step": 35370
    },
    {
      "epoch": 106.25,
      "grad_norm": 30.567750930786133,
      "learning_rate": 6.458458458458458e-06,
      "loss": 0.289,
      "step": 35380
    },
    {
      "epoch": 106.28,
      "grad_norm": 12.413239479064941,
      "learning_rate": 6.457457457457458e-06,
      "loss": 0.2957,
      "step": 35390
    },
    {
      "epoch": 106.31,
      "grad_norm": 15.685585021972656,
      "learning_rate": 6.456456456456457e-06,
      "loss": 0.2762,
      "step": 35400
    },
    {
      "epoch": 106.34,
      "grad_norm": 10.053318977355957,
      "learning_rate": 6.455455455455456e-06,
      "loss": 0.3122,
      "step": 35410
    },
    {
      "epoch": 106.37,
      "grad_norm": 7.502995491027832,
      "learning_rate": 6.454454454454455e-06,
      "loss": 0.3043,
      "step": 35420
    },
    {
      "epoch": 106.4,
      "grad_norm": 9.915437698364258,
      "learning_rate": 6.453453453453455e-06,
      "loss": 0.3288,
      "step": 35430
    },
    {
      "epoch": 106.43,
      "grad_norm": 8.805083274841309,
      "learning_rate": 6.452452452452453e-06,
      "loss": 0.2957,
      "step": 35440
    },
    {
      "epoch": 106.46,
      "grad_norm": 15.331893920898438,
      "learning_rate": 6.451451451451452e-06,
      "loss": 0.2544,
      "step": 35450
    },
    {
      "epoch": 106.49,
      "grad_norm": 12.615826606750488,
      "learning_rate": 6.45045045045045e-06,
      "loss": 0.2831,
      "step": 35460
    },
    {
      "epoch": 106.52,
      "grad_norm": 12.040678977966309,
      "learning_rate": 6.44944944944945e-06,
      "loss": 0.2424,
      "step": 35470
    },
    {
      "epoch": 106.55,
      "grad_norm": 11.511984825134277,
      "learning_rate": 6.448448448448449e-06,
      "loss": 0.2787,
      "step": 35480
    },
    {
      "epoch": 106.58,
      "grad_norm": 6.976409435272217,
      "learning_rate": 6.447447447447448e-06,
      "loss": 0.2812,
      "step": 35490
    },
    {
      "epoch": 106.61,
      "grad_norm": 8.973828315734863,
      "learning_rate": 6.446446446446447e-06,
      "loss": 0.3139,
      "step": 35500
    },
    {
      "epoch": 106.64,
      "grad_norm": 24.109346389770508,
      "learning_rate": 6.445445445445446e-06,
      "loss": 0.2471,
      "step": 35510
    },
    {
      "epoch": 106.67,
      "grad_norm": 13.288726806640625,
      "learning_rate": 6.444444444444445e-06,
      "loss": 0.2625,
      "step": 35520
    },
    {
      "epoch": 106.7,
      "grad_norm": 11.946005821228027,
      "learning_rate": 6.443443443443445e-06,
      "loss": 0.2946,
      "step": 35530
    },
    {
      "epoch": 106.73,
      "grad_norm": 8.492148399353027,
      "learning_rate": 6.4424424424424425e-06,
      "loss": 0.2692,
      "step": 35540
    },
    {
      "epoch": 106.76,
      "grad_norm": 7.895975589752197,
      "learning_rate": 6.441441441441442e-06,
      "loss": 0.3029,
      "step": 35550
    },
    {
      "epoch": 106.79,
      "grad_norm": 8.393953323364258,
      "learning_rate": 6.4404404404404406e-06,
      "loss": 0.2297,
      "step": 35560
    },
    {
      "epoch": 106.82,
      "grad_norm": 12.425718307495117,
      "learning_rate": 6.43943943943944e-06,
      "loss": 0.2934,
      "step": 35570
    },
    {
      "epoch": 106.85,
      "grad_norm": 6.605587482452393,
      "learning_rate": 6.4384384384384394e-06,
      "loss": 0.2504,
      "step": 35580
    },
    {
      "epoch": 106.88,
      "grad_norm": 8.43719482421875,
      "learning_rate": 6.437437437437438e-06,
      "loss": 0.2617,
      "step": 35590
    },
    {
      "epoch": 106.91,
      "grad_norm": 7.005990505218506,
      "learning_rate": 6.4364364364364375e-06,
      "loss": 0.249,
      "step": 35600
    },
    {
      "epoch": 106.94,
      "grad_norm": 8.103694915771484,
      "learning_rate": 6.435435435435436e-06,
      "loss": 0.2833,
      "step": 35610
    },
    {
      "epoch": 106.97,
      "grad_norm": 7.097627639770508,
      "learning_rate": 6.434434434434435e-06,
      "loss": 0.2365,
      "step": 35620
    },
    {
      "epoch": 107.0,
      "grad_norm": 7.680332183837891,
      "learning_rate": 6.433433433433433e-06,
      "loss": 0.2739,
      "step": 35630
    },
    {
      "epoch": 107.0,
      "eval_accuracy": 0.9263,
      "eval_loss": 0.30387455224990845,
      "eval_runtime": 30.3961,
      "eval_samples_per_second": 328.989,
      "eval_steps_per_second": 1.316,
      "step": 35631
    },
    {
      "epoch": 107.03,
      "grad_norm": 7.509460926055908,
      "learning_rate": 6.432432432432433e-06,
      "loss": 0.2444,
      "step": 35640
    },
    {
      "epoch": 107.06,
      "grad_norm": 7.079344749450684,
      "learning_rate": 6.431431431431432e-06,
      "loss": 0.3139,
      "step": 35650
    },
    {
      "epoch": 107.09,
      "grad_norm": 11.213482856750488,
      "learning_rate": 6.430430430430431e-06,
      "loss": 0.2858,
      "step": 35660
    },
    {
      "epoch": 107.12,
      "grad_norm": 8.375226974487305,
      "learning_rate": 6.42942942942943e-06,
      "loss": 0.304,
      "step": 35670
    },
    {
      "epoch": 107.15,
      "grad_norm": 5.8337626457214355,
      "learning_rate": 6.42842842842843e-06,
      "loss": 0.2812,
      "step": 35680
    },
    {
      "epoch": 107.18,
      "grad_norm": 11.302281379699707,
      "learning_rate": 6.427427427427428e-06,
      "loss": 0.2866,
      "step": 35690
    },
    {
      "epoch": 107.21,
      "grad_norm": 9.39327621459961,
      "learning_rate": 6.426426426426427e-06,
      "loss": 0.2584,
      "step": 35700
    },
    {
      "epoch": 107.24,
      "grad_norm": 8.674544334411621,
      "learning_rate": 6.425425425425425e-06,
      "loss": 0.2798,
      "step": 35710
    },
    {
      "epoch": 107.27,
      "grad_norm": 8.52653694152832,
      "learning_rate": 6.424424424424425e-06,
      "loss": 0.2571,
      "step": 35720
    },
    {
      "epoch": 107.3,
      "grad_norm": 6.627228736877441,
      "learning_rate": 6.4234234234234234e-06,
      "loss": 0.2508,
      "step": 35730
    },
    {
      "epoch": 107.33,
      "grad_norm": 9.842439651489258,
      "learning_rate": 6.422422422422423e-06,
      "loss": 0.2734,
      "step": 35740
    },
    {
      "epoch": 107.36,
      "grad_norm": 8.894174575805664,
      "learning_rate": 6.421421421421422e-06,
      "loss": 0.2889,
      "step": 35750
    },
    {
      "epoch": 107.39,
      "grad_norm": 8.68211555480957,
      "learning_rate": 6.420420420420421e-06,
      "loss": 0.2867,
      "step": 35760
    },
    {
      "epoch": 107.42,
      "grad_norm": 8.557766914367676,
      "learning_rate": 6.41941941941942e-06,
      "loss": 0.2828,
      "step": 35770
    },
    {
      "epoch": 107.45,
      "grad_norm": 10.898299217224121,
      "learning_rate": 6.418418418418418e-06,
      "loss": 0.2341,
      "step": 35780
    },
    {
      "epoch": 107.48,
      "grad_norm": 11.56132698059082,
      "learning_rate": 6.4174174174174176e-06,
      "loss": 0.299,
      "step": 35790
    },
    {
      "epoch": 107.51,
      "grad_norm": 7.104824542999268,
      "learning_rate": 6.416416416416417e-06,
      "loss": 0.2897,
      "step": 35800
    },
    {
      "epoch": 107.54,
      "grad_norm": 9.23714828491211,
      "learning_rate": 6.415415415415416e-06,
      "loss": 0.2657,
      "step": 35810
    },
    {
      "epoch": 107.57,
      "grad_norm": 11.958876609802246,
      "learning_rate": 6.414414414414415e-06,
      "loss": 0.2662,
      "step": 35820
    },
    {
      "epoch": 107.6,
      "grad_norm": 8.496806144714355,
      "learning_rate": 6.4134134134134145e-06,
      "loss": 0.2795,
      "step": 35830
    },
    {
      "epoch": 107.63,
      "grad_norm": 21.8720703125,
      "learning_rate": 6.412412412412413e-06,
      "loss": 0.2466,
      "step": 35840
    },
    {
      "epoch": 107.66,
      "grad_norm": 7.861093997955322,
      "learning_rate": 6.4114114114114125e-06,
      "loss": 0.2353,
      "step": 35850
    },
    {
      "epoch": 107.69,
      "grad_norm": 9.041421890258789,
      "learning_rate": 6.410410410410411e-06,
      "loss": 0.3112,
      "step": 35860
    },
    {
      "epoch": 107.72,
      "grad_norm": 13.362008094787598,
      "learning_rate": 6.40940940940941e-06,
      "loss": 0.3017,
      "step": 35870
    },
    {
      "epoch": 107.75,
      "grad_norm": 4.964626312255859,
      "learning_rate": 6.408408408408408e-06,
      "loss": 0.2575,
      "step": 35880
    },
    {
      "epoch": 107.78,
      "grad_norm": 8.395686149597168,
      "learning_rate": 6.407407407407408e-06,
      "loss": 0.2638,
      "step": 35890
    },
    {
      "epoch": 107.81,
      "grad_norm": 12.27729606628418,
      "learning_rate": 6.406406406406407e-06,
      "loss": 0.3211,
      "step": 35900
    },
    {
      "epoch": 107.84,
      "grad_norm": 7.980066776275635,
      "learning_rate": 6.405405405405406e-06,
      "loss": 0.2787,
      "step": 35910
    },
    {
      "epoch": 107.87,
      "grad_norm": 5.118128299713135,
      "learning_rate": 6.404404404404405e-06,
      "loss": 0.2575,
      "step": 35920
    },
    {
      "epoch": 107.9,
      "grad_norm": 6.134520530700684,
      "learning_rate": 6.403403403403405e-06,
      "loss": 0.2788,
      "step": 35930
    },
    {
      "epoch": 107.93,
      "grad_norm": 12.32839584350586,
      "learning_rate": 6.402402402402403e-06,
      "loss": 0.2881,
      "step": 35940
    },
    {
      "epoch": 107.96,
      "grad_norm": 9.371675491333008,
      "learning_rate": 6.401401401401402e-06,
      "loss": 0.2764,
      "step": 35950
    },
    {
      "epoch": 107.99,
      "grad_norm": 13.334471702575684,
      "learning_rate": 6.4004004004004004e-06,
      "loss": 0.2491,
      "step": 35960
    },
    {
      "epoch": 108.0,
      "eval_accuracy": 0.9277,
      "eval_loss": 0.3016342520713806,
      "eval_runtime": 30.1177,
      "eval_samples_per_second": 332.031,
      "eval_steps_per_second": 1.328,
      "step": 35964
    },
    {
      "epoch": 108.02,
      "grad_norm": 13.332372665405273,
      "learning_rate": 6.3993993993994e-06,
      "loss": 0.4171,
      "step": 35970
    },
    {
      "epoch": 108.05,
      "grad_norm": 9.753283500671387,
      "learning_rate": 6.3983983983983985e-06,
      "loss": 0.2442,
      "step": 35980
    },
    {
      "epoch": 108.08,
      "grad_norm": 12.546451568603516,
      "learning_rate": 6.397397397397398e-06,
      "loss": 0.2861,
      "step": 35990
    },
    {
      "epoch": 108.11,
      "grad_norm": 6.848812580108643,
      "learning_rate": 6.396396396396397e-06,
      "loss": 0.267,
      "step": 36000
    },
    {
      "epoch": 108.14,
      "grad_norm": 12.798257827758789,
      "learning_rate": 6.395395395395396e-06,
      "loss": 0.2832,
      "step": 36010
    },
    {
      "epoch": 108.17,
      "grad_norm": 8.41767406463623,
      "learning_rate": 6.394394394394395e-06,
      "loss": 0.2516,
      "step": 36020
    },
    {
      "epoch": 108.2,
      "grad_norm": 8.653974533081055,
      "learning_rate": 6.393393393393393e-06,
      "loss": 0.2716,
      "step": 36030
    },
    {
      "epoch": 108.23,
      "grad_norm": 7.919657230377197,
      "learning_rate": 6.392392392392393e-06,
      "loss": 0.2845,
      "step": 36040
    },
    {
      "epoch": 108.26,
      "grad_norm": 10.917180061340332,
      "learning_rate": 6.391391391391392e-06,
      "loss": 0.3233,
      "step": 36050
    },
    {
      "epoch": 108.29,
      "grad_norm": 6.5265021324157715,
      "learning_rate": 6.390390390390391e-06,
      "loss": 0.2383,
      "step": 36060
    },
    {
      "epoch": 108.32,
      "grad_norm": 10.434696197509766,
      "learning_rate": 6.38938938938939e-06,
      "loss": 0.2678,
      "step": 36070
    },
    {
      "epoch": 108.35,
      "grad_norm": 8.120145797729492,
      "learning_rate": 6.388388388388389e-06,
      "loss": 0.2724,
      "step": 36080
    },
    {
      "epoch": 108.38,
      "grad_norm": 11.494404792785645,
      "learning_rate": 6.387387387387388e-06,
      "loss": 0.2648,
      "step": 36090
    },
    {
      "epoch": 108.41,
      "grad_norm": 8.146100997924805,
      "learning_rate": 6.3863863863863875e-06,
      "loss": 0.2418,
      "step": 36100
    },
    {
      "epoch": 108.44,
      "grad_norm": 4.490426063537598,
      "learning_rate": 6.385385385385386e-06,
      "loss": 0.2642,
      "step": 36110
    },
    {
      "epoch": 108.47,
      "grad_norm": 6.038362979888916,
      "learning_rate": 6.384384384384385e-06,
      "loss": 0.2279,
      "step": 36120
    },
    {
      "epoch": 108.5,
      "grad_norm": 13.34914493560791,
      "learning_rate": 6.383383383383383e-06,
      "loss": 0.3254,
      "step": 36130
    },
    {
      "epoch": 108.53,
      "grad_norm": 7.512332439422607,
      "learning_rate": 6.382382382382383e-06,
      "loss": 0.316,
      "step": 36140
    },
    {
      "epoch": 108.56,
      "grad_norm": 8.170024871826172,
      "learning_rate": 6.381381381381382e-06,
      "loss": 0.2795,
      "step": 36150
    },
    {
      "epoch": 108.59,
      "grad_norm": 9.528767585754395,
      "learning_rate": 6.380380380380381e-06,
      "loss": 0.2789,
      "step": 36160
    },
    {
      "epoch": 108.62,
      "grad_norm": 5.785638332366943,
      "learning_rate": 6.37937937937938e-06,
      "loss": 0.2127,
      "step": 36170
    },
    {
      "epoch": 108.65,
      "grad_norm": 5.684356212615967,
      "learning_rate": 6.378378378378379e-06,
      "loss": 0.2837,
      "step": 36180
    },
    {
      "epoch": 108.68,
      "grad_norm": 35.682456970214844,
      "learning_rate": 6.377377377377378e-06,
      "loss": 0.2812,
      "step": 36190
    },
    {
      "epoch": 108.71,
      "grad_norm": 11.744751930236816,
      "learning_rate": 6.376376376376376e-06,
      "loss": 0.2559,
      "step": 36200
    },
    {
      "epoch": 108.74,
      "grad_norm": 18.65513801574707,
      "learning_rate": 6.3753753753753755e-06,
      "loss": 0.2693,
      "step": 36210
    },
    {
      "epoch": 108.77,
      "grad_norm": 10.153264045715332,
      "learning_rate": 6.374374374374375e-06,
      "loss": 0.2733,
      "step": 36220
    },
    {
      "epoch": 108.8,
      "grad_norm": 6.281116008758545,
      "learning_rate": 6.3733733733733735e-06,
      "loss": 0.2501,
      "step": 36230
    },
    {
      "epoch": 108.83,
      "grad_norm": 8.757946968078613,
      "learning_rate": 6.372372372372373e-06,
      "loss": 0.2907,
      "step": 36240
    },
    {
      "epoch": 108.86,
      "grad_norm": 12.618708610534668,
      "learning_rate": 6.371371371371372e-06,
      "loss": 0.2399,
      "step": 36250
    },
    {
      "epoch": 108.89,
      "grad_norm": 11.470333099365234,
      "learning_rate": 6.370370370370371e-06,
      "loss": 0.2937,
      "step": 36260
    },
    {
      "epoch": 108.92,
      "grad_norm": 8.379623413085938,
      "learning_rate": 6.3693693693693704e-06,
      "loss": 0.2478,
      "step": 36270
    },
    {
      "epoch": 108.95,
      "grad_norm": 13.023411750793457,
      "learning_rate": 6.368368368368368e-06,
      "loss": 0.2966,
      "step": 36280
    },
    {
      "epoch": 108.98,
      "grad_norm": 6.196298122406006,
      "learning_rate": 6.367367367367368e-06,
      "loss": 0.2491,
      "step": 36290
    },
    {
      "epoch": 109.0,
      "eval_accuracy": 0.9272,
      "eval_loss": 0.3024536669254303,
      "eval_runtime": 30.4663,
      "eval_samples_per_second": 328.232,
      "eval_steps_per_second": 1.313,
      "step": 36297
    },
    {
      "epoch": 109.01,
      "grad_norm": 7.113882541656494,
      "learning_rate": 6.366366366366366e-06,
      "loss": 0.3628,
      "step": 36300
    },
    {
      "epoch": 109.04,
      "grad_norm": 15.240525245666504,
      "learning_rate": 6.365365365365366e-06,
      "loss": 0.2282,
      "step": 36310
    },
    {
      "epoch": 109.07,
      "grad_norm": 9.534981727600098,
      "learning_rate": 6.364364364364365e-06,
      "loss": 0.2134,
      "step": 36320
    },
    {
      "epoch": 109.1,
      "grad_norm": 19.850101470947266,
      "learning_rate": 6.363363363363364e-06,
      "loss": 0.2853,
      "step": 36330
    },
    {
      "epoch": 109.13,
      "grad_norm": 9.00008773803711,
      "learning_rate": 6.362362362362363e-06,
      "loss": 0.247,
      "step": 36340
    },
    {
      "epoch": 109.16,
      "grad_norm": 10.873369216918945,
      "learning_rate": 6.361361361361363e-06,
      "loss": 0.2418,
      "step": 36350
    },
    {
      "epoch": 109.19,
      "grad_norm": 11.62416934967041,
      "learning_rate": 6.360360360360361e-06,
      "loss": 0.2615,
      "step": 36360
    },
    {
      "epoch": 109.22,
      "grad_norm": 10.065814971923828,
      "learning_rate": 6.35935935935936e-06,
      "loss": 0.2583,
      "step": 36370
    },
    {
      "epoch": 109.25,
      "grad_norm": 6.271431922912598,
      "learning_rate": 6.358358358358358e-06,
      "loss": 0.2759,
      "step": 36380
    },
    {
      "epoch": 109.28,
      "grad_norm": 7.234034538269043,
      "learning_rate": 6.357357357357358e-06,
      "loss": 0.2928,
      "step": 36390
    },
    {
      "epoch": 109.31,
      "grad_norm": 8.081804275512695,
      "learning_rate": 6.356356356356357e-06,
      "loss": 0.2417,
      "step": 36400
    },
    {
      "epoch": 109.34,
      "grad_norm": 8.412516593933105,
      "learning_rate": 6.355355355355356e-06,
      "loss": 0.2457,
      "step": 36410
    },
    {
      "epoch": 109.37,
      "grad_norm": 8.694528579711914,
      "learning_rate": 6.354354354354355e-06,
      "loss": 0.2697,
      "step": 36420
    },
    {
      "epoch": 109.4,
      "grad_norm": 9.56881332397461,
      "learning_rate": 6.353353353353354e-06,
      "loss": 0.2625,
      "step": 36430
    },
    {
      "epoch": 109.43,
      "grad_norm": 6.816729545593262,
      "learning_rate": 6.352352352352353e-06,
      "loss": 0.2524,
      "step": 36440
    },
    {
      "epoch": 109.46,
      "grad_norm": 7.425090789794922,
      "learning_rate": 6.351351351351351e-06,
      "loss": 0.2543,
      "step": 36450
    },
    {
      "epoch": 109.49,
      "grad_norm": 15.367293357849121,
      "learning_rate": 6.3503503503503505e-06,
      "loss": 0.2532,
      "step": 36460
    },
    {
      "epoch": 109.52,
      "grad_norm": 11.367491722106934,
      "learning_rate": 6.34934934934935e-06,
      "loss": 0.2929,
      "step": 36470
    },
    {
      "epoch": 109.55,
      "grad_norm": 6.371634006500244,
      "learning_rate": 6.3483483483483485e-06,
      "loss": 0.3144,
      "step": 36480
    },
    {
      "epoch": 109.58,
      "grad_norm": 4.545748710632324,
      "learning_rate": 6.347347347347348e-06,
      "loss": 0.2443,
      "step": 36490
    },
    {
      "epoch": 109.61,
      "grad_norm": 10.374063491821289,
      "learning_rate": 6.3463463463463474e-06,
      "loss": 0.2951,
      "step": 36500
    },
    {
      "epoch": 109.64,
      "grad_norm": 9.404377937316895,
      "learning_rate": 6.345345345345346e-06,
      "loss": 0.2813,
      "step": 36510
    },
    {
      "epoch": 109.67,
      "grad_norm": 9.327220916748047,
      "learning_rate": 6.3443443443443455e-06,
      "loss": 0.263,
      "step": 36520
    },
    {
      "epoch": 109.7,
      "grad_norm": 7.7693681716918945,
      "learning_rate": 6.343343343343343e-06,
      "loss": 0.2469,
      "step": 36530
    },
    {
      "epoch": 109.73,
      "grad_norm": 8.68012809753418,
      "learning_rate": 6.342342342342343e-06,
      "loss": 0.2703,
      "step": 36540
    },
    {
      "epoch": 109.76,
      "grad_norm": 7.067342758178711,
      "learning_rate": 6.341341341341341e-06,
      "loss": 0.2428,
      "step": 36550
    },
    {
      "epoch": 109.79,
      "grad_norm": 9.306110382080078,
      "learning_rate": 6.340340340340341e-06,
      "loss": 0.2362,
      "step": 36560
    },
    {
      "epoch": 109.82,
      "grad_norm": 6.1791510581970215,
      "learning_rate": 6.33933933933934e-06,
      "loss": 0.2843,
      "step": 36570
    },
    {
      "epoch": 109.85,
      "grad_norm": 7.0092973709106445,
      "learning_rate": 6.338338338338339e-06,
      "loss": 0.25,
      "step": 36580
    },
    {
      "epoch": 109.88,
      "grad_norm": 7.978646278381348,
      "learning_rate": 6.337337337337338e-06,
      "loss": 0.2552,
      "step": 36590
    },
    {
      "epoch": 109.91,
      "grad_norm": 11.600296020507812,
      "learning_rate": 6.336336336336338e-06,
      "loss": 0.3074,
      "step": 36600
    },
    {
      "epoch": 109.94,
      "grad_norm": 7.322022438049316,
      "learning_rate": 6.335335335335336e-06,
      "loss": 0.2272,
      "step": 36610
    },
    {
      "epoch": 109.97,
      "grad_norm": 8.67291259765625,
      "learning_rate": 6.334334334334335e-06,
      "loss": 0.24,
      "step": 36620
    },
    {
      "epoch": 110.0,
      "grad_norm": 92.77017974853516,
      "learning_rate": 6.333333333333333e-06,
      "loss": 0.291,
      "step": 36630
    },
    {
      "epoch": 110.0,
      "eval_accuracy": 0.9257,
      "eval_loss": 0.3018401563167572,
      "eval_runtime": 30.2119,
      "eval_samples_per_second": 330.995,
      "eval_steps_per_second": 1.324,
      "step": 36630
    },
    {
      "epoch": 110.03,
      "grad_norm": 10.17582893371582,
      "learning_rate": 6.332332332332333e-06,
      "loss": 0.2519,
      "step": 36640
    },
    {
      "epoch": 110.06,
      "grad_norm": 16.82842254638672,
      "learning_rate": 6.3313313313313314e-06,
      "loss": 0.2655,
      "step": 36650
    },
    {
      "epoch": 110.09,
      "grad_norm": 75.23151397705078,
      "learning_rate": 6.330330330330331e-06,
      "loss": 0.2537,
      "step": 36660
    },
    {
      "epoch": 110.12,
      "grad_norm": 18.17453956604004,
      "learning_rate": 6.32932932932933e-06,
      "loss": 0.3517,
      "step": 36670
    },
    {
      "epoch": 110.15,
      "grad_norm": 10.724184036254883,
      "learning_rate": 6.328328328328329e-06,
      "loss": 0.2901,
      "step": 36680
    },
    {
      "epoch": 110.18,
      "grad_norm": 5.396843910217285,
      "learning_rate": 6.327327327327328e-06,
      "loss": 0.2345,
      "step": 36690
    },
    {
      "epoch": 110.21,
      "grad_norm": 8.100613594055176,
      "learning_rate": 6.326326326326326e-06,
      "loss": 0.3274,
      "step": 36700
    },
    {
      "epoch": 110.24,
      "grad_norm": 9.576359748840332,
      "learning_rate": 6.3253253253253255e-06,
      "loss": 0.2769,
      "step": 36710
    },
    {
      "epoch": 110.27,
      "grad_norm": 8.627700805664062,
      "learning_rate": 6.324324324324325e-06,
      "loss": 0.2627,
      "step": 36720
    },
    {
      "epoch": 110.3,
      "grad_norm": 7.042660236358643,
      "learning_rate": 6.323323323323324e-06,
      "loss": 0.2804,
      "step": 36730
    },
    {
      "epoch": 110.33,
      "grad_norm": 10.036458969116211,
      "learning_rate": 6.322322322322323e-06,
      "loss": 0.2332,
      "step": 36740
    },
    {
      "epoch": 110.36,
      "grad_norm": 7.6549577713012695,
      "learning_rate": 6.321321321321322e-06,
      "loss": 0.2557,
      "step": 36750
    },
    {
      "epoch": 110.39,
      "grad_norm": 7.428267002105713,
      "learning_rate": 6.320320320320321e-06,
      "loss": 0.2243,
      "step": 36760
    },
    {
      "epoch": 110.42,
      "grad_norm": 10.294143676757812,
      "learning_rate": 6.3193193193193205e-06,
      "loss": 0.2604,
      "step": 36770
    },
    {
      "epoch": 110.45,
      "grad_norm": 9.73356819152832,
      "learning_rate": 6.318318318318318e-06,
      "loss": 0.281,
      "step": 36780
    },
    {
      "epoch": 110.48,
      "grad_norm": 5.904912948608398,
      "learning_rate": 6.317317317317318e-06,
      "loss": 0.2678,
      "step": 36790
    },
    {
      "epoch": 110.51,
      "grad_norm": 7.014834403991699,
      "learning_rate": 6.316316316316316e-06,
      "loss": 0.2878,
      "step": 36800
    },
    {
      "epoch": 110.54,
      "grad_norm": 8.551284790039062,
      "learning_rate": 6.315315315315316e-06,
      "loss": 0.2466,
      "step": 36810
    },
    {
      "epoch": 110.57,
      "grad_norm": 8.658544540405273,
      "learning_rate": 6.314314314314315e-06,
      "loss": 0.28,
      "step": 36820
    },
    {
      "epoch": 110.6,
      "grad_norm": 9.550053596496582,
      "learning_rate": 6.313313313313314e-06,
      "loss": 0.2835,
      "step": 36830
    },
    {
      "epoch": 110.63,
      "grad_norm": 8.523709297180176,
      "learning_rate": 6.312312312312313e-06,
      "loss": 0.2687,
      "step": 36840
    },
    {
      "epoch": 110.66,
      "grad_norm": 9.65658950805664,
      "learning_rate": 6.311311311311313e-06,
      "loss": 0.2414,
      "step": 36850
    },
    {
      "epoch": 110.69,
      "grad_norm": 8.401023864746094,
      "learning_rate": 6.31031031031031e-06,
      "loss": 0.2509,
      "step": 36860
    },
    {
      "epoch": 110.72,
      "grad_norm": 7.608580112457275,
      "learning_rate": 6.309309309309309e-06,
      "loss": 0.2507,
      "step": 36870
    },
    {
      "epoch": 110.75,
      "grad_norm": 7.5450663566589355,
      "learning_rate": 6.3083083083083084e-06,
      "loss": 0.2611,
      "step": 36880
    },
    {
      "epoch": 110.78,
      "grad_norm": 10.846982955932617,
      "learning_rate": 6.307307307307308e-06,
      "loss": 0.2553,
      "step": 36890
    },
    {
      "epoch": 110.81,
      "grad_norm": 10.077884674072266,
      "learning_rate": 6.3063063063063065e-06,
      "loss": 0.3046,
      "step": 36900
    },
    {
      "epoch": 110.84,
      "grad_norm": 13.551126480102539,
      "learning_rate": 6.305305305305306e-06,
      "loss": 0.2394,
      "step": 36910
    },
    {
      "epoch": 110.87,
      "grad_norm": 10.279696464538574,
      "learning_rate": 6.304304304304305e-06,
      "loss": 0.2364,
      "step": 36920
    },
    {
      "epoch": 110.9,
      "grad_norm": 11.258269309997559,
      "learning_rate": 6.303303303303304e-06,
      "loss": 0.309,
      "step": 36930
    },
    {
      "epoch": 110.93,
      "grad_norm": 13.547099113464355,
      "learning_rate": 6.302302302302303e-06,
      "loss": 0.2289,
      "step": 36940
    },
    {
      "epoch": 110.96,
      "grad_norm": 10.870450973510742,
      "learning_rate": 6.301301301301301e-06,
      "loss": 0.252,
      "step": 36950
    },
    {
      "epoch": 110.99,
      "grad_norm": 12.590580940246582,
      "learning_rate": 6.300300300300301e-06,
      "loss": 0.264,
      "step": 36960
    },
    {
      "epoch": 111.0,
      "eval_accuracy": 0.9255,
      "eval_loss": 0.30961936712265015,
      "eval_runtime": 30.3112,
      "eval_samples_per_second": 329.911,
      "eval_steps_per_second": 1.32,
      "step": 36963
    },
    {
      "epoch": 111.02,
      "grad_norm": 9.723773002624512,
      "learning_rate": 6.2992992992993e-06,
      "loss": 0.2342,
      "step": 36970
    },
    {
      "epoch": 111.05,
      "grad_norm": 24.76806640625,
      "learning_rate": 6.298298298298299e-06,
      "loss": 0.2389,
      "step": 36980
    },
    {
      "epoch": 111.08,
      "grad_norm": 9.734393119812012,
      "learning_rate": 6.297297297297298e-06,
      "loss": 0.2049,
      "step": 36990
    },
    {
      "epoch": 111.11,
      "grad_norm": 8.151266098022461,
      "learning_rate": 6.296296296296297e-06,
      "loss": 0.2848,
      "step": 37000
    },
    {
      "epoch": 111.14,
      "grad_norm": 12.509382247924805,
      "learning_rate": 6.295295295295296e-06,
      "loss": 0.3084,
      "step": 37010
    },
    {
      "epoch": 111.17,
      "grad_norm": 7.176764488220215,
      "learning_rate": 6.2942942942942955e-06,
      "loss": 0.2761,
      "step": 37020
    },
    {
      "epoch": 111.2,
      "grad_norm": 19.31537628173828,
      "learning_rate": 6.293293293293293e-06,
      "loss": 0.2456,
      "step": 37030
    },
    {
      "epoch": 111.23,
      "grad_norm": 9.653488159179688,
      "learning_rate": 6.292292292292293e-06,
      "loss": 0.2208,
      "step": 37040
    },
    {
      "epoch": 111.26,
      "grad_norm": 8.579176902770996,
      "learning_rate": 6.291291291291291e-06,
      "loss": 0.2705,
      "step": 37050
    },
    {
      "epoch": 111.29,
      "grad_norm": 6.012103080749512,
      "learning_rate": 6.290290290290291e-06,
      "loss": 0.2599,
      "step": 37060
    },
    {
      "epoch": 111.32,
      "grad_norm": 7.07248592376709,
      "learning_rate": 6.28928928928929e-06,
      "loss": 0.2285,
      "step": 37070
    },
    {
      "epoch": 111.35,
      "grad_norm": 8.732285499572754,
      "learning_rate": 6.288288288288289e-06,
      "loss": 0.2468,
      "step": 37080
    },
    {
      "epoch": 111.38,
      "grad_norm": 7.686263561248779,
      "learning_rate": 6.287287287287288e-06,
      "loss": 0.2641,
      "step": 37090
    },
    {
      "epoch": 111.41,
      "grad_norm": 23.046140670776367,
      "learning_rate": 6.286286286286287e-06,
      "loss": 0.2808,
      "step": 37100
    },
    {
      "epoch": 111.44,
      "grad_norm": 9.644116401672363,
      "learning_rate": 6.2852852852852854e-06,
      "loss": 0.2378,
      "step": 37110
    },
    {
      "epoch": 111.47,
      "grad_norm": 11.043275833129883,
      "learning_rate": 6.284284284284284e-06,
      "loss": 0.2432,
      "step": 37120
    },
    {
      "epoch": 111.5,
      "grad_norm": 13.334775924682617,
      "learning_rate": 6.2832832832832835e-06,
      "loss": 0.2606,
      "step": 37130
    },
    {
      "epoch": 111.53,
      "grad_norm": 11.83115291595459,
      "learning_rate": 6.282282282282283e-06,
      "loss": 0.2342,
      "step": 37140
    },
    {
      "epoch": 111.56,
      "grad_norm": 8.118093490600586,
      "learning_rate": 6.2812812812812815e-06,
      "loss": 0.2348,
      "step": 37150
    },
    {
      "epoch": 111.59,
      "grad_norm": 8.039462089538574,
      "learning_rate": 6.280280280280281e-06,
      "loss": 0.2811,
      "step": 37160
    },
    {
      "epoch": 111.62,
      "grad_norm": 10.979276657104492,
      "learning_rate": 6.27927927927928e-06,
      "loss": 0.3024,
      "step": 37170
    },
    {
      "epoch": 111.65,
      "grad_norm": 10.494451522827148,
      "learning_rate": 6.278278278278279e-06,
      "loss": 0.238,
      "step": 37180
    },
    {
      "epoch": 111.68,
      "grad_norm": 50.654876708984375,
      "learning_rate": 6.2772772772772784e-06,
      "loss": 0.3162,
      "step": 37190
    },
    {
      "epoch": 111.71,
      "grad_norm": 10.233278274536133,
      "learning_rate": 6.276276276276276e-06,
      "loss": 0.2543,
      "step": 37200
    },
    {
      "epoch": 111.74,
      "grad_norm": 16.490314483642578,
      "learning_rate": 6.275275275275276e-06,
      "loss": 0.251,
      "step": 37210
    },
    {
      "epoch": 111.77,
      "grad_norm": 18.70013427734375,
      "learning_rate": 6.274274274274274e-06,
      "loss": 0.2448,
      "step": 37220
    },
    {
      "epoch": 111.8,
      "grad_norm": 9.607115745544434,
      "learning_rate": 6.273273273273274e-06,
      "loss": 0.2879,
      "step": 37230
    },
    {
      "epoch": 111.83,
      "grad_norm": 10.99588680267334,
      "learning_rate": 6.272272272272273e-06,
      "loss": 0.2745,
      "step": 37240
    },
    {
      "epoch": 111.86,
      "grad_norm": 7.96452522277832,
      "learning_rate": 6.271271271271272e-06,
      "loss": 0.2906,
      "step": 37250
    },
    {
      "epoch": 111.89,
      "grad_norm": 7.671411514282227,
      "learning_rate": 6.270270270270271e-06,
      "loss": 0.239,
      "step": 37260
    },
    {
      "epoch": 111.92,
      "grad_norm": 16.43231964111328,
      "learning_rate": 6.2692692692692706e-06,
      "loss": 0.2691,
      "step": 37270
    },
    {
      "epoch": 111.95,
      "grad_norm": 11.086589813232422,
      "learning_rate": 6.268268268268268e-06,
      "loss": 0.2354,
      "step": 37280
    },
    {
      "epoch": 111.98,
      "grad_norm": 11.175224304199219,
      "learning_rate": 6.267267267267268e-06,
      "loss": 0.2931,
      "step": 37290
    },
    {
      "epoch": 112.0,
      "eval_accuracy": 0.9282,
      "eval_loss": 0.3092724084854126,
      "eval_runtime": 30.5726,
      "eval_samples_per_second": 327.09,
      "eval_steps_per_second": 1.308,
      "step": 37296
    },
    {
      "epoch": 112.01,
      "grad_norm": 11.362544059753418,
      "learning_rate": 6.266266266266266e-06,
      "loss": 0.2403,
      "step": 37300
    },
    {
      "epoch": 112.04,
      "grad_norm": 8.58780288696289,
      "learning_rate": 6.265265265265266e-06,
      "loss": 0.245,
      "step": 37310
    },
    {
      "epoch": 112.07,
      "grad_norm": 13.298986434936523,
      "learning_rate": 6.264264264264264e-06,
      "loss": 0.2391,
      "step": 37320
    },
    {
      "epoch": 112.1,
      "grad_norm": 16.34769630432129,
      "learning_rate": 6.263263263263264e-06,
      "loss": 0.2693,
      "step": 37330
    },
    {
      "epoch": 112.13,
      "grad_norm": 6.3154754638671875,
      "learning_rate": 6.262262262262263e-06,
      "loss": 0.275,
      "step": 37340
    },
    {
      "epoch": 112.16,
      "grad_norm": 8.140006065368652,
      "learning_rate": 6.261261261261262e-06,
      "loss": 0.2544,
      "step": 37350
    },
    {
      "epoch": 112.19,
      "grad_norm": 9.620804786682129,
      "learning_rate": 6.2602602602602605e-06,
      "loss": 0.2679,
      "step": 37360
    },
    {
      "epoch": 112.22,
      "grad_norm": 8.943511009216309,
      "learning_rate": 6.259259259259259e-06,
      "loss": 0.2664,
      "step": 37370
    },
    {
      "epoch": 112.25,
      "grad_norm": 10.687153816223145,
      "learning_rate": 6.2582582582582585e-06,
      "loss": 0.2448,
      "step": 37380
    },
    {
      "epoch": 112.28,
      "grad_norm": 10.60282039642334,
      "learning_rate": 6.257257257257258e-06,
      "loss": 0.3233,
      "step": 37390
    },
    {
      "epoch": 112.31,
      "grad_norm": 6.480803489685059,
      "learning_rate": 6.2562562562562565e-06,
      "loss": 0.257,
      "step": 37400
    },
    {
      "epoch": 112.34,
      "grad_norm": 7.788878917694092,
      "learning_rate": 6.255255255255256e-06,
      "loss": 0.259,
      "step": 37410
    },
    {
      "epoch": 112.37,
      "grad_norm": 11.828871726989746,
      "learning_rate": 6.2542542542542554e-06,
      "loss": 0.2291,
      "step": 37420
    },
    {
      "epoch": 112.4,
      "grad_norm": 10.823221206665039,
      "learning_rate": 6.253253253253254e-06,
      "loss": 0.272,
      "step": 37430
    },
    {
      "epoch": 112.43,
      "grad_norm": 6.257399559020996,
      "learning_rate": 6.2522522522522535e-06,
      "loss": 0.2452,
      "step": 37440
    },
    {
      "epoch": 112.46,
      "grad_norm": 6.918508052825928,
      "learning_rate": 6.251251251251251e-06,
      "loss": 0.2991,
      "step": 37450
    },
    {
      "epoch": 112.49,
      "grad_norm": 6.961529731750488,
      "learning_rate": 6.250250250250251e-06,
      "loss": 0.2273,
      "step": 37460
    },
    {
      "epoch": 112.52,
      "grad_norm": 9.322816848754883,
      "learning_rate": 6.249249249249249e-06,
      "loss": 0.2448,
      "step": 37470
    },
    {
      "epoch": 112.55,
      "grad_norm": 8.618579864501953,
      "learning_rate": 6.248248248248249e-06,
      "loss": 0.2423,
      "step": 37480
    },
    {
      "epoch": 112.58,
      "grad_norm": 6.894775390625,
      "learning_rate": 6.247247247247248e-06,
      "loss": 0.2084,
      "step": 37490
    },
    {
      "epoch": 112.61,
      "grad_norm": 15.428376197814941,
      "learning_rate": 6.246246246246247e-06,
      "loss": 0.2858,
      "step": 37500
    },
    {
      "epoch": 112.64,
      "grad_norm": 8.944445610046387,
      "learning_rate": 6.245245245245246e-06,
      "loss": 0.2291,
      "step": 37510
    },
    {
      "epoch": 112.67,
      "grad_norm": 9.761090278625488,
      "learning_rate": 6.244244244244246e-06,
      "loss": 0.2289,
      "step": 37520
    },
    {
      "epoch": 112.7,
      "grad_norm": 16.153339385986328,
      "learning_rate": 6.243243243243243e-06,
      "loss": 0.2721,
      "step": 37530
    },
    {
      "epoch": 112.73,
      "grad_norm": 11.7510347366333,
      "learning_rate": 6.242242242242243e-06,
      "loss": 0.2428,
      "step": 37540
    },
    {
      "epoch": 112.76,
      "grad_norm": 8.141000747680664,
      "learning_rate": 6.241241241241241e-06,
      "loss": 0.2531,
      "step": 37550
    },
    {
      "epoch": 112.79,
      "grad_norm": 6.3756608963012695,
      "learning_rate": 6.240240240240241e-06,
      "loss": 0.2779,
      "step": 37560
    },
    {
      "epoch": 112.82,
      "grad_norm": 13.401786804199219,
      "learning_rate": 6.2392392392392394e-06,
      "loss": 0.2206,
      "step": 37570
    },
    {
      "epoch": 112.85,
      "grad_norm": 9.27122688293457,
      "learning_rate": 6.238238238238239e-06,
      "loss": 0.2678,
      "step": 37580
    },
    {
      "epoch": 112.88,
      "grad_norm": 8.414907455444336,
      "learning_rate": 6.237237237237238e-06,
      "loss": 0.2539,
      "step": 37590
    },
    {
      "epoch": 112.91,
      "grad_norm": 12.065780639648438,
      "learning_rate": 6.236236236236237e-06,
      "loss": 0.2518,
      "step": 37600
    },
    {
      "epoch": 112.94,
      "grad_norm": 14.420062065124512,
      "learning_rate": 6.2352352352352355e-06,
      "loss": 0.2689,
      "step": 37610
    },
    {
      "epoch": 112.97,
      "grad_norm": 7.92146635055542,
      "learning_rate": 6.234234234234234e-06,
      "loss": 0.2407,
      "step": 37620
    },
    {
      "epoch": 113.0,
      "eval_accuracy": 0.927,
      "eval_loss": 0.3106383681297302,
      "eval_runtime": 30.4722,
      "eval_samples_per_second": 328.168,
      "eval_steps_per_second": 1.313,
      "step": 37629
    },
    {
      "epoch": 113.0,
      "grad_norm": 5.802934646606445,
      "learning_rate": 6.2332332332332335e-06,
      "loss": 0.3405,
      "step": 37630
    },
    {
      "epoch": 113.03,
      "grad_norm": 14.923483848571777,
      "learning_rate": 6.232232232232233e-06,
      "loss": 0.2603,
      "step": 37640
    },
    {
      "epoch": 113.06,
      "grad_norm": 4.4953532218933105,
      "learning_rate": 6.2312312312312316e-06,
      "loss": 0.2389,
      "step": 37650
    },
    {
      "epoch": 113.09,
      "grad_norm": 23.907745361328125,
      "learning_rate": 6.230230230230231e-06,
      "loss": 0.2888,
      "step": 37660
    },
    {
      "epoch": 113.12,
      "grad_norm": 14.52160358428955,
      "learning_rate": 6.22922922922923e-06,
      "loss": 0.2813,
      "step": 37670
    },
    {
      "epoch": 113.15,
      "grad_norm": 7.409430980682373,
      "learning_rate": 6.228228228228229e-06,
      "loss": 0.2358,
      "step": 37680
    },
    {
      "epoch": 113.18,
      "grad_norm": 14.513826370239258,
      "learning_rate": 6.2272272272272285e-06,
      "loss": 0.2429,
      "step": 37690
    },
    {
      "epoch": 113.21,
      "grad_norm": 8.91285514831543,
      "learning_rate": 6.226226226226226e-06,
      "loss": 0.2324,
      "step": 37700
    },
    {
      "epoch": 113.24,
      "grad_norm": 15.080068588256836,
      "learning_rate": 6.225225225225226e-06,
      "loss": 0.2488,
      "step": 37710
    },
    {
      "epoch": 113.27,
      "grad_norm": 20.20258331298828,
      "learning_rate": 6.224224224224224e-06,
      "loss": 0.2453,
      "step": 37720
    },
    {
      "epoch": 113.3,
      "grad_norm": 8.177850723266602,
      "learning_rate": 6.223223223223224e-06,
      "loss": 0.2553,
      "step": 37730
    },
    {
      "epoch": 113.33,
      "grad_norm": 15.764158248901367,
      "learning_rate": 6.222222222222223e-06,
      "loss": 0.2988,
      "step": 37740
    },
    {
      "epoch": 113.36,
      "grad_norm": 8.978083610534668,
      "learning_rate": 6.221221221221222e-06,
      "loss": 0.269,
      "step": 37750
    },
    {
      "epoch": 113.39,
      "grad_norm": 9.493274688720703,
      "learning_rate": 6.220220220220221e-06,
      "loss": 0.282,
      "step": 37760
    },
    {
      "epoch": 113.42,
      "grad_norm": 8.31375503540039,
      "learning_rate": 6.219219219219221e-06,
      "loss": 0.2032,
      "step": 37770
    },
    {
      "epoch": 113.45,
      "grad_norm": 6.166957855224609,
      "learning_rate": 6.218218218218218e-06,
      "loss": 0.2335,
      "step": 37780
    },
    {
      "epoch": 113.48,
      "grad_norm": 6.4353437423706055,
      "learning_rate": 6.217217217217217e-06,
      "loss": 0.2732,
      "step": 37790
    },
    {
      "epoch": 113.51,
      "grad_norm": 7.540570259094238,
      "learning_rate": 6.2162162162162164e-06,
      "loss": 0.218,
      "step": 37800
    },
    {
      "epoch": 113.54,
      "grad_norm": 23.458843231201172,
      "learning_rate": 6.215215215215216e-06,
      "loss": 0.1901,
      "step": 37810
    },
    {
      "epoch": 113.57,
      "grad_norm": 7.653669834136963,
      "learning_rate": 6.2142142142142145e-06,
      "loss": 0.2508,
      "step": 37820
    },
    {
      "epoch": 113.6,
      "grad_norm": 9.41202449798584,
      "learning_rate": 6.213213213213214e-06,
      "loss": 0.2522,
      "step": 37830
    },
    {
      "epoch": 113.63,
      "grad_norm": 6.9623308181762695,
      "learning_rate": 6.212212212212213e-06,
      "loss": 0.2323,
      "step": 37840
    },
    {
      "epoch": 113.66,
      "grad_norm": 5.778870582580566,
      "learning_rate": 6.211211211211212e-06,
      "loss": 0.233,
      "step": 37850
    },
    {
      "epoch": 113.69,
      "grad_norm": 9.826271057128906,
      "learning_rate": 6.2102102102102105e-06,
      "loss": 0.2669,
      "step": 37860
    },
    {
      "epoch": 113.72,
      "grad_norm": 11.334074020385742,
      "learning_rate": 6.209209209209209e-06,
      "loss": 0.208,
      "step": 37870
    },
    {
      "epoch": 113.75,
      "grad_norm": 6.076084613800049,
      "learning_rate": 6.2082082082082086e-06,
      "loss": 0.2405,
      "step": 37880
    },
    {
      "epoch": 113.78,
      "grad_norm": 9.796159744262695,
      "learning_rate": 6.207207207207208e-06,
      "loss": 0.2397,
      "step": 37890
    },
    {
      "epoch": 113.81,
      "grad_norm": 13.73332405090332,
      "learning_rate": 6.206206206206207e-06,
      "loss": 0.2826,
      "step": 37900
    },
    {
      "epoch": 113.84,
      "grad_norm": 10.02869987487793,
      "learning_rate": 6.205205205205206e-06,
      "loss": 0.2234,
      "step": 37910
    },
    {
      "epoch": 113.87,
      "grad_norm": 8.99661922454834,
      "learning_rate": 6.204204204204205e-06,
      "loss": 0.2582,
      "step": 37920
    },
    {
      "epoch": 113.9,
      "grad_norm": 6.223546981811523,
      "learning_rate": 6.203203203203204e-06,
      "loss": 0.2893,
      "step": 37930
    },
    {
      "epoch": 113.93,
      "grad_norm": 8.981083869934082,
      "learning_rate": 6.2022022022022035e-06,
      "loss": 0.2476,
      "step": 37940
    },
    {
      "epoch": 113.96,
      "grad_norm": 36.35367202758789,
      "learning_rate": 6.201201201201201e-06,
      "loss": 0.2406,
      "step": 37950
    },
    {
      "epoch": 113.99,
      "grad_norm": 8.822578430175781,
      "learning_rate": 6.200200200200201e-06,
      "loss": 0.2583,
      "step": 37960
    },
    {
      "epoch": 114.0,
      "eval_accuracy": 0.9252,
      "eval_loss": 0.31160488724708557,
      "eval_runtime": 30.134,
      "eval_samples_per_second": 331.852,
      "eval_steps_per_second": 1.327,
      "step": 37962
    },
    {
      "epoch": 114.02,
      "grad_norm": 11.892762184143066,
      "learning_rate": 6.199199199199199e-06,
      "loss": 0.2184,
      "step": 37970
    },
    {
      "epoch": 114.05,
      "grad_norm": 18.71360206604004,
      "learning_rate": 6.198198198198199e-06,
      "loss": 0.2716,
      "step": 37980
    },
    {
      "epoch": 114.08,
      "grad_norm": 8.133407592773438,
      "learning_rate": 6.197197197197198e-06,
      "loss": 0.2735,
      "step": 37990
    },
    {
      "epoch": 114.11,
      "grad_norm": 9.986114501953125,
      "learning_rate": 6.196196196196197e-06,
      "loss": 0.246,
      "step": 38000
    },
    {
      "epoch": 114.14,
      "grad_norm": 12.83951187133789,
      "learning_rate": 6.195195195195196e-06,
      "loss": 0.2663,
      "step": 38010
    },
    {
      "epoch": 114.17,
      "grad_norm": 8.195913314819336,
      "learning_rate": 6.194194194194195e-06,
      "loss": 0.2153,
      "step": 38020
    },
    {
      "epoch": 114.2,
      "grad_norm": 11.989972114562988,
      "learning_rate": 6.1931931931931934e-06,
      "loss": 0.2853,
      "step": 38030
    },
    {
      "epoch": 114.23,
      "grad_norm": 14.459035873413086,
      "learning_rate": 6.192192192192192e-06,
      "loss": 0.2395,
      "step": 38040
    },
    {
      "epoch": 114.26,
      "grad_norm": 11.784107208251953,
      "learning_rate": 6.1911911911911915e-06,
      "loss": 0.2414,
      "step": 38050
    },
    {
      "epoch": 114.29,
      "grad_norm": 11.579268455505371,
      "learning_rate": 6.190190190190191e-06,
      "loss": 0.2691,
      "step": 38060
    },
    {
      "epoch": 114.32,
      "grad_norm": 10.584566116333008,
      "learning_rate": 6.1891891891891895e-06,
      "loss": 0.2315,
      "step": 38070
    },
    {
      "epoch": 114.35,
      "grad_norm": 8.21994686126709,
      "learning_rate": 6.188188188188189e-06,
      "loss": 0.3157,
      "step": 38080
    },
    {
      "epoch": 114.38,
      "grad_norm": 7.388727188110352,
      "learning_rate": 6.187187187187188e-06,
      "loss": 0.2489,
      "step": 38090
    },
    {
      "epoch": 114.41,
      "grad_norm": 8.766919136047363,
      "learning_rate": 6.186186186186187e-06,
      "loss": 0.2693,
      "step": 38100
    },
    {
      "epoch": 114.44,
      "grad_norm": 13.673382759094238,
      "learning_rate": 6.1851851851851856e-06,
      "loss": 0.2377,
      "step": 38110
    },
    {
      "epoch": 114.47,
      "grad_norm": 11.159707069396973,
      "learning_rate": 6.184184184184184e-06,
      "loss": 0.2698,
      "step": 38120
    },
    {
      "epoch": 114.5,
      "grad_norm": 19.50588035583496,
      "learning_rate": 6.183183183183184e-06,
      "loss": 0.2616,
      "step": 38130
    },
    {
      "epoch": 114.53,
      "grad_norm": 14.307573318481445,
      "learning_rate": 6.182182182182182e-06,
      "loss": 0.2595,
      "step": 38140
    },
    {
      "epoch": 114.56,
      "grad_norm": 11.039061546325684,
      "learning_rate": 6.181181181181182e-06,
      "loss": 0.2596,
      "step": 38150
    },
    {
      "epoch": 114.59,
      "grad_norm": 10.978849411010742,
      "learning_rate": 6.180180180180181e-06,
      "loss": 0.2819,
      "step": 38160
    },
    {
      "epoch": 114.62,
      "grad_norm": 6.877181529998779,
      "learning_rate": 6.17917917917918e-06,
      "loss": 0.2831,
      "step": 38170
    },
    {
      "epoch": 114.65,
      "grad_norm": 7.3886494636535645,
      "learning_rate": 6.178178178178179e-06,
      "loss": 0.262,
      "step": 38180
    },
    {
      "epoch": 114.68,
      "grad_norm": 8.441142082214355,
      "learning_rate": 6.177177177177177e-06,
      "loss": 0.2695,
      "step": 38190
    },
    {
      "epoch": 114.71,
      "grad_norm": 13.224555015563965,
      "learning_rate": 6.176176176176176e-06,
      "loss": 0.267,
      "step": 38200
    },
    {
      "epoch": 114.74,
      "grad_norm": 5.132302284240723,
      "learning_rate": 6.175175175175176e-06,
      "loss": 0.2726,
      "step": 38210
    },
    {
      "epoch": 114.77,
      "grad_norm": 6.108293056488037,
      "learning_rate": 6.174174174174174e-06,
      "loss": 0.2836,
      "step": 38220
    },
    {
      "epoch": 114.8,
      "grad_norm": 4.285937786102295,
      "learning_rate": 6.173173173173174e-06,
      "loss": 0.1959,
      "step": 38230
    },
    {
      "epoch": 114.83,
      "grad_norm": 9.031401634216309,
      "learning_rate": 6.172172172172172e-06,
      "loss": 0.2411,
      "step": 38240
    },
    {
      "epoch": 114.86,
      "grad_norm": 8.77663803100586,
      "learning_rate": 6.171171171171172e-06,
      "loss": 0.2625,
      "step": 38250
    },
    {
      "epoch": 114.89,
      "grad_norm": 7.049267768859863,
      "learning_rate": 6.170170170170171e-06,
      "loss": 0.3067,
      "step": 38260
    },
    {
      "epoch": 114.92,
      "grad_norm": 6.148614883422852,
      "learning_rate": 6.16916916916917e-06,
      "loss": 0.2609,
      "step": 38270
    },
    {
      "epoch": 114.95,
      "grad_norm": 13.919294357299805,
      "learning_rate": 6.1681681681681685e-06,
      "loss": 0.2487,
      "step": 38280
    },
    {
      "epoch": 114.98,
      "grad_norm": 10.846518516540527,
      "learning_rate": 6.167167167167167e-06,
      "loss": 0.2628,
      "step": 38290
    },
    {
      "epoch": 115.0,
      "eval_accuracy": 0.9254,
      "eval_loss": 0.30684566497802734,
      "eval_runtime": 30.2912,
      "eval_samples_per_second": 330.129,
      "eval_steps_per_second": 1.321,
      "step": 38295
    },
    {
      "epoch": 115.02,
      "grad_norm": 10.886296272277832,
      "learning_rate": 6.1661661661661665e-06,
      "loss": 0.2219,
      "step": 38300
    },
    {
      "epoch": 115.05,
      "grad_norm": 8.721637725830078,
      "learning_rate": 6.165165165165166e-06,
      "loss": 0.2488,
      "step": 38310
    },
    {
      "epoch": 115.08,
      "grad_norm": 6.254786968231201,
      "learning_rate": 6.1641641641641645e-06,
      "loss": 0.265,
      "step": 38320
    },
    {
      "epoch": 115.11,
      "grad_norm": 6.652961730957031,
      "learning_rate": 6.163163163163164e-06,
      "loss": 0.2725,
      "step": 38330
    },
    {
      "epoch": 115.14,
      "grad_norm": 10.97523307800293,
      "learning_rate": 6.162162162162163e-06,
      "loss": 0.2615,
      "step": 38340
    },
    {
      "epoch": 115.17,
      "grad_norm": 6.531844139099121,
      "learning_rate": 6.161161161161162e-06,
      "loss": 0.2593,
      "step": 38350
    },
    {
      "epoch": 115.2,
      "grad_norm": 7.340628147125244,
      "learning_rate": 6.16016016016016e-06,
      "loss": 0.2571,
      "step": 38360
    },
    {
      "epoch": 115.23,
      "grad_norm": 10.43176555633545,
      "learning_rate": 6.159159159159159e-06,
      "loss": 0.1943,
      "step": 38370
    },
    {
      "epoch": 115.26,
      "grad_norm": 11.14797306060791,
      "learning_rate": 6.158158158158159e-06,
      "loss": 0.237,
      "step": 38380
    },
    {
      "epoch": 115.29,
      "grad_norm": 12.404497146606445,
      "learning_rate": 6.157157157157157e-06,
      "loss": 0.2321,
      "step": 38390
    },
    {
      "epoch": 115.32,
      "grad_norm": 13.135299682617188,
      "learning_rate": 6.156156156156157e-06,
      "loss": 0.277,
      "step": 38400
    },
    {
      "epoch": 115.35,
      "grad_norm": 11.94929027557373,
      "learning_rate": 6.155155155155156e-06,
      "loss": 0.2483,
      "step": 38410
    },
    {
      "epoch": 115.38,
      "grad_norm": 9.649328231811523,
      "learning_rate": 6.154154154154155e-06,
      "loss": 0.247,
      "step": 38420
    },
    {
      "epoch": 115.41,
      "grad_norm": 9.397720336914062,
      "learning_rate": 6.153153153153154e-06,
      "loss": 0.2755,
      "step": 38430
    },
    {
      "epoch": 115.44,
      "grad_norm": 7.3867645263671875,
      "learning_rate": 6.152152152152152e-06,
      "loss": 0.2172,
      "step": 38440
    },
    {
      "epoch": 115.47,
      "grad_norm": 5.67253303527832,
      "learning_rate": 6.151151151151151e-06,
      "loss": 0.2541,
      "step": 38450
    },
    {
      "epoch": 115.5,
      "grad_norm": 7.603451728820801,
      "learning_rate": 6.150150150150151e-06,
      "loss": 0.2923,
      "step": 38460
    },
    {
      "epoch": 115.53,
      "grad_norm": 10.766825675964355,
      "learning_rate": 6.149149149149149e-06,
      "loss": 0.2579,
      "step": 38470
    },
    {
      "epoch": 115.56,
      "grad_norm": 7.571752071380615,
      "learning_rate": 6.148148148148149e-06,
      "loss": 0.2279,
      "step": 38480
    },
    {
      "epoch": 115.59,
      "grad_norm": 11.34520149230957,
      "learning_rate": 6.147147147147147e-06,
      "loss": 0.2903,
      "step": 38490
    },
    {
      "epoch": 115.62,
      "grad_norm": 5.858510971069336,
      "learning_rate": 6.146146146146147e-06,
      "loss": 0.2756,
      "step": 38500
    },
    {
      "epoch": 115.65,
      "grad_norm": 10.454940795898438,
      "learning_rate": 6.145145145145146e-06,
      "loss": 0.2421,
      "step": 38510
    },
    {
      "epoch": 115.68,
      "grad_norm": 13.678767204284668,
      "learning_rate": 6.144144144144145e-06,
      "loss": 0.2727,
      "step": 38520
    },
    {
      "epoch": 115.71,
      "grad_norm": 9.905545234680176,
      "learning_rate": 6.1431431431431435e-06,
      "loss": 0.235,
      "step": 38530
    },
    {
      "epoch": 115.74,
      "grad_norm": 9.871652603149414,
      "learning_rate": 6.142142142142142e-06,
      "loss": 0.255,
      "step": 38540
    },
    {
      "epoch": 115.77,
      "grad_norm": 9.707547187805176,
      "learning_rate": 6.1411411411411415e-06,
      "loss": 0.2841,
      "step": 38550
    },
    {
      "epoch": 115.8,
      "grad_norm": 25.047733306884766,
      "learning_rate": 6.140140140140141e-06,
      "loss": 0.2652,
      "step": 38560
    },
    {
      "epoch": 115.83,
      "grad_norm": 7.018586158752441,
      "learning_rate": 6.1391391391391396e-06,
      "loss": 0.2896,
      "step": 38570
    },
    {
      "epoch": 115.86,
      "grad_norm": 11.931769371032715,
      "learning_rate": 6.138138138138139e-06,
      "loss": 0.2653,
      "step": 38580
    },
    {
      "epoch": 115.89,
      "grad_norm": 11.637248039245605,
      "learning_rate": 6.137137137137138e-06,
      "loss": 0.2382,
      "step": 38590
    },
    {
      "epoch": 115.92,
      "grad_norm": 15.05288028717041,
      "learning_rate": 6.136136136136137e-06,
      "loss": 0.282,
      "step": 38600
    },
    {
      "epoch": 115.95,
      "grad_norm": 6.8541340827941895,
      "learning_rate": 6.135135135135135e-06,
      "loss": 0.2569,
      "step": 38610
    },
    {
      "epoch": 115.98,
      "grad_norm": 10.473970413208008,
      "learning_rate": 6.134134134134134e-06,
      "loss": 0.2806,
      "step": 38620
    },
    {
      "epoch": 116.0,
      "eval_accuracy": 0.9257,
      "eval_loss": 0.31143227219581604,
      "eval_runtime": 30.466,
      "eval_samples_per_second": 328.235,
      "eval_steps_per_second": 1.313,
      "step": 38628
    },
    {
      "epoch": 116.01,
      "grad_norm": 8.30316162109375,
      "learning_rate": 6.133133133133134e-06,
      "loss": 0.2695,
      "step": 38630
    },
    {
      "epoch": 116.04,
      "grad_norm": 12.206727027893066,
      "learning_rate": 6.132132132132132e-06,
      "loss": 0.2583,
      "step": 38640
    },
    {
      "epoch": 116.07,
      "grad_norm": 16.808364868164062,
      "learning_rate": 6.131131131131132e-06,
      "loss": 0.2218,
      "step": 38650
    },
    {
      "epoch": 116.1,
      "grad_norm": 15.95702075958252,
      "learning_rate": 6.130130130130131e-06,
      "loss": 0.2467,
      "step": 38660
    },
    {
      "epoch": 116.13,
      "grad_norm": 23.2890625,
      "learning_rate": 6.12912912912913e-06,
      "loss": 0.2661,
      "step": 38670
    },
    {
      "epoch": 116.16,
      "grad_norm": 9.221982955932617,
      "learning_rate": 6.128128128128129e-06,
      "loss": 0.2791,
      "step": 38680
    },
    {
      "epoch": 116.19,
      "grad_norm": 15.007526397705078,
      "learning_rate": 6.127127127127127e-06,
      "loss": 0.2934,
      "step": 38690
    },
    {
      "epoch": 116.22,
      "grad_norm": 19.490705490112305,
      "learning_rate": 6.126126126126126e-06,
      "loss": 0.2563,
      "step": 38700
    },
    {
      "epoch": 116.25,
      "grad_norm": 5.2466864585876465,
      "learning_rate": 6.125125125125125e-06,
      "loss": 0.243,
      "step": 38710
    },
    {
      "epoch": 116.28,
      "grad_norm": 22.895421981811523,
      "learning_rate": 6.124124124124124e-06,
      "loss": 0.2743,
      "step": 38720
    },
    {
      "epoch": 116.31,
      "grad_norm": 12.177725791931152,
      "learning_rate": 6.123123123123124e-06,
      "loss": 0.2716,
      "step": 38730
    },
    {
      "epoch": 116.34,
      "grad_norm": 7.783472537994385,
      "learning_rate": 6.1221221221221225e-06,
      "loss": 0.3009,
      "step": 38740
    },
    {
      "epoch": 116.37,
      "grad_norm": 10.593048095703125,
      "learning_rate": 6.121121121121122e-06,
      "loss": 0.2653,
      "step": 38750
    },
    {
      "epoch": 116.4,
      "grad_norm": 5.460694789886475,
      "learning_rate": 6.120120120120121e-06,
      "loss": 0.2289,
      "step": 38760
    },
    {
      "epoch": 116.43,
      "grad_norm": 11.950486183166504,
      "learning_rate": 6.11911911911912e-06,
      "loss": 0.2438,
      "step": 38770
    },
    {
      "epoch": 116.46,
      "grad_norm": 17.39990997314453,
      "learning_rate": 6.1181181181181185e-06,
      "loss": 0.2452,
      "step": 38780
    },
    {
      "epoch": 116.49,
      "grad_norm": 9.127978324890137,
      "learning_rate": 6.117117117117117e-06,
      "loss": 0.2909,
      "step": 38790
    },
    {
      "epoch": 116.52,
      "grad_norm": 10.356189727783203,
      "learning_rate": 6.1161161161161166e-06,
      "loss": 0.2245,
      "step": 38800
    },
    {
      "epoch": 116.55,
      "grad_norm": 9.58715534210205,
      "learning_rate": 6.115115115115115e-06,
      "loss": 0.2733,
      "step": 38810
    },
    {
      "epoch": 116.58,
      "grad_norm": 11.63288688659668,
      "learning_rate": 6.114114114114115e-06,
      "loss": 0.2958,
      "step": 38820
    },
    {
      "epoch": 116.61,
      "grad_norm": 11.007725715637207,
      "learning_rate": 6.113113113113114e-06,
      "loss": 0.2614,
      "step": 38830
    },
    {
      "epoch": 116.64,
      "grad_norm": 23.13662338256836,
      "learning_rate": 6.112112112112113e-06,
      "loss": 0.2915,
      "step": 38840
    },
    {
      "epoch": 116.67,
      "grad_norm": 10.476344108581543,
      "learning_rate": 6.111111111111112e-06,
      "loss": 0.2473,
      "step": 38850
    },
    {
      "epoch": 116.7,
      "grad_norm": 9.86097526550293,
      "learning_rate": 6.11011011011011e-06,
      "loss": 0.2724,
      "step": 38860
    },
    {
      "epoch": 116.73,
      "grad_norm": 6.1971611976623535,
      "learning_rate": 6.109109109109109e-06,
      "loss": 0.206,
      "step": 38870
    },
    {
      "epoch": 116.76,
      "grad_norm": 10.36914348602295,
      "learning_rate": 6.108108108108109e-06,
      "loss": 0.2225,
      "step": 38880
    },
    {
      "epoch": 116.79,
      "grad_norm": 9.427523612976074,
      "learning_rate": 6.107107107107107e-06,
      "loss": 0.2692,
      "step": 38890
    },
    {
      "epoch": 116.82,
      "grad_norm": 7.271234035491943,
      "learning_rate": 6.106106106106107e-06,
      "loss": 0.2505,
      "step": 38900
    },
    {
      "epoch": 116.85,
      "grad_norm": 9.281888008117676,
      "learning_rate": 6.105105105105106e-06,
      "loss": 0.2552,
      "step": 38910
    },
    {
      "epoch": 116.88,
      "grad_norm": 8.888927459716797,
      "learning_rate": 6.104104104104105e-06,
      "loss": 0.2361,
      "step": 38920
    },
    {
      "epoch": 116.91,
      "grad_norm": 20.15721321105957,
      "learning_rate": 6.103103103103104e-06,
      "loss": 0.3003,
      "step": 38930
    },
    {
      "epoch": 116.94,
      "grad_norm": 7.050634860992432,
      "learning_rate": 6.102102102102102e-06,
      "loss": 0.262,
      "step": 38940
    },
    {
      "epoch": 116.97,
      "grad_norm": 7.86729621887207,
      "learning_rate": 6.101101101101101e-06,
      "loss": 0.2643,
      "step": 38950
    },
    {
      "epoch": 117.0,
      "grad_norm": 10.700554847717285,
      "learning_rate": 6.1001001001001e-06,
      "loss": 0.2441,
      "step": 38960
    },
    {
      "epoch": 117.0,
      "eval_accuracy": 0.9251,
      "eval_loss": 0.307223379611969,
      "eval_runtime": 29.7689,
      "eval_samples_per_second": 335.921,
      "eval_steps_per_second": 1.344,
      "step": 38961
    },
    {
      "epoch": 117.03,
      "grad_norm": 6.565662384033203,
      "learning_rate": 6.0990990990990995e-06,
      "loss": 0.2177,
      "step": 38970
    },
    {
      "epoch": 117.06,
      "grad_norm": 14.648101806640625,
      "learning_rate": 6.098098098098099e-06,
      "loss": 0.2454,
      "step": 38980
    },
    {
      "epoch": 117.09,
      "grad_norm": 4.978806972503662,
      "learning_rate": 6.0970970970970975e-06,
      "loss": 0.2033,
      "step": 38990
    },
    {
      "epoch": 117.12,
      "grad_norm": 9.626849174499512,
      "learning_rate": 6.096096096096097e-06,
      "loss": 0.289,
      "step": 39000
    },
    {
      "epoch": 117.15,
      "grad_norm": 7.168361186981201,
      "learning_rate": 6.095095095095096e-06,
      "loss": 0.227,
      "step": 39010
    },
    {
      "epoch": 117.18,
      "grad_norm": 5.5972466468811035,
      "learning_rate": 6.094094094094095e-06,
      "loss": 0.258,
      "step": 39020
    },
    {
      "epoch": 117.21,
      "grad_norm": 13.208948135375977,
      "learning_rate": 6.0930930930930936e-06,
      "loss": 0.2645,
      "step": 39030
    },
    {
      "epoch": 117.24,
      "grad_norm": 5.960218906402588,
      "learning_rate": 6.092092092092092e-06,
      "loss": 0.2238,
      "step": 39040
    },
    {
      "epoch": 117.27,
      "grad_norm": 9.033866882324219,
      "learning_rate": 6.091091091091092e-06,
      "loss": 0.2662,
      "step": 39050
    },
    {
      "epoch": 117.3,
      "grad_norm": 8.043190002441406,
      "learning_rate": 6.09009009009009e-06,
      "loss": 0.2371,
      "step": 39060
    },
    {
      "epoch": 117.33,
      "grad_norm": 10.197691917419434,
      "learning_rate": 6.08908908908909e-06,
      "loss": 0.3083,
      "step": 39070
    },
    {
      "epoch": 117.36,
      "grad_norm": 9.9664306640625,
      "learning_rate": 6.088088088088089e-06,
      "loss": 0.2743,
      "step": 39080
    },
    {
      "epoch": 117.39,
      "grad_norm": 10.844202041625977,
      "learning_rate": 6.087087087087088e-06,
      "loss": 0.2648,
      "step": 39090
    },
    {
      "epoch": 117.42,
      "grad_norm": 10.703262329101562,
      "learning_rate": 6.086086086086087e-06,
      "loss": 0.2338,
      "step": 39100
    },
    {
      "epoch": 117.45,
      "grad_norm": 8.967732429504395,
      "learning_rate": 6.085085085085085e-06,
      "loss": 0.2779,
      "step": 39110
    },
    {
      "epoch": 117.48,
      "grad_norm": 14.183002471923828,
      "learning_rate": 6.084084084084084e-06,
      "loss": 0.2744,
      "step": 39120
    },
    {
      "epoch": 117.51,
      "grad_norm": 8.994462013244629,
      "learning_rate": 6.083083083083084e-06,
      "loss": 0.2577,
      "step": 39130
    },
    {
      "epoch": 117.54,
      "grad_norm": 10.487449645996094,
      "learning_rate": 6.082082082082082e-06,
      "loss": 0.2408,
      "step": 39140
    },
    {
      "epoch": 117.57,
      "grad_norm": 12.621175765991211,
      "learning_rate": 6.081081081081082e-06,
      "loss": 0.2955,
      "step": 39150
    },
    {
      "epoch": 117.6,
      "grad_norm": 12.260003089904785,
      "learning_rate": 6.08008008008008e-06,
      "loss": 0.2974,
      "step": 39160
    },
    {
      "epoch": 117.63,
      "grad_norm": 8.300865173339844,
      "learning_rate": 6.07907907907908e-06,
      "loss": 0.2763,
      "step": 39170
    },
    {
      "epoch": 117.66,
      "grad_norm": 9.601490020751953,
      "learning_rate": 6.078078078078079e-06,
      "loss": 0.2352,
      "step": 39180
    },
    {
      "epoch": 117.69,
      "grad_norm": 11.082831382751465,
      "learning_rate": 6.077077077077077e-06,
      "loss": 0.2416,
      "step": 39190
    },
    {
      "epoch": 117.72,
      "grad_norm": 9.764662742614746,
      "learning_rate": 6.0760760760760765e-06,
      "loss": 0.2754,
      "step": 39200
    },
    {
      "epoch": 117.75,
      "grad_norm": 6.564511299133301,
      "learning_rate": 6.075075075075075e-06,
      "loss": 0.2788,
      "step": 39210
    },
    {
      "epoch": 117.78,
      "grad_norm": 9.920165061950684,
      "learning_rate": 6.0740740740740745e-06,
      "loss": 0.3017,
      "step": 39220
    },
    {
      "epoch": 117.81,
      "grad_norm": 7.549356937408447,
      "learning_rate": 6.073073073073074e-06,
      "loss": 0.2798,
      "step": 39230
    },
    {
      "epoch": 117.84,
      "grad_norm": 7.985033988952637,
      "learning_rate": 6.0720720720720725e-06,
      "loss": 0.2662,
      "step": 39240
    },
    {
      "epoch": 117.87,
      "grad_norm": 12.356758117675781,
      "learning_rate": 6.071071071071072e-06,
      "loss": 0.2333,
      "step": 39250
    },
    {
      "epoch": 117.9,
      "grad_norm": 7.240966796875,
      "learning_rate": 6.0700700700700706e-06,
      "loss": 0.2757,
      "step": 39260
    },
    {
      "epoch": 117.93,
      "grad_norm": 8.923054695129395,
      "learning_rate": 6.06906906906907e-06,
      "loss": 0.2784,
      "step": 39270
    },
    {
      "epoch": 117.96,
      "grad_norm": 9.298643112182617,
      "learning_rate": 6.068068068068068e-06,
      "loss": 0.2407,
      "step": 39280
    },
    {
      "epoch": 117.99,
      "grad_norm": 6.764179229736328,
      "learning_rate": 6.067067067067067e-06,
      "loss": 0.2204,
      "step": 39290
    },
    {
      "epoch": 118.0,
      "eval_accuracy": 0.9258,
      "eval_loss": 0.31253013014793396,
      "eval_runtime": 30.0138,
      "eval_samples_per_second": 333.18,
      "eval_steps_per_second": 1.333,
      "step": 39294
    },
    {
      "epoch": 118.02,
      "grad_norm": 9.066445350646973,
      "learning_rate": 6.066066066066067e-06,
      "loss": 0.2656,
      "step": 39300
    },
    {
      "epoch": 118.05,
      "grad_norm": 8.549710273742676,
      "learning_rate": 6.065065065065065e-06,
      "loss": 0.2599,
      "step": 39310
    },
    {
      "epoch": 118.08,
      "grad_norm": 6.807999134063721,
      "learning_rate": 6.064064064064065e-06,
      "loss": 0.2525,
      "step": 39320
    },
    {
      "epoch": 118.11,
      "grad_norm": 12.126139640808105,
      "learning_rate": 6.063063063063064e-06,
      "loss": 0.2441,
      "step": 39330
    },
    {
      "epoch": 118.14,
      "grad_norm": 9.12575912475586,
      "learning_rate": 6.062062062062063e-06,
      "loss": 0.2342,
      "step": 39340
    },
    {
      "epoch": 118.17,
      "grad_norm": 19.12923812866211,
      "learning_rate": 6.061061061061062e-06,
      "loss": 0.2296,
      "step": 39350
    },
    {
      "epoch": 118.2,
      "grad_norm": 13.029208183288574,
      "learning_rate": 6.06006006006006e-06,
      "loss": 0.2955,
      "step": 39360
    },
    {
      "epoch": 118.23,
      "grad_norm": 13.006131172180176,
      "learning_rate": 6.059059059059059e-06,
      "loss": 0.2541,
      "step": 39370
    },
    {
      "epoch": 118.26,
      "grad_norm": 3.190408706665039,
      "learning_rate": 6.058058058058058e-06,
      "loss": 0.2268,
      "step": 39380
    },
    {
      "epoch": 118.29,
      "grad_norm": 12.788191795349121,
      "learning_rate": 6.057057057057057e-06,
      "loss": 0.2708,
      "step": 39390
    },
    {
      "epoch": 118.32,
      "grad_norm": 7.785965442657471,
      "learning_rate": 6.056056056056057e-06,
      "loss": 0.2531,
      "step": 39400
    },
    {
      "epoch": 118.35,
      "grad_norm": 11.416372299194336,
      "learning_rate": 6.055055055055055e-06,
      "loss": 0.2489,
      "step": 39410
    },
    {
      "epoch": 118.38,
      "grad_norm": 10.879304885864258,
      "learning_rate": 6.054054054054055e-06,
      "loss": 0.2422,
      "step": 39420
    },
    {
      "epoch": 118.41,
      "grad_norm": 8.678972244262695,
      "learning_rate": 6.053053053053054e-06,
      "loss": 0.2599,
      "step": 39430
    },
    {
      "epoch": 118.44,
      "grad_norm": 6.798366069793701,
      "learning_rate": 6.052052052052052e-06,
      "loss": 0.2422,
      "step": 39440
    },
    {
      "epoch": 118.47,
      "grad_norm": 12.092879295349121,
      "learning_rate": 6.0510510510510515e-06,
      "loss": 0.2272,
      "step": 39450
    },
    {
      "epoch": 118.5,
      "grad_norm": 7.562942028045654,
      "learning_rate": 6.05005005005005e-06,
      "loss": 0.2757,
      "step": 39460
    },
    {
      "epoch": 118.53,
      "grad_norm": 15.33298397064209,
      "learning_rate": 6.0490490490490495e-06,
      "loss": 0.2815,
      "step": 39470
    },
    {
      "epoch": 118.56,
      "grad_norm": 7.9630446434021,
      "learning_rate": 6.048048048048049e-06,
      "loss": 0.2727,
      "step": 39480
    },
    {
      "epoch": 118.59,
      "grad_norm": 10.132438659667969,
      "learning_rate": 6.0470470470470476e-06,
      "loss": 0.2501,
      "step": 39490
    },
    {
      "epoch": 118.62,
      "grad_norm": 6.624090671539307,
      "learning_rate": 6.046046046046047e-06,
      "loss": 0.2757,
      "step": 39500
    },
    {
      "epoch": 118.65,
      "grad_norm": 21.6922550201416,
      "learning_rate": 6.045045045045046e-06,
      "loss": 0.2535,
      "step": 39510
    },
    {
      "epoch": 118.68,
      "grad_norm": 5.3113203048706055,
      "learning_rate": 6.044044044044044e-06,
      "loss": 0.197,
      "step": 39520
    },
    {
      "epoch": 118.71,
      "grad_norm": 9.037190437316895,
      "learning_rate": 6.043043043043043e-06,
      "loss": 0.2522,
      "step": 39530
    },
    {
      "epoch": 118.74,
      "grad_norm": 12.999390602111816,
      "learning_rate": 6.042042042042042e-06,
      "loss": 0.2723,
      "step": 39540
    },
    {
      "epoch": 118.77,
      "grad_norm": 6.362791061401367,
      "learning_rate": 6.041041041041042e-06,
      "loss": 0.2495,
      "step": 39550
    },
    {
      "epoch": 118.8,
      "grad_norm": 9.592296600341797,
      "learning_rate": 6.04004004004004e-06,
      "loss": 0.252,
      "step": 39560
    },
    {
      "epoch": 118.83,
      "grad_norm": 8.545079231262207,
      "learning_rate": 6.03903903903904e-06,
      "loss": 0.2469,
      "step": 39570
    },
    {
      "epoch": 118.86,
      "grad_norm": 7.989108562469482,
      "learning_rate": 6.038038038038039e-06,
      "loss": 0.2845,
      "step": 39580
    },
    {
      "epoch": 118.89,
      "grad_norm": 7.143494606018066,
      "learning_rate": 6.037037037037038e-06,
      "loss": 0.2718,
      "step": 39590
    },
    {
      "epoch": 118.92,
      "grad_norm": 4.50592041015625,
      "learning_rate": 6.036036036036037e-06,
      "loss": 0.2302,
      "step": 39600
    },
    {
      "epoch": 118.95,
      "grad_norm": 6.568984031677246,
      "learning_rate": 6.035035035035035e-06,
      "loss": 0.2337,
      "step": 39610
    },
    {
      "epoch": 118.98,
      "grad_norm": 12.400115966796875,
      "learning_rate": 6.034034034034034e-06,
      "loss": 0.2819,
      "step": 39620
    },
    {
      "epoch": 119.0,
      "eval_accuracy": 0.9237,
      "eval_loss": 0.31779733300209045,
      "eval_runtime": 30.4534,
      "eval_samples_per_second": 328.37,
      "eval_steps_per_second": 1.313,
      "step": 39627
    },
    {
      "epoch": 119.01,
      "grad_norm": 5.731043815612793,
      "learning_rate": 6.033033033033033e-06,
      "loss": 0.2378,
      "step": 39630
    },
    {
      "epoch": 119.04,
      "grad_norm": 12.324281692504883,
      "learning_rate": 6.032032032032032e-06,
      "loss": 0.2666,
      "step": 39640
    },
    {
      "epoch": 119.07,
      "grad_norm": 6.980433940887451,
      "learning_rate": 6.031031031031032e-06,
      "loss": 0.2466,
      "step": 39650
    },
    {
      "epoch": 119.1,
      "grad_norm": 7.767157077789307,
      "learning_rate": 6.0300300300300304e-06,
      "loss": 0.2384,
      "step": 39660
    },
    {
      "epoch": 119.13,
      "grad_norm": 9.548602104187012,
      "learning_rate": 6.02902902902903e-06,
      "loss": 0.2424,
      "step": 39670
    },
    {
      "epoch": 119.16,
      "grad_norm": 22.531282424926758,
      "learning_rate": 6.028028028028029e-06,
      "loss": 0.3198,
      "step": 39680
    },
    {
      "epoch": 119.19,
      "grad_norm": 8.061245918273926,
      "learning_rate": 6.027027027027027e-06,
      "loss": 0.245,
      "step": 39690
    },
    {
      "epoch": 119.22,
      "grad_norm": 12.93807601928711,
      "learning_rate": 6.0260260260260265e-06,
      "loss": 0.2861,
      "step": 39700
    },
    {
      "epoch": 119.25,
      "grad_norm": 12.445990562438965,
      "learning_rate": 6.025025025025025e-06,
      "loss": 0.2826,
      "step": 39710
    },
    {
      "epoch": 119.28,
      "grad_norm": 3.987046957015991,
      "learning_rate": 6.0240240240240246e-06,
      "loss": 0.2441,
      "step": 39720
    },
    {
      "epoch": 119.31,
      "grad_norm": 6.052858352661133,
      "learning_rate": 6.023023023023023e-06,
      "loss": 0.238,
      "step": 39730
    },
    {
      "epoch": 119.34,
      "grad_norm": 7.917327404022217,
      "learning_rate": 6.022022022022023e-06,
      "loss": 0.2778,
      "step": 39740
    },
    {
      "epoch": 119.37,
      "grad_norm": 13.133591651916504,
      "learning_rate": 6.021021021021022e-06,
      "loss": 0.2356,
      "step": 39750
    },
    {
      "epoch": 119.4,
      "grad_norm": 14.234613418579102,
      "learning_rate": 6.020020020020021e-06,
      "loss": 0.2808,
      "step": 39760
    },
    {
      "epoch": 119.43,
      "grad_norm": 13.570443153381348,
      "learning_rate": 6.019019019019019e-06,
      "loss": 0.2588,
      "step": 39770
    },
    {
      "epoch": 119.46,
      "grad_norm": 6.468785762786865,
      "learning_rate": 6.018018018018018e-06,
      "loss": 0.2056,
      "step": 39780
    },
    {
      "epoch": 119.49,
      "grad_norm": 6.313859462738037,
      "learning_rate": 6.017017017017017e-06,
      "loss": 0.2162,
      "step": 39790
    },
    {
      "epoch": 119.52,
      "grad_norm": 8.881738662719727,
      "learning_rate": 6.016016016016017e-06,
      "loss": 0.3329,
      "step": 39800
    },
    {
      "epoch": 119.55,
      "grad_norm": 6.6433820724487305,
      "learning_rate": 6.015015015015015e-06,
      "loss": 0.2485,
      "step": 39810
    },
    {
      "epoch": 119.58,
      "grad_norm": 11.015219688415527,
      "learning_rate": 6.014014014014015e-06,
      "loss": 0.2462,
      "step": 39820
    },
    {
      "epoch": 119.61,
      "grad_norm": 9.088865280151367,
      "learning_rate": 6.013013013013014e-06,
      "loss": 0.2332,
      "step": 39830
    },
    {
      "epoch": 119.64,
      "grad_norm": 18.086044311523438,
      "learning_rate": 6.012012012012013e-06,
      "loss": 0.298,
      "step": 39840
    },
    {
      "epoch": 119.67,
      "grad_norm": 9.720559120178223,
      "learning_rate": 6.011011011011012e-06,
      "loss": 0.2521,
      "step": 39850
    },
    {
      "epoch": 119.7,
      "grad_norm": 5.5593791007995605,
      "learning_rate": 6.01001001001001e-06,
      "loss": 0.2331,
      "step": 39860
    },
    {
      "epoch": 119.73,
      "grad_norm": 13.440938949584961,
      "learning_rate": 6.009009009009009e-06,
      "loss": 0.2752,
      "step": 39870
    },
    {
      "epoch": 119.76,
      "grad_norm": 14.153654098510742,
      "learning_rate": 6.008008008008008e-06,
      "loss": 0.2733,
      "step": 39880
    },
    {
      "epoch": 119.79,
      "grad_norm": 16.146432876586914,
      "learning_rate": 6.0070070070070074e-06,
      "loss": 0.2462,
      "step": 39890
    },
    {
      "epoch": 119.82,
      "grad_norm": 12.803178787231445,
      "learning_rate": 6.006006006006007e-06,
      "loss": 0.2716,
      "step": 39900
    },
    {
      "epoch": 119.85,
      "grad_norm": 14.097338676452637,
      "learning_rate": 6.0050050050050055e-06,
      "loss": 0.2475,
      "step": 39910
    },
    {
      "epoch": 119.88,
      "grad_norm": 11.741219520568848,
      "learning_rate": 6.004004004004005e-06,
      "loss": 0.2537,
      "step": 39920
    },
    {
      "epoch": 119.91,
      "grad_norm": 7.352508068084717,
      "learning_rate": 6.003003003003004e-06,
      "loss": 0.2536,
      "step": 39930
    },
    {
      "epoch": 119.94,
      "grad_norm": 15.242510795593262,
      "learning_rate": 6.002002002002002e-06,
      "loss": 0.2612,
      "step": 39940
    },
    {
      "epoch": 119.97,
      "grad_norm": 10.075516700744629,
      "learning_rate": 6.0010010010010016e-06,
      "loss": 0.241,
      "step": 39950
    },
    {
      "epoch": 120.0,
      "grad_norm": 12.009880065917969,
      "learning_rate": 6e-06,
      "loss": 0.2466,
      "step": 39960
    },
    {
      "epoch": 120.0,
      "eval_accuracy": 0.9248,
      "eval_loss": 0.3189446032047272,
      "eval_runtime": 29.9861,
      "eval_samples_per_second": 333.488,
      "eval_steps_per_second": 1.334,
      "step": 39960
    },
    {
      "epoch": 120.03,
      "grad_norm": 8.160889625549316,
      "learning_rate": 5.998998998999e-06,
      "loss": 0.2573,
      "step": 39970
    },
    {
      "epoch": 120.06,
      "grad_norm": 8.350765228271484,
      "learning_rate": 5.997997997997998e-06,
      "loss": 0.2779,
      "step": 39980
    },
    {
      "epoch": 120.09,
      "grad_norm": 7.69767951965332,
      "learning_rate": 5.996996996996998e-06,
      "loss": 0.2439,
      "step": 39990
    },
    {
      "epoch": 120.12,
      "grad_norm": 14.085836410522461,
      "learning_rate": 5.995995995995997e-06,
      "loss": 0.2372,
      "step": 40000
    },
    {
      "epoch": 120.15,
      "grad_norm": 9.45754337310791,
      "learning_rate": 5.994994994994996e-06,
      "loss": 0.2583,
      "step": 40010
    },
    {
      "epoch": 120.18,
      "grad_norm": 10.212403297424316,
      "learning_rate": 5.993993993993994e-06,
      "loss": 0.2578,
      "step": 40020
    },
    {
      "epoch": 120.21,
      "grad_norm": 12.724573135375977,
      "learning_rate": 5.992992992992993e-06,
      "loss": 0.2295,
      "step": 40030
    },
    {
      "epoch": 120.24,
      "grad_norm": 9.895998001098633,
      "learning_rate": 5.991991991991992e-06,
      "loss": 0.2563,
      "step": 40040
    },
    {
      "epoch": 120.27,
      "grad_norm": 10.176663398742676,
      "learning_rate": 5.990990990990992e-06,
      "loss": 0.2434,
      "step": 40050
    },
    {
      "epoch": 120.3,
      "grad_norm": 8.139554023742676,
      "learning_rate": 5.98998998998999e-06,
      "loss": 0.277,
      "step": 40060
    },
    {
      "epoch": 120.33,
      "grad_norm": 9.283276557922363,
      "learning_rate": 5.98898898898899e-06,
      "loss": 0.2484,
      "step": 40070
    },
    {
      "epoch": 120.36,
      "grad_norm": 55.262939453125,
      "learning_rate": 5.987987987987988e-06,
      "loss": 0.2677,
      "step": 40080
    },
    {
      "epoch": 120.39,
      "grad_norm": 9.766742706298828,
      "learning_rate": 5.986986986986988e-06,
      "loss": 0.2413,
      "step": 40090
    },
    {
      "epoch": 120.42,
      "grad_norm": 10.060392379760742,
      "learning_rate": 5.985985985985987e-06,
      "loss": 0.1816,
      "step": 40100
    },
    {
      "epoch": 120.45,
      "grad_norm": 8.930059432983398,
      "learning_rate": 5.984984984984985e-06,
      "loss": 0.2277,
      "step": 40110
    },
    {
      "epoch": 120.48,
      "grad_norm": 7.988299369812012,
      "learning_rate": 5.9839839839839844e-06,
      "loss": 0.2851,
      "step": 40120
    },
    {
      "epoch": 120.51,
      "grad_norm": 7.063798427581787,
      "learning_rate": 5.982982982982983e-06,
      "loss": 0.1896,
      "step": 40130
    },
    {
      "epoch": 120.54,
      "grad_norm": 7.3409013748168945,
      "learning_rate": 5.9819819819819825e-06,
      "loss": 0.3091,
      "step": 40140
    },
    {
      "epoch": 120.57,
      "grad_norm": 7.962474346160889,
      "learning_rate": 5.980980980980982e-06,
      "loss": 0.2483,
      "step": 40150
    },
    {
      "epoch": 120.6,
      "grad_norm": 13.670761108398438,
      "learning_rate": 5.9799799799799805e-06,
      "loss": 0.2462,
      "step": 40160
    },
    {
      "epoch": 120.63,
      "grad_norm": 17.727678298950195,
      "learning_rate": 5.97897897897898e-06,
      "loss": 0.2876,
      "step": 40170
    },
    {
      "epoch": 120.66,
      "grad_norm": 7.870440483093262,
      "learning_rate": 5.9779779779779786e-06,
      "loss": 0.2222,
      "step": 40180
    },
    {
      "epoch": 120.69,
      "grad_norm": 13.822956085205078,
      "learning_rate": 5.976976976976977e-06,
      "loss": 0.2465,
      "step": 40190
    },
    {
      "epoch": 120.72,
      "grad_norm": 9.53570556640625,
      "learning_rate": 5.975975975975976e-06,
      "loss": 0.2565,
      "step": 40200
    },
    {
      "epoch": 120.75,
      "grad_norm": 8.40291976928711,
      "learning_rate": 5.974974974974975e-06,
      "loss": 0.254,
      "step": 40210
    },
    {
      "epoch": 120.78,
      "grad_norm": 10.406725883483887,
      "learning_rate": 5.973973973973975e-06,
      "loss": 0.2229,
      "step": 40220
    },
    {
      "epoch": 120.81,
      "grad_norm": 9.038362503051758,
      "learning_rate": 5.972972972972973e-06,
      "loss": 0.2815,
      "step": 40230
    },
    {
      "epoch": 120.84,
      "grad_norm": 7.042052745819092,
      "learning_rate": 5.971971971971973e-06,
      "loss": 0.2515,
      "step": 40240
    },
    {
      "epoch": 120.87,
      "grad_norm": 8.427806854248047,
      "learning_rate": 5.970970970970972e-06,
      "loss": 0.2426,
      "step": 40250
    },
    {
      "epoch": 120.9,
      "grad_norm": 14.390480041503906,
      "learning_rate": 5.969969969969971e-06,
      "loss": 0.2725,
      "step": 40260
    },
    {
      "epoch": 120.93,
      "grad_norm": 10.28323745727539,
      "learning_rate": 5.968968968968969e-06,
      "loss": 0.251,
      "step": 40270
    },
    {
      "epoch": 120.96,
      "grad_norm": 7.910725116729736,
      "learning_rate": 5.967967967967968e-06,
      "loss": 0.2694,
      "step": 40280
    },
    {
      "epoch": 120.99,
      "grad_norm": 8.420743942260742,
      "learning_rate": 5.966966966966967e-06,
      "loss": 0.2284,
      "step": 40290
    },
    {
      "epoch": 121.0,
      "eval_accuracy": 0.9267,
      "eval_loss": 0.31067851185798645,
      "eval_runtime": 29.8228,
      "eval_samples_per_second": 335.314,
      "eval_steps_per_second": 1.341,
      "step": 40293
    },
    {
      "epoch": 121.02,
      "grad_norm": 17.652820587158203,
      "learning_rate": 5.965965965965966e-06,
      "loss": 0.2107,
      "step": 40300
    },
    {
      "epoch": 121.05,
      "grad_norm": 7.924627304077148,
      "learning_rate": 5.964964964964965e-06,
      "loss": 0.2225,
      "step": 40310
    },
    {
      "epoch": 121.08,
      "grad_norm": 9.036622047424316,
      "learning_rate": 5.963963963963965e-06,
      "loss": 0.2677,
      "step": 40320
    },
    {
      "epoch": 121.11,
      "grad_norm": 8.5034818649292,
      "learning_rate": 5.962962962962963e-06,
      "loss": 0.263,
      "step": 40330
    },
    {
      "epoch": 121.14,
      "grad_norm": 10.92290210723877,
      "learning_rate": 5.961961961961963e-06,
      "loss": 0.2023,
      "step": 40340
    },
    {
      "epoch": 121.17,
      "grad_norm": 11.789287567138672,
      "learning_rate": 5.960960960960962e-06,
      "loss": 0.2561,
      "step": 40350
    },
    {
      "epoch": 121.2,
      "grad_norm": 7.607210159301758,
      "learning_rate": 5.95995995995996e-06,
      "loss": 0.2297,
      "step": 40360
    },
    {
      "epoch": 121.23,
      "grad_norm": 8.325346946716309,
      "learning_rate": 5.9589589589589595e-06,
      "loss": 0.2513,
      "step": 40370
    },
    {
      "epoch": 121.26,
      "grad_norm": 20.17912483215332,
      "learning_rate": 5.957957957957958e-06,
      "loss": 0.2855,
      "step": 40380
    },
    {
      "epoch": 121.29,
      "grad_norm": 12.189083099365234,
      "learning_rate": 5.9569569569569575e-06,
      "loss": 0.2362,
      "step": 40390
    },
    {
      "epoch": 121.32,
      "grad_norm": 7.388943195343018,
      "learning_rate": 5.955955955955957e-06,
      "loss": 0.2394,
      "step": 40400
    },
    {
      "epoch": 121.35,
      "grad_norm": 11.162872314453125,
      "learning_rate": 5.9549549549549556e-06,
      "loss": 0.2963,
      "step": 40410
    },
    {
      "epoch": 121.38,
      "grad_norm": 8.132135391235352,
      "learning_rate": 5.953953953953955e-06,
      "loss": 0.249,
      "step": 40420
    },
    {
      "epoch": 121.41,
      "grad_norm": 5.640919208526611,
      "learning_rate": 5.952952952952954e-06,
      "loss": 0.2741,
      "step": 40430
    },
    {
      "epoch": 121.44,
      "grad_norm": 6.086001873016357,
      "learning_rate": 5.951951951951952e-06,
      "loss": 0.2585,
      "step": 40440
    },
    {
      "epoch": 121.47,
      "grad_norm": 8.988040924072266,
      "learning_rate": 5.950950950950951e-06,
      "loss": 0.2518,
      "step": 40450
    },
    {
      "epoch": 121.5,
      "grad_norm": 10.865507125854492,
      "learning_rate": 5.94994994994995e-06,
      "loss": 0.2751,
      "step": 40460
    },
    {
      "epoch": 121.53,
      "grad_norm": 15.225364685058594,
      "learning_rate": 5.94894894894895e-06,
      "loss": 0.2635,
      "step": 40470
    },
    {
      "epoch": 121.56,
      "grad_norm": 9.430177688598633,
      "learning_rate": 5.947947947947948e-06,
      "loss": 0.2529,
      "step": 40480
    },
    {
      "epoch": 121.59,
      "grad_norm": 12.637765884399414,
      "learning_rate": 5.946946946946948e-06,
      "loss": 0.2905,
      "step": 40490
    },
    {
      "epoch": 121.62,
      "grad_norm": 8.610780715942383,
      "learning_rate": 5.945945945945947e-06,
      "loss": 0.2494,
      "step": 40500
    },
    {
      "epoch": 121.65,
      "grad_norm": 6.879619598388672,
      "learning_rate": 5.944944944944946e-06,
      "loss": 0.3092,
      "step": 40510
    },
    {
      "epoch": 121.68,
      "grad_norm": 9.414328575134277,
      "learning_rate": 5.943943943943944e-06,
      "loss": 0.2435,
      "step": 40520
    },
    {
      "epoch": 121.71,
      "grad_norm": 16.73748207092285,
      "learning_rate": 5.942942942942943e-06,
      "loss": 0.3011,
      "step": 40530
    },
    {
      "epoch": 121.74,
      "grad_norm": 18.478534698486328,
      "learning_rate": 5.941941941941942e-06,
      "loss": 0.2251,
      "step": 40540
    },
    {
      "epoch": 121.77,
      "grad_norm": 9.700529098510742,
      "learning_rate": 5.940940940940941e-06,
      "loss": 0.2384,
      "step": 40550
    },
    {
      "epoch": 121.8,
      "grad_norm": 15.877737998962402,
      "learning_rate": 5.93993993993994e-06,
      "loss": 0.2366,
      "step": 40560
    },
    {
      "epoch": 121.83,
      "grad_norm": 20.974533081054688,
      "learning_rate": 5.93893893893894e-06,
      "loss": 0.252,
      "step": 40570
    },
    {
      "epoch": 121.86,
      "grad_norm": 10.506731033325195,
      "learning_rate": 5.9379379379379384e-06,
      "loss": 0.2553,
      "step": 40580
    },
    {
      "epoch": 121.89,
      "grad_norm": 10.3152437210083,
      "learning_rate": 5.936936936936938e-06,
      "loss": 0.2108,
      "step": 40590
    },
    {
      "epoch": 121.92,
      "grad_norm": 8.070435523986816,
      "learning_rate": 5.935935935935936e-06,
      "loss": 0.2523,
      "step": 40600
    },
    {
      "epoch": 121.95,
      "grad_norm": 8.442362785339355,
      "learning_rate": 5.934934934934935e-06,
      "loss": 0.2315,
      "step": 40610
    },
    {
      "epoch": 121.98,
      "grad_norm": 6.422471523284912,
      "learning_rate": 5.9339339339339345e-06,
      "loss": 0.217,
      "step": 40620
    },
    {
      "epoch": 122.0,
      "eval_accuracy": 0.9263,
      "eval_loss": 0.313806414604187,
      "eval_runtime": 30.0494,
      "eval_samples_per_second": 332.785,
      "eval_steps_per_second": 1.331,
      "step": 40626
    },
    {
      "epoch": 122.01,
      "grad_norm": 10.457489013671875,
      "learning_rate": 5.932932932932933e-06,
      "loss": 0.2459,
      "step": 40630
    },
    {
      "epoch": 122.04,
      "grad_norm": 7.527783393859863,
      "learning_rate": 5.9319319319319326e-06,
      "loss": 0.2666,
      "step": 40640
    },
    {
      "epoch": 122.07,
      "grad_norm": 6.706661701202393,
      "learning_rate": 5.930930930930931e-06,
      "loss": 0.2711,
      "step": 40650
    },
    {
      "epoch": 122.1,
      "grad_norm": 8.492745399475098,
      "learning_rate": 5.929929929929931e-06,
      "loss": 0.2285,
      "step": 40660
    },
    {
      "epoch": 122.13,
      "grad_norm": 7.409997463226318,
      "learning_rate": 5.92892892892893e-06,
      "loss": 0.1976,
      "step": 40670
    },
    {
      "epoch": 122.16,
      "grad_norm": 6.869611740112305,
      "learning_rate": 5.927927927927929e-06,
      "loss": 0.2145,
      "step": 40680
    },
    {
      "epoch": 122.19,
      "grad_norm": 7.216169834136963,
      "learning_rate": 5.926926926926927e-06,
      "loss": 0.2789,
      "step": 40690
    },
    {
      "epoch": 122.22,
      "grad_norm": 8.947678565979004,
      "learning_rate": 5.925925925925926e-06,
      "loss": 0.2849,
      "step": 40700
    },
    {
      "epoch": 122.25,
      "grad_norm": 7.528345584869385,
      "learning_rate": 5.924924924924925e-06,
      "loss": 0.2735,
      "step": 40710
    },
    {
      "epoch": 122.28,
      "grad_norm": 33.67399978637695,
      "learning_rate": 5.923923923923925e-06,
      "loss": 0.2959,
      "step": 40720
    },
    {
      "epoch": 122.31,
      "grad_norm": 12.050817489624023,
      "learning_rate": 5.922922922922923e-06,
      "loss": 0.2777,
      "step": 40730
    },
    {
      "epoch": 122.34,
      "grad_norm": 10.078775405883789,
      "learning_rate": 5.921921921921923e-06,
      "loss": 0.2785,
      "step": 40740
    },
    {
      "epoch": 122.37,
      "grad_norm": 7.564576148986816,
      "learning_rate": 5.920920920920921e-06,
      "loss": 0.2746,
      "step": 40750
    },
    {
      "epoch": 122.4,
      "grad_norm": 9.415987968444824,
      "learning_rate": 5.919919919919921e-06,
      "loss": 0.2314,
      "step": 40760
    },
    {
      "epoch": 122.43,
      "grad_norm": 55.57491683959961,
      "learning_rate": 5.9189189189189185e-06,
      "loss": 0.2344,
      "step": 40770
    },
    {
      "epoch": 122.46,
      "grad_norm": 8.717290878295898,
      "learning_rate": 5.917917917917918e-06,
      "loss": 0.2757,
      "step": 40780
    },
    {
      "epoch": 122.49,
      "grad_norm": 11.224674224853516,
      "learning_rate": 5.916916916916917e-06,
      "loss": 0.2291,
      "step": 40790
    },
    {
      "epoch": 122.52,
      "grad_norm": 7.371959686279297,
      "learning_rate": 5.915915915915916e-06,
      "loss": 0.2432,
      "step": 40800
    },
    {
      "epoch": 122.55,
      "grad_norm": 9.831378936767578,
      "learning_rate": 5.9149149149149154e-06,
      "loss": 0.2658,
      "step": 40810
    },
    {
      "epoch": 122.58,
      "grad_norm": 8.524735450744629,
      "learning_rate": 5.913913913913915e-06,
      "loss": 0.2284,
      "step": 40820
    },
    {
      "epoch": 122.61,
      "grad_norm": 18.738880157470703,
      "learning_rate": 5.9129129129129135e-06,
      "loss": 0.2511,
      "step": 40830
    },
    {
      "epoch": 122.64,
      "grad_norm": 6.88028621673584,
      "learning_rate": 5.911911911911913e-06,
      "loss": 0.2617,
      "step": 40840
    },
    {
      "epoch": 122.67,
      "grad_norm": 10.203615188598633,
      "learning_rate": 5.910910910910911e-06,
      "loss": 0.2065,
      "step": 40850
    },
    {
      "epoch": 122.7,
      "grad_norm": 11.979241371154785,
      "learning_rate": 5.90990990990991e-06,
      "loss": 0.2544,
      "step": 40860
    },
    {
      "epoch": 122.73,
      "grad_norm": 10.397370338439941,
      "learning_rate": 5.908908908908909e-06,
      "loss": 0.2315,
      "step": 40870
    },
    {
      "epoch": 122.76,
      "grad_norm": 8.946357727050781,
      "learning_rate": 5.907907907907908e-06,
      "loss": 0.2552,
      "step": 40880
    },
    {
      "epoch": 122.79,
      "grad_norm": 7.387339115142822,
      "learning_rate": 5.906906906906908e-06,
      "loss": 0.1964,
      "step": 40890
    },
    {
      "epoch": 122.82,
      "grad_norm": 8.894357681274414,
      "learning_rate": 5.905905905905906e-06,
      "loss": 0.2004,
      "step": 40900
    },
    {
      "epoch": 122.85,
      "grad_norm": 5.423921585083008,
      "learning_rate": 5.904904904904906e-06,
      "loss": 0.2676,
      "step": 40910
    },
    {
      "epoch": 122.88,
      "grad_norm": 6.91759729385376,
      "learning_rate": 5.903903903903905e-06,
      "loss": 0.2336,
      "step": 40920
    },
    {
      "epoch": 122.91,
      "grad_norm": 7.051758289337158,
      "learning_rate": 5.902902902902904e-06,
      "loss": 0.2323,
      "step": 40930
    },
    {
      "epoch": 122.94,
      "grad_norm": 11.5159912109375,
      "learning_rate": 5.901901901901902e-06,
      "loss": 0.2784,
      "step": 40940
    },
    {
      "epoch": 122.97,
      "grad_norm": 9.094457626342773,
      "learning_rate": 5.900900900900901e-06,
      "loss": 0.2405,
      "step": 40950
    },
    {
      "epoch": 123.0,
      "eval_accuracy": 0.9275,
      "eval_loss": 0.3087535500526428,
      "eval_runtime": 29.7476,
      "eval_samples_per_second": 336.162,
      "eval_steps_per_second": 1.345,
      "step": 40959
    },
    {
      "epoch": 123.0,
      "grad_norm": 10.527048110961914,
      "learning_rate": 5.8998998998999e-06,
      "loss": 0.2464,
      "step": 40960
    },
    {
      "epoch": 123.03,
      "grad_norm": 5.695865154266357,
      "learning_rate": 5.8988988988989e-06,
      "loss": 0.2176,
      "step": 40970
    },
    {
      "epoch": 123.06,
      "grad_norm": 51.22569274902344,
      "learning_rate": 5.897897897897898e-06,
      "loss": 0.2914,
      "step": 40980
    },
    {
      "epoch": 123.09,
      "grad_norm": 7.0794267654418945,
      "learning_rate": 5.896896896896898e-06,
      "loss": 0.2318,
      "step": 40990
    },
    {
      "epoch": 123.12,
      "grad_norm": 7.256864070892334,
      "learning_rate": 5.895895895895896e-06,
      "loss": 0.2482,
      "step": 41000
    },
    {
      "epoch": 123.15,
      "grad_norm": 5.383802890777588,
      "learning_rate": 5.894894894894896e-06,
      "loss": 0.25,
      "step": 41010
    },
    {
      "epoch": 123.18,
      "grad_norm": 4.483356475830078,
      "learning_rate": 5.8938938938938936e-06,
      "loss": 0.2232,
      "step": 41020
    },
    {
      "epoch": 123.21,
      "grad_norm": 7.832893371582031,
      "learning_rate": 5.892892892892893e-06,
      "loss": 0.2913,
      "step": 41030
    },
    {
      "epoch": 123.24,
      "grad_norm": 6.656818866729736,
      "learning_rate": 5.8918918918918924e-06,
      "loss": 0.2501,
      "step": 41040
    },
    {
      "epoch": 123.27,
      "grad_norm": 8.236028671264648,
      "learning_rate": 5.890890890890891e-06,
      "loss": 0.2537,
      "step": 41050
    },
    {
      "epoch": 123.3,
      "grad_norm": 23.580747604370117,
      "learning_rate": 5.8898898898898905e-06,
      "loss": 0.265,
      "step": 41060
    },
    {
      "epoch": 123.33,
      "grad_norm": 9.049362182617188,
      "learning_rate": 5.88888888888889e-06,
      "loss": 0.2567,
      "step": 41070
    },
    {
      "epoch": 123.36,
      "grad_norm": 11.91850471496582,
      "learning_rate": 5.8878878878878885e-06,
      "loss": 0.2728,
      "step": 41080
    },
    {
      "epoch": 123.39,
      "grad_norm": 8.23604965209961,
      "learning_rate": 5.886886886886888e-06,
      "loss": 0.2631,
      "step": 41090
    },
    {
      "epoch": 123.42,
      "grad_norm": 6.493373870849609,
      "learning_rate": 5.885885885885886e-06,
      "loss": 0.2223,
      "step": 41100
    },
    {
      "epoch": 123.45,
      "grad_norm": 11.811920166015625,
      "learning_rate": 5.884884884884885e-06,
      "loss": 0.2395,
      "step": 41110
    },
    {
      "epoch": 123.48,
      "grad_norm": 11.986960411071777,
      "learning_rate": 5.883883883883884e-06,
      "loss": 0.2194,
      "step": 41120
    },
    {
      "epoch": 123.51,
      "grad_norm": 10.33935546875,
      "learning_rate": 5.882882882882883e-06,
      "loss": 0.2463,
      "step": 41130
    },
    {
      "epoch": 123.54,
      "grad_norm": 27.385391235351562,
      "learning_rate": 5.881881881881883e-06,
      "loss": 0.2806,
      "step": 41140
    },
    {
      "epoch": 123.57,
      "grad_norm": 26.815433502197266,
      "learning_rate": 5.880880880880881e-06,
      "loss": 0.3172,
      "step": 41150
    },
    {
      "epoch": 123.6,
      "grad_norm": 6.159441947937012,
      "learning_rate": 5.879879879879881e-06,
      "loss": 0.2337,
      "step": 41160
    },
    {
      "epoch": 123.63,
      "grad_norm": 12.259489059448242,
      "learning_rate": 5.87887887887888e-06,
      "loss": 0.2762,
      "step": 41170
    },
    {
      "epoch": 123.66,
      "grad_norm": 9.332571029663086,
      "learning_rate": 5.877877877877879e-06,
      "loss": 0.249,
      "step": 41180
    },
    {
      "epoch": 123.69,
      "grad_norm": 9.329368591308594,
      "learning_rate": 5.876876876876877e-06,
      "loss": 0.2697,
      "step": 41190
    },
    {
      "epoch": 123.72,
      "grad_norm": 10.854620933532715,
      "learning_rate": 5.875875875875876e-06,
      "loss": 0.2689,
      "step": 41200
    },
    {
      "epoch": 123.75,
      "grad_norm": 13.480330467224121,
      "learning_rate": 5.874874874874875e-06,
      "loss": 0.2777,
      "step": 41210
    },
    {
      "epoch": 123.78,
      "grad_norm": 11.054851531982422,
      "learning_rate": 5.873873873873874e-06,
      "loss": 0.2777,
      "step": 41220
    },
    {
      "epoch": 123.81,
      "grad_norm": 18.788312911987305,
      "learning_rate": 5.872872872872873e-06,
      "loss": 0.2716,
      "step": 41230
    },
    {
      "epoch": 123.84,
      "grad_norm": 6.95937967300415,
      "learning_rate": 5.871871871871873e-06,
      "loss": 0.2803,
      "step": 41240
    },
    {
      "epoch": 123.87,
      "grad_norm": 8.353614807128906,
      "learning_rate": 5.870870870870871e-06,
      "loss": 0.2448,
      "step": 41250
    },
    {
      "epoch": 123.9,
      "grad_norm": 10.957862854003906,
      "learning_rate": 5.869869869869871e-06,
      "loss": 0.2771,
      "step": 41260
    },
    {
      "epoch": 123.93,
      "grad_norm": 7.600067615509033,
      "learning_rate": 5.868868868868869e-06,
      "loss": 0.2573,
      "step": 41270
    },
    {
      "epoch": 123.96,
      "grad_norm": 8.460509300231934,
      "learning_rate": 5.867867867867868e-06,
      "loss": 0.2618,
      "step": 41280
    },
    {
      "epoch": 123.99,
      "grad_norm": 8.201217651367188,
      "learning_rate": 5.8668668668668675e-06,
      "loss": 0.2972,
      "step": 41290
    },
    {
      "epoch": 124.0,
      "eval_accuracy": 0.9255,
      "eval_loss": 0.3067146837711334,
      "eval_runtime": 30.1613,
      "eval_samples_per_second": 331.55,
      "eval_steps_per_second": 1.326,
      "step": 41292
    },
    {
      "epoch": 124.02,
      "grad_norm": 5.98701286315918,
      "learning_rate": 5.865865865865866e-06,
      "loss": 0.2171,
      "step": 41300
    },
    {
      "epoch": 124.05,
      "grad_norm": 8.219392776489258,
      "learning_rate": 5.8648648648648655e-06,
      "loss": 0.2684,
      "step": 41310
    },
    {
      "epoch": 124.08,
      "grad_norm": 20.47637367248535,
      "learning_rate": 5.863863863863864e-06,
      "loss": 0.2484,
      "step": 41320
    },
    {
      "epoch": 124.11,
      "grad_norm": 10.43238353729248,
      "learning_rate": 5.8628628628628635e-06,
      "loss": 0.2591,
      "step": 41330
    },
    {
      "epoch": 124.14,
      "grad_norm": 8.776142120361328,
      "learning_rate": 5.861861861861863e-06,
      "loss": 0.2927,
      "step": 41340
    },
    {
      "epoch": 124.17,
      "grad_norm": 14.211291313171387,
      "learning_rate": 5.860860860860861e-06,
      "loss": 0.2449,
      "step": 41350
    },
    {
      "epoch": 124.2,
      "grad_norm": 8.627349853515625,
      "learning_rate": 5.85985985985986e-06,
      "loss": 0.2679,
      "step": 41360
    },
    {
      "epoch": 124.23,
      "grad_norm": 11.986977577209473,
      "learning_rate": 5.858858858858859e-06,
      "loss": 0.2792,
      "step": 41370
    },
    {
      "epoch": 124.26,
      "grad_norm": 4.700423240661621,
      "learning_rate": 5.857857857857858e-06,
      "loss": 0.2446,
      "step": 41380
    },
    {
      "epoch": 124.29,
      "grad_norm": 9.038325309753418,
      "learning_rate": 5.856856856856858e-06,
      "loss": 0.2597,
      "step": 41390
    },
    {
      "epoch": 124.32,
      "grad_norm": 8.30774974822998,
      "learning_rate": 5.855855855855856e-06,
      "loss": 0.2379,
      "step": 41400
    },
    {
      "epoch": 124.35,
      "grad_norm": 6.741456031799316,
      "learning_rate": 5.854854854854856e-06,
      "loss": 0.2505,
      "step": 41410
    },
    {
      "epoch": 124.38,
      "grad_norm": 4.52980375289917,
      "learning_rate": 5.853853853853855e-06,
      "loss": 0.2472,
      "step": 41420
    },
    {
      "epoch": 124.41,
      "grad_norm": 15.186450004577637,
      "learning_rate": 5.852852852852854e-06,
      "loss": 0.2934,
      "step": 41430
    },
    {
      "epoch": 124.44,
      "grad_norm": 8.189556121826172,
      "learning_rate": 5.8518518518518515e-06,
      "loss": 0.2691,
      "step": 41440
    },
    {
      "epoch": 124.47,
      "grad_norm": 4.249034404754639,
      "learning_rate": 5.850850850850851e-06,
      "loss": 0.2795,
      "step": 41450
    },
    {
      "epoch": 124.5,
      "grad_norm": 6.574668884277344,
      "learning_rate": 5.84984984984985e-06,
      "loss": 0.2238,
      "step": 41460
    },
    {
      "epoch": 124.53,
      "grad_norm": 4.8524980545043945,
      "learning_rate": 5.848848848848849e-06,
      "loss": 0.2025,
      "step": 41470
    },
    {
      "epoch": 124.56,
      "grad_norm": 11.420845031738281,
      "learning_rate": 5.847847847847848e-06,
      "loss": 0.2823,
      "step": 41480
    },
    {
      "epoch": 124.59,
      "grad_norm": 9.009490013122559,
      "learning_rate": 5.846846846846848e-06,
      "loss": 0.2428,
      "step": 41490
    },
    {
      "epoch": 124.62,
      "grad_norm": 8.688403129577637,
      "learning_rate": 5.8458458458458464e-06,
      "loss": 0.2672,
      "step": 41500
    },
    {
      "epoch": 124.65,
      "grad_norm": 15.645672798156738,
      "learning_rate": 5.844844844844846e-06,
      "loss": 0.2743,
      "step": 41510
    },
    {
      "epoch": 124.68,
      "grad_norm": 12.134678840637207,
      "learning_rate": 5.843843843843844e-06,
      "loss": 0.2288,
      "step": 41520
    },
    {
      "epoch": 124.71,
      "grad_norm": 7.707067966461182,
      "learning_rate": 5.842842842842843e-06,
      "loss": 0.279,
      "step": 41530
    },
    {
      "epoch": 124.74,
      "grad_norm": 9.46369457244873,
      "learning_rate": 5.8418418418418425e-06,
      "loss": 0.2381,
      "step": 41540
    },
    {
      "epoch": 124.77,
      "grad_norm": 6.625874996185303,
      "learning_rate": 5.840840840840841e-06,
      "loss": 0.2527,
      "step": 41550
    },
    {
      "epoch": 124.8,
      "grad_norm": 6.67770528793335,
      "learning_rate": 5.8398398398398405e-06,
      "loss": 0.2818,
      "step": 41560
    },
    {
      "epoch": 124.83,
      "grad_norm": 6.213510036468506,
      "learning_rate": 5.838838838838839e-06,
      "loss": 0.235,
      "step": 41570
    },
    {
      "epoch": 124.86,
      "grad_norm": 8.969176292419434,
      "learning_rate": 5.837837837837839e-06,
      "loss": 0.2592,
      "step": 41580
    },
    {
      "epoch": 124.89,
      "grad_norm": 6.711195468902588,
      "learning_rate": 5.836836836836838e-06,
      "loss": 0.2331,
      "step": 41590
    },
    {
      "epoch": 124.92,
      "grad_norm": 11.762642860412598,
      "learning_rate": 5.835835835835836e-06,
      "loss": 0.2649,
      "step": 41600
    },
    {
      "epoch": 124.95,
      "grad_norm": 14.33794116973877,
      "learning_rate": 5.834834834834835e-06,
      "loss": 0.2755,
      "step": 41610
    },
    {
      "epoch": 124.98,
      "grad_norm": 51.5206184387207,
      "learning_rate": 5.833833833833834e-06,
      "loss": 0.246,
      "step": 41620
    },
    {
      "epoch": 125.0,
      "eval_accuracy": 0.9252,
      "eval_loss": 0.31195691227912903,
      "eval_runtime": 29.8465,
      "eval_samples_per_second": 335.048,
      "eval_steps_per_second": 1.34,
      "step": 41625
    },
    {
      "epoch": 125.02,
      "grad_norm": 10.778359413146973,
      "learning_rate": 5.832832832832833e-06,
      "loss": 0.1979,
      "step": 41630
    },
    {
      "epoch": 125.05,
      "grad_norm": 4.487590789794922,
      "learning_rate": 5.831831831831833e-06,
      "loss": 0.2464,
      "step": 41640
    },
    {
      "epoch": 125.08,
      "grad_norm": 22.7575626373291,
      "learning_rate": 5.830830830830831e-06,
      "loss": 0.2884,
      "step": 41650
    },
    {
      "epoch": 125.11,
      "grad_norm": 8.09821891784668,
      "learning_rate": 5.829829829829831e-06,
      "loss": 0.2328,
      "step": 41660
    },
    {
      "epoch": 125.14,
      "grad_norm": 10.964784622192383,
      "learning_rate": 5.828828828828829e-06,
      "loss": 0.2723,
      "step": 41670
    },
    {
      "epoch": 125.17,
      "grad_norm": 9.29461669921875,
      "learning_rate": 5.827827827827829e-06,
      "loss": 0.2487,
      "step": 41680
    },
    {
      "epoch": 125.2,
      "grad_norm": 6.174034595489502,
      "learning_rate": 5.8268268268268265e-06,
      "loss": 0.2492,
      "step": 41690
    },
    {
      "epoch": 125.23,
      "grad_norm": 11.633269309997559,
      "learning_rate": 5.825825825825826e-06,
      "loss": 0.2682,
      "step": 41700
    },
    {
      "epoch": 125.26,
      "grad_norm": 10.736008644104004,
      "learning_rate": 5.824824824824825e-06,
      "loss": 0.2543,
      "step": 41710
    },
    {
      "epoch": 125.29,
      "grad_norm": 8.981684684753418,
      "learning_rate": 5.823823823823824e-06,
      "loss": 0.2547,
      "step": 41720
    },
    {
      "epoch": 125.32,
      "grad_norm": 6.287871360778809,
      "learning_rate": 5.8228228228228234e-06,
      "loss": 0.2698,
      "step": 41730
    },
    {
      "epoch": 125.35,
      "grad_norm": 17.21749496459961,
      "learning_rate": 5.821821821821823e-06,
      "loss": 0.2296,
      "step": 41740
    },
    {
      "epoch": 125.38,
      "grad_norm": 8.046335220336914,
      "learning_rate": 5.8208208208208215e-06,
      "loss": 0.2109,
      "step": 41750
    },
    {
      "epoch": 125.41,
      "grad_norm": 8.993094444274902,
      "learning_rate": 5.819819819819821e-06,
      "loss": 0.2497,
      "step": 41760
    },
    {
      "epoch": 125.44,
      "grad_norm": 15.984443664550781,
      "learning_rate": 5.818818818818819e-06,
      "loss": 0.2621,
      "step": 41770
    },
    {
      "epoch": 125.47,
      "grad_norm": 9.101130485534668,
      "learning_rate": 5.817817817817818e-06,
      "loss": 0.2571,
      "step": 41780
    },
    {
      "epoch": 125.5,
      "grad_norm": 6.95971155166626,
      "learning_rate": 5.816816816816817e-06,
      "loss": 0.2264,
      "step": 41790
    },
    {
      "epoch": 125.53,
      "grad_norm": 11.636286735534668,
      "learning_rate": 5.815815815815816e-06,
      "loss": 0.2311,
      "step": 41800
    },
    {
      "epoch": 125.56,
      "grad_norm": 22.06410026550293,
      "learning_rate": 5.814814814814816e-06,
      "loss": 0.3051,
      "step": 41810
    },
    {
      "epoch": 125.59,
      "grad_norm": 8.873347282409668,
      "learning_rate": 5.813813813813814e-06,
      "loss": 0.2283,
      "step": 41820
    },
    {
      "epoch": 125.62,
      "grad_norm": 7.64946174621582,
      "learning_rate": 5.812812812812814e-06,
      "loss": 0.255,
      "step": 41830
    },
    {
      "epoch": 125.65,
      "grad_norm": 8.641168594360352,
      "learning_rate": 5.811811811811813e-06,
      "loss": 0.198,
      "step": 41840
    },
    {
      "epoch": 125.68,
      "grad_norm": 11.588932991027832,
      "learning_rate": 5.810810810810811e-06,
      "loss": 0.2682,
      "step": 41850
    },
    {
      "epoch": 125.71,
      "grad_norm": 10.1900634765625,
      "learning_rate": 5.80980980980981e-06,
      "loss": 0.2299,
      "step": 41860
    },
    {
      "epoch": 125.74,
      "grad_norm": 39.35232925415039,
      "learning_rate": 5.808808808808809e-06,
      "loss": 0.2763,
      "step": 41870
    },
    {
      "epoch": 125.77,
      "grad_norm": 7.716683387756348,
      "learning_rate": 5.807807807807808e-06,
      "loss": 0.2516,
      "step": 41880
    },
    {
      "epoch": 125.8,
      "grad_norm": 8.220368385314941,
      "learning_rate": 5.806806806806808e-06,
      "loss": 0.267,
      "step": 41890
    },
    {
      "epoch": 125.83,
      "grad_norm": 10.147479057312012,
      "learning_rate": 5.805805805805806e-06,
      "loss": 0.2888,
      "step": 41900
    },
    {
      "epoch": 125.86,
      "grad_norm": 9.059749603271484,
      "learning_rate": 5.804804804804806e-06,
      "loss": 0.2267,
      "step": 41910
    },
    {
      "epoch": 125.89,
      "grad_norm": 12.708621978759766,
      "learning_rate": 5.803803803803804e-06,
      "loss": 0.2591,
      "step": 41920
    },
    {
      "epoch": 125.92,
      "grad_norm": 12.902271270751953,
      "learning_rate": 5.802802802802803e-06,
      "loss": 0.2797,
      "step": 41930
    },
    {
      "epoch": 125.95,
      "grad_norm": 10.018285751342773,
      "learning_rate": 5.8018018018018015e-06,
      "loss": 0.2821,
      "step": 41940
    },
    {
      "epoch": 125.98,
      "grad_norm": 13.820377349853516,
      "learning_rate": 5.800800800800801e-06,
      "loss": 0.273,
      "step": 41950
    },
    {
      "epoch": 126.0,
      "eval_accuracy": 0.9247,
      "eval_loss": 0.3164510130882263,
      "eval_runtime": 30.1258,
      "eval_samples_per_second": 331.942,
      "eval_steps_per_second": 1.328,
      "step": 41958
    },
    {
      "epoch": 126.01,
      "grad_norm": 10.589327812194824,
      "learning_rate": 5.7997997997998004e-06,
      "loss": 0.2414,
      "step": 41960
    },
    {
      "epoch": 126.04,
      "grad_norm": 7.8583831787109375,
      "learning_rate": 5.798798798798799e-06,
      "loss": 0.2801,
      "step": 41970
    },
    {
      "epoch": 126.07,
      "grad_norm": 6.744598388671875,
      "learning_rate": 5.7977977977977985e-06,
      "loss": 0.2136,
      "step": 41980
    },
    {
      "epoch": 126.1,
      "grad_norm": 7.299524307250977,
      "learning_rate": 5.796796796796798e-06,
      "loss": 0.2671,
      "step": 41990
    },
    {
      "epoch": 126.13,
      "grad_norm": 9.397318840026855,
      "learning_rate": 5.7957957957957965e-06,
      "loss": 0.2159,
      "step": 42000
    },
    {
      "epoch": 126.16,
      "grad_norm": 12.326234817504883,
      "learning_rate": 5.794794794794796e-06,
      "loss": 0.2431,
      "step": 42010
    },
    {
      "epoch": 126.19,
      "grad_norm": 8.582897186279297,
      "learning_rate": 5.793793793793794e-06,
      "loss": 0.1781,
      "step": 42020
    },
    {
      "epoch": 126.22,
      "grad_norm": 8.4203462600708,
      "learning_rate": 5.792792792792793e-06,
      "loss": 0.2535,
      "step": 42030
    },
    {
      "epoch": 126.25,
      "grad_norm": 12.643244743347168,
      "learning_rate": 5.791791791791792e-06,
      "loss": 0.3288,
      "step": 42040
    },
    {
      "epoch": 126.28,
      "grad_norm": 9.014124870300293,
      "learning_rate": 5.790790790790791e-06,
      "loss": 0.2408,
      "step": 42050
    },
    {
      "epoch": 126.31,
      "grad_norm": 7.7596025466918945,
      "learning_rate": 5.789789789789791e-06,
      "loss": 0.2793,
      "step": 42060
    },
    {
      "epoch": 126.34,
      "grad_norm": 12.365252494812012,
      "learning_rate": 5.788788788788789e-06,
      "loss": 0.3036,
      "step": 42070
    },
    {
      "epoch": 126.37,
      "grad_norm": 6.975355625152588,
      "learning_rate": 5.787787787787789e-06,
      "loss": 0.23,
      "step": 42080
    },
    {
      "epoch": 126.4,
      "grad_norm": 6.220841884613037,
      "learning_rate": 5.786786786786788e-06,
      "loss": 0.2826,
      "step": 42090
    },
    {
      "epoch": 126.43,
      "grad_norm": 10.2344331741333,
      "learning_rate": 5.785785785785786e-06,
      "loss": 0.2644,
      "step": 42100
    },
    {
      "epoch": 126.46,
      "grad_norm": 6.122129917144775,
      "learning_rate": 5.784784784784785e-06,
      "loss": 0.2332,
      "step": 42110
    },
    {
      "epoch": 126.49,
      "grad_norm": 8.929717063903809,
      "learning_rate": 5.783783783783784e-06,
      "loss": 0.2756,
      "step": 42120
    },
    {
      "epoch": 126.52,
      "grad_norm": 24.192073822021484,
      "learning_rate": 5.782782782782783e-06,
      "loss": 0.2715,
      "step": 42130
    },
    {
      "epoch": 126.55,
      "grad_norm": 11.633360862731934,
      "learning_rate": 5.781781781781782e-06,
      "loss": 0.2076,
      "step": 42140
    },
    {
      "epoch": 126.58,
      "grad_norm": 12.009920120239258,
      "learning_rate": 5.780780780780781e-06,
      "loss": 0.2438,
      "step": 42150
    },
    {
      "epoch": 126.61,
      "grad_norm": 9.346396446228027,
      "learning_rate": 5.779779779779781e-06,
      "loss": 0.2527,
      "step": 42160
    },
    {
      "epoch": 126.64,
      "grad_norm": 6.676176071166992,
      "learning_rate": 5.778778778778779e-06,
      "loss": 0.2744,
      "step": 42170
    },
    {
      "epoch": 126.67,
      "grad_norm": 18.55327033996582,
      "learning_rate": 5.777777777777778e-06,
      "loss": 0.2564,
      "step": 42180
    },
    {
      "epoch": 126.7,
      "grad_norm": 6.718024730682373,
      "learning_rate": 5.776776776776777e-06,
      "loss": 0.2165,
      "step": 42190
    },
    {
      "epoch": 126.73,
      "grad_norm": 7.504716873168945,
      "learning_rate": 5.775775775775776e-06,
      "loss": 0.236,
      "step": 42200
    },
    {
      "epoch": 126.76,
      "grad_norm": 8.345107078552246,
      "learning_rate": 5.7747747747747755e-06,
      "loss": 0.2582,
      "step": 42210
    },
    {
      "epoch": 126.79,
      "grad_norm": 13.784915924072266,
      "learning_rate": 5.773773773773774e-06,
      "loss": 0.2308,
      "step": 42220
    },
    {
      "epoch": 126.82,
      "grad_norm": 8.982242584228516,
      "learning_rate": 5.7727727727727735e-06,
      "loss": 0.2461,
      "step": 42230
    },
    {
      "epoch": 126.85,
      "grad_norm": 12.292370796203613,
      "learning_rate": 5.771771771771772e-06,
      "loss": 0.2421,
      "step": 42240
    },
    {
      "epoch": 126.88,
      "grad_norm": 11.715461730957031,
      "learning_rate": 5.7707707707707715e-06,
      "loss": 0.2563,
      "step": 42250
    },
    {
      "epoch": 126.91,
      "grad_norm": 9.39710521697998,
      "learning_rate": 5.769769769769771e-06,
      "loss": 0.258,
      "step": 42260
    },
    {
      "epoch": 126.94,
      "grad_norm": 28.933481216430664,
      "learning_rate": 5.768768768768769e-06,
      "loss": 0.2698,
      "step": 42270
    },
    {
      "epoch": 126.97,
      "grad_norm": 11.193105697631836,
      "learning_rate": 5.767767767767768e-06,
      "loss": 0.2609,
      "step": 42280
    },
    {
      "epoch": 127.0,
      "grad_norm": 8.381494522094727,
      "learning_rate": 5.766766766766767e-06,
      "loss": 0.2837,
      "step": 42290
    },
    {
      "epoch": 127.0,
      "eval_accuracy": 0.925,
      "eval_loss": 0.3159259259700775,
      "eval_runtime": 30.1853,
      "eval_samples_per_second": 331.287,
      "eval_steps_per_second": 1.325,
      "step": 42291
    },
    {
      "epoch": 127.03,
      "grad_norm": 29.495254516601562,
      "learning_rate": 5.765765765765766e-06,
      "loss": 0.2287,
      "step": 42300
    },
    {
      "epoch": 127.06,
      "grad_norm": 7.834287643432617,
      "learning_rate": 5.764764764764766e-06,
      "loss": 0.1945,
      "step": 42310
    },
    {
      "epoch": 127.09,
      "grad_norm": 7.395016193389893,
      "learning_rate": 5.763763763763764e-06,
      "loss": 0.2596,
      "step": 42320
    },
    {
      "epoch": 127.12,
      "grad_norm": 9.775337219238281,
      "learning_rate": 5.762762762762764e-06,
      "loss": 0.2763,
      "step": 42330
    },
    {
      "epoch": 127.15,
      "grad_norm": 15.93709945678711,
      "learning_rate": 5.761761761761763e-06,
      "loss": 0.3016,
      "step": 42340
    },
    {
      "epoch": 127.18,
      "grad_norm": 6.418456077575684,
      "learning_rate": 5.760760760760761e-06,
      "loss": 0.2435,
      "step": 42350
    },
    {
      "epoch": 127.21,
      "grad_norm": 7.613146781921387,
      "learning_rate": 5.7597597597597595e-06,
      "loss": 0.2588,
      "step": 42360
    },
    {
      "epoch": 127.24,
      "grad_norm": 5.041161060333252,
      "learning_rate": 5.758758758758759e-06,
      "loss": 0.2497,
      "step": 42370
    },
    {
      "epoch": 127.27,
      "grad_norm": 9.764585494995117,
      "learning_rate": 5.757757757757758e-06,
      "loss": 0.2892,
      "step": 42380
    },
    {
      "epoch": 127.3,
      "grad_norm": 16.15337562561035,
      "learning_rate": 5.756756756756757e-06,
      "loss": 0.2398,
      "step": 42390
    },
    {
      "epoch": 127.33,
      "grad_norm": 8.539102554321289,
      "learning_rate": 5.755755755755756e-06,
      "loss": 0.2548,
      "step": 42400
    },
    {
      "epoch": 127.36,
      "grad_norm": 9.03654956817627,
      "learning_rate": 5.754754754754756e-06,
      "loss": 0.2233,
      "step": 42410
    },
    {
      "epoch": 127.39,
      "grad_norm": 6.903131484985352,
      "learning_rate": 5.7537537537537544e-06,
      "loss": 0.2463,
      "step": 42420
    },
    {
      "epoch": 127.42,
      "grad_norm": 12.4562406539917,
      "learning_rate": 5.752752752752753e-06,
      "loss": 0.2949,
      "step": 42430
    },
    {
      "epoch": 127.45,
      "grad_norm": 8.9762601852417,
      "learning_rate": 5.751751751751752e-06,
      "loss": 0.2508,
      "step": 42440
    },
    {
      "epoch": 127.48,
      "grad_norm": 7.099888324737549,
      "learning_rate": 5.750750750750751e-06,
      "loss": 0.2528,
      "step": 42450
    },
    {
      "epoch": 127.51,
      "grad_norm": 12.848033905029297,
      "learning_rate": 5.7497497497497505e-06,
      "loss": 0.2235,
      "step": 42460
    },
    {
      "epoch": 127.54,
      "grad_norm": 9.66506576538086,
      "learning_rate": 5.748748748748749e-06,
      "loss": 0.2465,
      "step": 42470
    },
    {
      "epoch": 127.57,
      "grad_norm": 12.026232719421387,
      "learning_rate": 5.7477477477477485e-06,
      "loss": 0.254,
      "step": 42480
    },
    {
      "epoch": 127.6,
      "grad_norm": 13.918740272521973,
      "learning_rate": 5.746746746746747e-06,
      "loss": 0.2585,
      "step": 42490
    },
    {
      "epoch": 127.63,
      "grad_norm": 11.105405807495117,
      "learning_rate": 5.7457457457457466e-06,
      "loss": 0.2615,
      "step": 42500
    },
    {
      "epoch": 127.66,
      "grad_norm": 5.088107585906982,
      "learning_rate": 5.744744744744746e-06,
      "loss": 0.2454,
      "step": 42510
    },
    {
      "epoch": 127.69,
      "grad_norm": 11.766748428344727,
      "learning_rate": 5.743743743743744e-06,
      "loss": 0.223,
      "step": 42520
    },
    {
      "epoch": 127.72,
      "grad_norm": 6.142477989196777,
      "learning_rate": 5.742742742742743e-06,
      "loss": 0.2698,
      "step": 42530
    },
    {
      "epoch": 127.75,
      "grad_norm": 7.024839878082275,
      "learning_rate": 5.741741741741742e-06,
      "loss": 0.2681,
      "step": 42540
    },
    {
      "epoch": 127.78,
      "grad_norm": 6.171216011047363,
      "learning_rate": 5.740740740740741e-06,
      "loss": 0.2739,
      "step": 42550
    },
    {
      "epoch": 127.81,
      "grad_norm": 13.886873245239258,
      "learning_rate": 5.739739739739741e-06,
      "loss": 0.2426,
      "step": 42560
    },
    {
      "epoch": 127.84,
      "grad_norm": 12.119695663452148,
      "learning_rate": 5.738738738738739e-06,
      "loss": 0.2796,
      "step": 42570
    },
    {
      "epoch": 127.87,
      "grad_norm": 10.00767707824707,
      "learning_rate": 5.737737737737739e-06,
      "loss": 0.2376,
      "step": 42580
    },
    {
      "epoch": 127.9,
      "grad_norm": 8.43599796295166,
      "learning_rate": 5.736736736736737e-06,
      "loss": 0.242,
      "step": 42590
    },
    {
      "epoch": 127.93,
      "grad_norm": 34.079925537109375,
      "learning_rate": 5.735735735735736e-06,
      "loss": 0.2906,
      "step": 42600
    },
    {
      "epoch": 127.96,
      "grad_norm": 9.975250244140625,
      "learning_rate": 5.7347347347347345e-06,
      "loss": 0.2602,
      "step": 42610
    },
    {
      "epoch": 127.99,
      "grad_norm": 25.59150505065918,
      "learning_rate": 5.733733733733734e-06,
      "loss": 0.2741,
      "step": 42620
    },
    {
      "epoch": 128.0,
      "eval_accuracy": 0.9256,
      "eval_loss": 0.3169495463371277,
      "eval_runtime": 30.3724,
      "eval_samples_per_second": 329.247,
      "eval_steps_per_second": 1.317,
      "step": 42624
    },
    {
      "epoch": 128.02,
      "grad_norm": 4.481472492218018,
      "learning_rate": 5.732732732732733e-06,
      "loss": 0.19,
      "step": 42630
    },
    {
      "epoch": 128.05,
      "grad_norm": 6.698709487915039,
      "learning_rate": 5.731731731731732e-06,
      "loss": 0.2534,
      "step": 42640
    },
    {
      "epoch": 128.08,
      "grad_norm": 6.116823196411133,
      "learning_rate": 5.7307307307307314e-06,
      "loss": 0.2621,
      "step": 42650
    },
    {
      "epoch": 128.11,
      "grad_norm": 8.991177558898926,
      "learning_rate": 5.729729729729731e-06,
      "loss": 0.2577,
      "step": 42660
    },
    {
      "epoch": 128.14,
      "grad_norm": 10.007184028625488,
      "learning_rate": 5.7287287287287295e-06,
      "loss": 0.227,
      "step": 42670
    },
    {
      "epoch": 128.17,
      "grad_norm": 15.651451110839844,
      "learning_rate": 5.727727727727728e-06,
      "loss": 0.2757,
      "step": 42680
    },
    {
      "epoch": 128.2,
      "grad_norm": 7.4448442459106445,
      "learning_rate": 5.726726726726727e-06,
      "loss": 0.2617,
      "step": 42690
    },
    {
      "epoch": 128.23,
      "grad_norm": 8.075090408325195,
      "learning_rate": 5.725725725725726e-06,
      "loss": 0.2307,
      "step": 42700
    },
    {
      "epoch": 128.26,
      "grad_norm": 6.565578937530518,
      "learning_rate": 5.724724724724725e-06,
      "loss": 0.2822,
      "step": 42710
    },
    {
      "epoch": 128.29,
      "grad_norm": 11.238733291625977,
      "learning_rate": 5.723723723723724e-06,
      "loss": 0.2838,
      "step": 42720
    },
    {
      "epoch": 128.32,
      "grad_norm": 6.175077438354492,
      "learning_rate": 5.7227227227227236e-06,
      "loss": 0.254,
      "step": 42730
    },
    {
      "epoch": 128.35,
      "grad_norm": 11.878280639648438,
      "learning_rate": 5.721721721721722e-06,
      "loss": 0.2512,
      "step": 42740
    },
    {
      "epoch": 128.38,
      "grad_norm": 8.817439079284668,
      "learning_rate": 5.720720720720722e-06,
      "loss": 0.2564,
      "step": 42750
    },
    {
      "epoch": 128.41,
      "grad_norm": 11.72377872467041,
      "learning_rate": 5.719719719719721e-06,
      "loss": 0.2242,
      "step": 42760
    },
    {
      "epoch": 128.44,
      "grad_norm": 9.267855644226074,
      "learning_rate": 5.718718718718719e-06,
      "loss": 0.2038,
      "step": 42770
    },
    {
      "epoch": 128.47,
      "grad_norm": 6.359633922576904,
      "learning_rate": 5.717717717717718e-06,
      "loss": 0.2686,
      "step": 42780
    },
    {
      "epoch": 128.5,
      "grad_norm": 10.478836059570312,
      "learning_rate": 5.716716716716717e-06,
      "loss": 0.2679,
      "step": 42790
    },
    {
      "epoch": 128.53,
      "grad_norm": 7.848367214202881,
      "learning_rate": 5.715715715715716e-06,
      "loss": 0.2905,
      "step": 42800
    },
    {
      "epoch": 128.56,
      "grad_norm": 6.314141273498535,
      "learning_rate": 5.714714714714715e-06,
      "loss": 0.2459,
      "step": 42810
    },
    {
      "epoch": 128.59,
      "grad_norm": 10.546624183654785,
      "learning_rate": 5.713713713713714e-06,
      "loss": 0.221,
      "step": 42820
    },
    {
      "epoch": 128.62,
      "grad_norm": 10.648294448852539,
      "learning_rate": 5.712712712712714e-06,
      "loss": 0.2403,
      "step": 42830
    },
    {
      "epoch": 128.65,
      "grad_norm": 8.058470726013184,
      "learning_rate": 5.711711711711712e-06,
      "loss": 0.2853,
      "step": 42840
    },
    {
      "epoch": 128.68,
      "grad_norm": 25.285390853881836,
      "learning_rate": 5.710710710710711e-06,
      "loss": 0.2664,
      "step": 42850
    },
    {
      "epoch": 128.71,
      "grad_norm": 13.980432510375977,
      "learning_rate": 5.7097097097097095e-06,
      "loss": 0.231,
      "step": 42860
    },
    {
      "epoch": 128.74,
      "grad_norm": 13.816667556762695,
      "learning_rate": 5.708708708708709e-06,
      "loss": 0.2721,
      "step": 42870
    },
    {
      "epoch": 128.77,
      "grad_norm": 21.366363525390625,
      "learning_rate": 5.707707707707708e-06,
      "loss": 0.2189,
      "step": 42880
    },
    {
      "epoch": 128.8,
      "grad_norm": 8.817442893981934,
      "learning_rate": 5.706706706706707e-06,
      "loss": 0.237,
      "step": 42890
    },
    {
      "epoch": 128.83,
      "grad_norm": 8.07828426361084,
      "learning_rate": 5.7057057057057065e-06,
      "loss": 0.2565,
      "step": 42900
    },
    {
      "epoch": 128.86,
      "grad_norm": 16.40593719482422,
      "learning_rate": 5.704704704704706e-06,
      "loss": 0.2526,
      "step": 42910
    },
    {
      "epoch": 128.89,
      "grad_norm": 9.361808776855469,
      "learning_rate": 5.7037037037037045e-06,
      "loss": 0.2379,
      "step": 42920
    },
    {
      "epoch": 128.92,
      "grad_norm": 7.4654645919799805,
      "learning_rate": 5.702702702702702e-06,
      "loss": 0.2459,
      "step": 42930
    },
    {
      "epoch": 128.95,
      "grad_norm": 6.993584156036377,
      "learning_rate": 5.701701701701702e-06,
      "loss": 0.256,
      "step": 42940
    },
    {
      "epoch": 128.98,
      "grad_norm": 12.155355453491211,
      "learning_rate": 5.700700700700701e-06,
      "loss": 0.26,
      "step": 42950
    },
    {
      "epoch": 129.0,
      "eval_accuracy": 0.9251,
      "eval_loss": 0.3156406879425049,
      "eval_runtime": 29.8688,
      "eval_samples_per_second": 334.797,
      "eval_steps_per_second": 1.339,
      "step": 42957
    },
    {
      "epoch": 129.01,
      "grad_norm": 10.493075370788574,
      "learning_rate": 5.6996996996997e-06,
      "loss": 0.2236,
      "step": 42960
    },
    {
      "epoch": 129.04,
      "grad_norm": 13.21835708618164,
      "learning_rate": 5.698698698698699e-06,
      "loss": 0.2517,
      "step": 42970
    },
    {
      "epoch": 129.07,
      "grad_norm": 11.07337474822998,
      "learning_rate": 5.697697697697699e-06,
      "loss": 0.2272,
      "step": 42980
    },
    {
      "epoch": 129.1,
      "grad_norm": 12.820321083068848,
      "learning_rate": 5.696696696696697e-06,
      "loss": 0.2913,
      "step": 42990
    },
    {
      "epoch": 129.13,
      "grad_norm": 14.408442497253418,
      "learning_rate": 5.695695695695697e-06,
      "loss": 0.2512,
      "step": 43000
    },
    {
      "epoch": 129.16,
      "grad_norm": 13.177003860473633,
      "learning_rate": 5.694694694694696e-06,
      "loss": 0.241,
      "step": 43010
    },
    {
      "epoch": 129.19,
      "grad_norm": 7.193375587463379,
      "learning_rate": 5.693693693693694e-06,
      "loss": 0.2545,
      "step": 43020
    },
    {
      "epoch": 129.22,
      "grad_norm": 6.5010552406311035,
      "learning_rate": 5.692692692692693e-06,
      "loss": 0.284,
      "step": 43030
    },
    {
      "epoch": 129.25,
      "grad_norm": 11.184617042541504,
      "learning_rate": 5.691691691691692e-06,
      "loss": 0.2051,
      "step": 43040
    },
    {
      "epoch": 129.28,
      "grad_norm": 19.71173858642578,
      "learning_rate": 5.690690690690691e-06,
      "loss": 0.233,
      "step": 43050
    },
    {
      "epoch": 129.31,
      "grad_norm": 10.366852760314941,
      "learning_rate": 5.68968968968969e-06,
      "loss": 0.2329,
      "step": 43060
    },
    {
      "epoch": 129.34,
      "grad_norm": 18.277088165283203,
      "learning_rate": 5.688688688688689e-06,
      "loss": 0.2569,
      "step": 43070
    },
    {
      "epoch": 129.37,
      "grad_norm": 7.512056827545166,
      "learning_rate": 5.687687687687689e-06,
      "loss": 0.2293,
      "step": 43080
    },
    {
      "epoch": 129.4,
      "grad_norm": 13.846866607666016,
      "learning_rate": 5.686686686686687e-06,
      "loss": 0.2382,
      "step": 43090
    },
    {
      "epoch": 129.43,
      "grad_norm": 11.275999069213867,
      "learning_rate": 5.685685685685686e-06,
      "loss": 0.2514,
      "step": 43100
    },
    {
      "epoch": 129.46,
      "grad_norm": 5.63031005859375,
      "learning_rate": 5.6846846846846846e-06,
      "loss": 0.253,
      "step": 43110
    },
    {
      "epoch": 129.49,
      "grad_norm": 8.575756072998047,
      "learning_rate": 5.683683683683684e-06,
      "loss": 0.2418,
      "step": 43120
    },
    {
      "epoch": 129.52,
      "grad_norm": 12.638221740722656,
      "learning_rate": 5.6826826826826835e-06,
      "loss": 0.2415,
      "step": 43130
    },
    {
      "epoch": 129.55,
      "grad_norm": 20.10300064086914,
      "learning_rate": 5.681681681681682e-06,
      "loss": 0.2663,
      "step": 43140
    },
    {
      "epoch": 129.58,
      "grad_norm": 6.696385383605957,
      "learning_rate": 5.6806806806806815e-06,
      "loss": 0.2462,
      "step": 43150
    },
    {
      "epoch": 129.61,
      "grad_norm": 7.696376323699951,
      "learning_rate": 5.67967967967968e-06,
      "loss": 0.2132,
      "step": 43160
    },
    {
      "epoch": 129.64,
      "grad_norm": 16.079687118530273,
      "learning_rate": 5.6786786786786795e-06,
      "loss": 0.2311,
      "step": 43170
    },
    {
      "epoch": 129.67,
      "grad_norm": 8.228824615478516,
      "learning_rate": 5.677677677677677e-06,
      "loss": 0.2626,
      "step": 43180
    },
    {
      "epoch": 129.7,
      "grad_norm": 8.054405212402344,
      "learning_rate": 5.676676676676677e-06,
      "loss": 0.2201,
      "step": 43190
    },
    {
      "epoch": 129.73,
      "grad_norm": 7.436009883880615,
      "learning_rate": 5.675675675675676e-06,
      "loss": 0.2569,
      "step": 43200
    },
    {
      "epoch": 129.76,
      "grad_norm": 9.121776580810547,
      "learning_rate": 5.674674674674675e-06,
      "loss": 0.261,
      "step": 43210
    },
    {
      "epoch": 129.79,
      "grad_norm": 55.9494743347168,
      "learning_rate": 5.673673673673674e-06,
      "loss": 0.2682,
      "step": 43220
    },
    {
      "epoch": 129.82,
      "grad_norm": 8.971183776855469,
      "learning_rate": 5.672672672672674e-06,
      "loss": 0.2283,
      "step": 43230
    },
    {
      "epoch": 129.85,
      "grad_norm": 9.586995124816895,
      "learning_rate": 5.671671671671672e-06,
      "loss": 0.2887,
      "step": 43240
    },
    {
      "epoch": 129.88,
      "grad_norm": 9.426704406738281,
      "learning_rate": 5.670670670670672e-06,
      "loss": 0.2729,
      "step": 43250
    },
    {
      "epoch": 129.91,
      "grad_norm": 6.200656890869141,
      "learning_rate": 5.6696696696696694e-06,
      "loss": 0.2486,
      "step": 43260
    },
    {
      "epoch": 129.94,
      "grad_norm": 7.423763275146484,
      "learning_rate": 5.668668668668669e-06,
      "loss": 0.2231,
      "step": 43270
    },
    {
      "epoch": 129.97,
      "grad_norm": 31.08656120300293,
      "learning_rate": 5.6676676676676675e-06,
      "loss": 0.2135,
      "step": 43280
    },
    {
      "epoch": 130.0,
      "grad_norm": 51.70720291137695,
      "learning_rate": 5.666666666666667e-06,
      "loss": 0.2739,
      "step": 43290
    },
    {
      "epoch": 130.0,
      "eval_accuracy": 0.9257,
      "eval_loss": 0.3166477084159851,
      "eval_runtime": 30.6279,
      "eval_samples_per_second": 326.5,
      "eval_steps_per_second": 1.306,
      "step": 43290
    },
    {
      "epoch": 130.03,
      "grad_norm": 14.185604095458984,
      "learning_rate": 5.665665665665666e-06,
      "loss": 0.2608,
      "step": 43300
    },
    {
      "epoch": 130.06,
      "grad_norm": 7.713788032531738,
      "learning_rate": 5.664664664664665e-06,
      "loss": 0.2624,
      "step": 43310
    },
    {
      "epoch": 130.09,
      "grad_norm": 6.608882427215576,
      "learning_rate": 5.663663663663664e-06,
      "loss": 0.2464,
      "step": 43320
    },
    {
      "epoch": 130.12,
      "grad_norm": 11.650530815124512,
      "learning_rate": 5.662662662662664e-06,
      "loss": 0.2204,
      "step": 43330
    },
    {
      "epoch": 130.15,
      "grad_norm": 8.378704071044922,
      "learning_rate": 5.661661661661662e-06,
      "loss": 0.2191,
      "step": 43340
    },
    {
      "epoch": 130.18,
      "grad_norm": 8.41157341003418,
      "learning_rate": 5.660660660660661e-06,
      "loss": 0.2064,
      "step": 43350
    },
    {
      "epoch": 130.21,
      "grad_norm": 5.540666580200195,
      "learning_rate": 5.65965965965966e-06,
      "loss": 0.227,
      "step": 43360
    },
    {
      "epoch": 130.24,
      "grad_norm": 8.475688934326172,
      "learning_rate": 5.658658658658659e-06,
      "loss": 0.2186,
      "step": 43370
    },
    {
      "epoch": 130.27,
      "grad_norm": 12.348843574523926,
      "learning_rate": 5.657657657657658e-06,
      "loss": 0.204,
      "step": 43380
    },
    {
      "epoch": 130.3,
      "grad_norm": 6.419233322143555,
      "learning_rate": 5.656656656656657e-06,
      "loss": 0.2692,
      "step": 43390
    },
    {
      "epoch": 130.33,
      "grad_norm": 9.321989059448242,
      "learning_rate": 5.6556556556556565e-06,
      "loss": 0.2798,
      "step": 43400
    },
    {
      "epoch": 130.36,
      "grad_norm": 7.808183670043945,
      "learning_rate": 5.654654654654655e-06,
      "loss": 0.2693,
      "step": 43410
    },
    {
      "epoch": 130.39,
      "grad_norm": 11.445246696472168,
      "learning_rate": 5.6536536536536546e-06,
      "loss": 0.2544,
      "step": 43420
    },
    {
      "epoch": 130.42,
      "grad_norm": 13.170942306518555,
      "learning_rate": 5.652652652652652e-06,
      "loss": 0.2175,
      "step": 43430
    },
    {
      "epoch": 130.45,
      "grad_norm": 11.774553298950195,
      "learning_rate": 5.651651651651652e-06,
      "loss": 0.2464,
      "step": 43440
    },
    {
      "epoch": 130.48,
      "grad_norm": 11.6845703125,
      "learning_rate": 5.650650650650651e-06,
      "loss": 0.2468,
      "step": 43450
    },
    {
      "epoch": 130.51,
      "grad_norm": 7.616000652313232,
      "learning_rate": 5.64964964964965e-06,
      "loss": 0.2976,
      "step": 43460
    },
    {
      "epoch": 130.54,
      "grad_norm": 7.6922807693481445,
      "learning_rate": 5.648648648648649e-06,
      "loss": 0.2543,
      "step": 43470
    },
    {
      "epoch": 130.57,
      "grad_norm": 11.096413612365723,
      "learning_rate": 5.647647647647649e-06,
      "loss": 0.2186,
      "step": 43480
    },
    {
      "epoch": 130.6,
      "grad_norm": 14.802783012390137,
      "learning_rate": 5.646646646646647e-06,
      "loss": 0.2715,
      "step": 43490
    },
    {
      "epoch": 130.63,
      "grad_norm": 5.72413969039917,
      "learning_rate": 5.645645645645647e-06,
      "loss": 0.2411,
      "step": 43500
    },
    {
      "epoch": 130.66,
      "grad_norm": 5.2050862312316895,
      "learning_rate": 5.6446446446446445e-06,
      "loss": 0.2333,
      "step": 43510
    },
    {
      "epoch": 130.69,
      "grad_norm": 13.707327842712402,
      "learning_rate": 5.643643643643644e-06,
      "loss": 0.302,
      "step": 43520
    },
    {
      "epoch": 130.72,
      "grad_norm": 8.018102645874023,
      "learning_rate": 5.6426426426426425e-06,
      "loss": 0.2498,
      "step": 43530
    },
    {
      "epoch": 130.75,
      "grad_norm": 7.006072044372559,
      "learning_rate": 5.641641641641642e-06,
      "loss": 0.2239,
      "step": 43540
    },
    {
      "epoch": 130.78,
      "grad_norm": 5.737383842468262,
      "learning_rate": 5.640640640640641e-06,
      "loss": 0.2182,
      "step": 43550
    },
    {
      "epoch": 130.81,
      "grad_norm": 8.991841316223145,
      "learning_rate": 5.63963963963964e-06,
      "loss": 0.2668,
      "step": 43560
    },
    {
      "epoch": 130.84,
      "grad_norm": 13.852638244628906,
      "learning_rate": 5.638638638638639e-06,
      "loss": 0.2519,
      "step": 43570
    },
    {
      "epoch": 130.87,
      "grad_norm": 6.3344316482543945,
      "learning_rate": 5.637637637637639e-06,
      "loss": 0.2342,
      "step": 43580
    },
    {
      "epoch": 130.9,
      "grad_norm": 8.236894607543945,
      "learning_rate": 5.6366366366366375e-06,
      "loss": 0.2953,
      "step": 43590
    },
    {
      "epoch": 130.93,
      "grad_norm": 5.9617486000061035,
      "learning_rate": 5.635635635635636e-06,
      "loss": 0.2512,
      "step": 43600
    },
    {
      "epoch": 130.96,
      "grad_norm": 12.033308029174805,
      "learning_rate": 5.634634634634635e-06,
      "loss": 0.2447,
      "step": 43610
    },
    {
      "epoch": 130.99,
      "grad_norm": 6.736783027648926,
      "learning_rate": 5.633633633633634e-06,
      "loss": 0.3104,
      "step": 43620
    },
    {
      "epoch": 131.0,
      "eval_accuracy": 0.9237,
      "eval_loss": 0.32382944226264954,
      "eval_runtime": 30.5324,
      "eval_samples_per_second": 327.521,
      "eval_steps_per_second": 1.31,
      "step": 43623
    },
    {
      "epoch": 131.02,
      "grad_norm": 16.332277297973633,
      "learning_rate": 5.632632632632633e-06,
      "loss": 0.2648,
      "step": 43630
    },
    {
      "epoch": 131.05,
      "grad_norm": 10.108380317687988,
      "learning_rate": 5.631631631631632e-06,
      "loss": 0.2722,
      "step": 43640
    },
    {
      "epoch": 131.08,
      "grad_norm": 3.6680498123168945,
      "learning_rate": 5.6306306306306316e-06,
      "loss": 0.2277,
      "step": 43650
    },
    {
      "epoch": 131.11,
      "grad_norm": 10.554811477661133,
      "learning_rate": 5.62962962962963e-06,
      "loss": 0.2273,
      "step": 43660
    },
    {
      "epoch": 131.14,
      "grad_norm": 15.333688735961914,
      "learning_rate": 5.62862862862863e-06,
      "loss": 0.2588,
      "step": 43670
    },
    {
      "epoch": 131.17,
      "grad_norm": 9.523752212524414,
      "learning_rate": 5.627627627627627e-06,
      "loss": 0.2267,
      "step": 43680
    },
    {
      "epoch": 131.2,
      "grad_norm": 11.516450881958008,
      "learning_rate": 5.626626626626627e-06,
      "loss": 0.2112,
      "step": 43690
    },
    {
      "epoch": 131.23,
      "grad_norm": 10.82993221282959,
      "learning_rate": 5.625625625625626e-06,
      "loss": 0.2391,
      "step": 43700
    },
    {
      "epoch": 131.26,
      "grad_norm": 6.517051696777344,
      "learning_rate": 5.624624624624625e-06,
      "loss": 0.3113,
      "step": 43710
    },
    {
      "epoch": 131.29,
      "grad_norm": 6.70371150970459,
      "learning_rate": 5.623623623623624e-06,
      "loss": 0.2264,
      "step": 43720
    },
    {
      "epoch": 131.32,
      "grad_norm": 13.239338874816895,
      "learning_rate": 5.622622622622623e-06,
      "loss": 0.261,
      "step": 43730
    },
    {
      "epoch": 131.35,
      "grad_norm": 8.029996871948242,
      "learning_rate": 5.621621621621622e-06,
      "loss": 0.2535,
      "step": 43740
    },
    {
      "epoch": 131.38,
      "grad_norm": 7.73216438293457,
      "learning_rate": 5.620620620620622e-06,
      "loss": 0.274,
      "step": 43750
    },
    {
      "epoch": 131.41,
      "grad_norm": 28.197452545166016,
      "learning_rate": 5.6196196196196195e-06,
      "loss": 0.2245,
      "step": 43760
    },
    {
      "epoch": 131.44,
      "grad_norm": 4.217005252838135,
      "learning_rate": 5.618618618618619e-06,
      "loss": 0.2287,
      "step": 43770
    },
    {
      "epoch": 131.47,
      "grad_norm": 3.936948537826538,
      "learning_rate": 5.6176176176176175e-06,
      "loss": 0.241,
      "step": 43780
    },
    {
      "epoch": 131.5,
      "grad_norm": 12.856735229492188,
      "learning_rate": 5.616616616616617e-06,
      "loss": 0.2545,
      "step": 43790
    },
    {
      "epoch": 131.53,
      "grad_norm": 2.9664061069488525,
      "learning_rate": 5.615615615615616e-06,
      "loss": 0.2041,
      "step": 43800
    },
    {
      "epoch": 131.56,
      "grad_norm": 12.546573638916016,
      "learning_rate": 5.614614614614615e-06,
      "loss": 0.2269,
      "step": 43810
    },
    {
      "epoch": 131.59,
      "grad_norm": 6.791029453277588,
      "learning_rate": 5.6136136136136145e-06,
      "loss": 0.2307,
      "step": 43820
    },
    {
      "epoch": 131.62,
      "grad_norm": 9.110831260681152,
      "learning_rate": 5.612612612612614e-06,
      "loss": 0.2215,
      "step": 43830
    },
    {
      "epoch": 131.65,
      "grad_norm": 9.960329055786133,
      "learning_rate": 5.6116116116116125e-06,
      "loss": 0.2885,
      "step": 43840
    },
    {
      "epoch": 131.68,
      "grad_norm": 11.46825885772705,
      "learning_rate": 5.61061061061061e-06,
      "loss": 0.2678,
      "step": 43850
    },
    {
      "epoch": 131.71,
      "grad_norm": 10.932429313659668,
      "learning_rate": 5.60960960960961e-06,
      "loss": 0.2789,
      "step": 43860
    },
    {
      "epoch": 131.74,
      "grad_norm": 7.486680507659912,
      "learning_rate": 5.608608608608609e-06,
      "loss": 0.2836,
      "step": 43870
    },
    {
      "epoch": 131.77,
      "grad_norm": 12.398481369018555,
      "learning_rate": 5.607607607607608e-06,
      "loss": 0.2404,
      "step": 43880
    },
    {
      "epoch": 131.8,
      "grad_norm": 8.975481986999512,
      "learning_rate": 5.606606606606607e-06,
      "loss": 0.2341,
      "step": 43890
    },
    {
      "epoch": 131.83,
      "grad_norm": 5.981738567352295,
      "learning_rate": 5.605605605605607e-06,
      "loss": 0.2113,
      "step": 43900
    },
    {
      "epoch": 131.86,
      "grad_norm": 6.918594837188721,
      "learning_rate": 5.604604604604605e-06,
      "loss": 0.2286,
      "step": 43910
    },
    {
      "epoch": 131.89,
      "grad_norm": 11.477506637573242,
      "learning_rate": 5.603603603603605e-06,
      "loss": 0.2594,
      "step": 43920
    },
    {
      "epoch": 131.92,
      "grad_norm": 9.140840530395508,
      "learning_rate": 5.602602602602602e-06,
      "loss": 0.2607,
      "step": 43930
    },
    {
      "epoch": 131.95,
      "grad_norm": 9.214582443237305,
      "learning_rate": 5.601601601601602e-06,
      "loss": 0.2855,
      "step": 43940
    },
    {
      "epoch": 131.98,
      "grad_norm": 10.120545387268066,
      "learning_rate": 5.600600600600601e-06,
      "loss": 0.264,
      "step": 43950
    },
    {
      "epoch": 132.0,
      "eval_accuracy": 0.9257,
      "eval_loss": 0.3163975477218628,
      "eval_runtime": 30.4835,
      "eval_samples_per_second": 328.046,
      "eval_steps_per_second": 1.312,
      "step": 43956
    },
    {
      "epoch": 132.01,
      "grad_norm": 8.14413833618164,
      "learning_rate": 5.5995995995996e-06,
      "loss": 0.2202,
      "step": 43960
    },
    {
      "epoch": 132.04,
      "grad_norm": 6.804251670837402,
      "learning_rate": 5.598598598598599e-06,
      "loss": 0.2151,
      "step": 43970
    },
    {
      "epoch": 132.07,
      "grad_norm": 19.682838439941406,
      "learning_rate": 5.597597597597598e-06,
      "loss": 0.246,
      "step": 43980
    },
    {
      "epoch": 132.1,
      "grad_norm": 9.842948913574219,
      "learning_rate": 5.596596596596597e-06,
      "loss": 0.2259,
      "step": 43990
    },
    {
      "epoch": 132.13,
      "grad_norm": 9.724662780761719,
      "learning_rate": 5.595595595595597e-06,
      "loss": 0.2493,
      "step": 44000
    },
    {
      "epoch": 132.16,
      "grad_norm": 9.480175018310547,
      "learning_rate": 5.5945945945945945e-06,
      "loss": 0.1878,
      "step": 44010
    },
    {
      "epoch": 132.19,
      "grad_norm": 9.458671569824219,
      "learning_rate": 5.593593593593594e-06,
      "loss": 0.2365,
      "step": 44020
    },
    {
      "epoch": 132.22,
      "grad_norm": 17.808467864990234,
      "learning_rate": 5.5925925925925926e-06,
      "loss": 0.2395,
      "step": 44030
    },
    {
      "epoch": 132.25,
      "grad_norm": 5.014270305633545,
      "learning_rate": 5.591591591591592e-06,
      "loss": 0.2567,
      "step": 44040
    },
    {
      "epoch": 132.28,
      "grad_norm": 9.704211235046387,
      "learning_rate": 5.5905905905905915e-06,
      "loss": 0.2619,
      "step": 44050
    },
    {
      "epoch": 132.31,
      "grad_norm": 7.440254211425781,
      "learning_rate": 5.58958958958959e-06,
      "loss": 0.2614,
      "step": 44060
    },
    {
      "epoch": 132.34,
      "grad_norm": 9.685340881347656,
      "learning_rate": 5.5885885885885895e-06,
      "loss": 0.2358,
      "step": 44070
    },
    {
      "epoch": 132.37,
      "grad_norm": 9.575346946716309,
      "learning_rate": 5.587587587587588e-06,
      "loss": 0.2148,
      "step": 44080
    },
    {
      "epoch": 132.4,
      "grad_norm": 6.136656761169434,
      "learning_rate": 5.5865865865865875e-06,
      "loss": 0.2377,
      "step": 44090
    },
    {
      "epoch": 132.43,
      "grad_norm": 8.793882369995117,
      "learning_rate": 5.585585585585585e-06,
      "loss": 0.2696,
      "step": 44100
    },
    {
      "epoch": 132.46,
      "grad_norm": 10.615421295166016,
      "learning_rate": 5.584584584584585e-06,
      "loss": 0.2726,
      "step": 44110
    },
    {
      "epoch": 132.49,
      "grad_norm": 8.622566223144531,
      "learning_rate": 5.583583583583584e-06,
      "loss": 0.2953,
      "step": 44120
    },
    {
      "epoch": 132.52,
      "grad_norm": 10.60853385925293,
      "learning_rate": 5.582582582582583e-06,
      "loss": 0.2654,
      "step": 44130
    },
    {
      "epoch": 132.55,
      "grad_norm": 10.511491775512695,
      "learning_rate": 5.581581581581582e-06,
      "loss": 0.2549,
      "step": 44140
    },
    {
      "epoch": 132.58,
      "grad_norm": 7.334187984466553,
      "learning_rate": 5.580580580580582e-06,
      "loss": 0.2509,
      "step": 44150
    },
    {
      "epoch": 132.61,
      "grad_norm": 10.446490287780762,
      "learning_rate": 5.57957957957958e-06,
      "loss": 0.2038,
      "step": 44160
    },
    {
      "epoch": 132.64,
      "grad_norm": 9.279346466064453,
      "learning_rate": 5.57857857857858e-06,
      "loss": 0.2603,
      "step": 44170
    },
    {
      "epoch": 132.67,
      "grad_norm": 8.275561332702637,
      "learning_rate": 5.577577577577577e-06,
      "loss": 0.2657,
      "step": 44180
    },
    {
      "epoch": 132.7,
      "grad_norm": 7.315482139587402,
      "learning_rate": 5.576576576576577e-06,
      "loss": 0.2974,
      "step": 44190
    },
    {
      "epoch": 132.73,
      "grad_norm": 7.413948059082031,
      "learning_rate": 5.5755755755755755e-06,
      "loss": 0.2369,
      "step": 44200
    },
    {
      "epoch": 132.76,
      "grad_norm": 7.87332010269165,
      "learning_rate": 5.574574574574575e-06,
      "loss": 0.2928,
      "step": 44210
    },
    {
      "epoch": 132.79,
      "grad_norm": 6.36109733581543,
      "learning_rate": 5.573573573573574e-06,
      "loss": 0.2725,
      "step": 44220
    },
    {
      "epoch": 132.82,
      "grad_norm": 8.793583869934082,
      "learning_rate": 5.572572572572573e-06,
      "loss": 0.2671,
      "step": 44230
    },
    {
      "epoch": 132.85,
      "grad_norm": 9.82668399810791,
      "learning_rate": 5.571571571571572e-06,
      "loss": 0.2423,
      "step": 44240
    },
    {
      "epoch": 132.88,
      "grad_norm": 9.607233047485352,
      "learning_rate": 5.570570570570572e-06,
      "loss": 0.2506,
      "step": 44250
    },
    {
      "epoch": 132.91,
      "grad_norm": 7.931602478027344,
      "learning_rate": 5.5695695695695696e-06,
      "loss": 0.2508,
      "step": 44260
    },
    {
      "epoch": 132.94,
      "grad_norm": 8.103853225708008,
      "learning_rate": 5.568568568568569e-06,
      "loss": 0.2852,
      "step": 44270
    },
    {
      "epoch": 132.97,
      "grad_norm": 7.635595321655273,
      "learning_rate": 5.567567567567568e-06,
      "loss": 0.2485,
      "step": 44280
    },
    {
      "epoch": 133.0,
      "eval_accuracy": 0.9232,
      "eval_loss": 0.3235946595668793,
      "eval_runtime": 30.513,
      "eval_samples_per_second": 327.729,
      "eval_steps_per_second": 1.311,
      "step": 44289
    },
    {
      "epoch": 133.0,
      "grad_norm": 8.922901153564453,
      "learning_rate": 5.566566566566567e-06,
      "loss": 0.2031,
      "step": 44290
    },
    {
      "epoch": 133.03,
      "grad_norm": 9.796833038330078,
      "learning_rate": 5.565565565565566e-06,
      "loss": 0.2412,
      "step": 44300
    },
    {
      "epoch": 133.06,
      "grad_norm": 7.982030868530273,
      "learning_rate": 5.564564564564565e-06,
      "loss": 0.2426,
      "step": 44310
    },
    {
      "epoch": 133.09,
      "grad_norm": 8.264145851135254,
      "learning_rate": 5.5635635635635645e-06,
      "loss": 0.2539,
      "step": 44320
    },
    {
      "epoch": 133.12,
      "grad_norm": 7.580409526824951,
      "learning_rate": 5.562562562562563e-06,
      "loss": 0.3008,
      "step": 44330
    },
    {
      "epoch": 133.15,
      "grad_norm": 12.055278778076172,
      "learning_rate": 5.561561561561562e-06,
      "loss": 0.2488,
      "step": 44340
    },
    {
      "epoch": 133.18,
      "grad_norm": 15.445891380310059,
      "learning_rate": 5.56056056056056e-06,
      "loss": 0.2095,
      "step": 44350
    },
    {
      "epoch": 133.21,
      "grad_norm": 7.011239051818848,
      "learning_rate": 5.55955955955956e-06,
      "loss": 0.2383,
      "step": 44360
    },
    {
      "epoch": 133.24,
      "grad_norm": 12.266694068908691,
      "learning_rate": 5.558558558558559e-06,
      "loss": 0.2225,
      "step": 44370
    },
    {
      "epoch": 133.27,
      "grad_norm": 6.499008655548096,
      "learning_rate": 5.557557557557558e-06,
      "loss": 0.2559,
      "step": 44380
    },
    {
      "epoch": 133.3,
      "grad_norm": 8.397539138793945,
      "learning_rate": 5.556556556556557e-06,
      "loss": 0.2393,
      "step": 44390
    },
    {
      "epoch": 133.33,
      "grad_norm": 8.487497329711914,
      "learning_rate": 5.555555555555557e-06,
      "loss": 0.2243,
      "step": 44400
    },
    {
      "epoch": 133.36,
      "grad_norm": 9.089244842529297,
      "learning_rate": 5.554554554554555e-06,
      "loss": 0.2497,
      "step": 44410
    },
    {
      "epoch": 133.39,
      "grad_norm": 12.40311050415039,
      "learning_rate": 5.553553553553555e-06,
      "loss": 0.2602,
      "step": 44420
    },
    {
      "epoch": 133.42,
      "grad_norm": 9.551292419433594,
      "learning_rate": 5.5525525525525525e-06,
      "loss": 0.2538,
      "step": 44430
    },
    {
      "epoch": 133.45,
      "grad_norm": 12.145597457885742,
      "learning_rate": 5.551551551551552e-06,
      "loss": 0.2401,
      "step": 44440
    },
    {
      "epoch": 133.48,
      "grad_norm": 7.237033843994141,
      "learning_rate": 5.5505505505505505e-06,
      "loss": 0.18,
      "step": 44450
    },
    {
      "epoch": 133.51,
      "grad_norm": 18.890939712524414,
      "learning_rate": 5.54954954954955e-06,
      "loss": 0.2696,
      "step": 44460
    },
    {
      "epoch": 133.54,
      "grad_norm": 10.421886444091797,
      "learning_rate": 5.548548548548549e-06,
      "loss": 0.2991,
      "step": 44470
    },
    {
      "epoch": 133.57,
      "grad_norm": 11.946776390075684,
      "learning_rate": 5.547547547547548e-06,
      "loss": 0.2412,
      "step": 44480
    },
    {
      "epoch": 133.6,
      "grad_norm": 8.986438751220703,
      "learning_rate": 5.546546546546547e-06,
      "loss": 0.2255,
      "step": 44490
    },
    {
      "epoch": 133.63,
      "grad_norm": 10.821784973144531,
      "learning_rate": 5.545545545545547e-06,
      "loss": 0.2428,
      "step": 44500
    },
    {
      "epoch": 133.66,
      "grad_norm": 31.132150650024414,
      "learning_rate": 5.544544544544545e-06,
      "loss": 0.2277,
      "step": 44510
    },
    {
      "epoch": 133.69,
      "grad_norm": 26.001802444458008,
      "learning_rate": 5.543543543543544e-06,
      "loss": 0.1946,
      "step": 44520
    },
    {
      "epoch": 133.72,
      "grad_norm": 9.066503524780273,
      "learning_rate": 5.542542542542543e-06,
      "loss": 0.2283,
      "step": 44530
    },
    {
      "epoch": 133.75,
      "grad_norm": 20.735992431640625,
      "learning_rate": 5.541541541541542e-06,
      "loss": 0.2326,
      "step": 44540
    },
    {
      "epoch": 133.78,
      "grad_norm": 6.0079193115234375,
      "learning_rate": 5.540540540540541e-06,
      "loss": 0.2416,
      "step": 44550
    },
    {
      "epoch": 133.81,
      "grad_norm": 10.404984474182129,
      "learning_rate": 5.53953953953954e-06,
      "loss": 0.2947,
      "step": 44560
    },
    {
      "epoch": 133.84,
      "grad_norm": 7.288845539093018,
      "learning_rate": 5.5385385385385396e-06,
      "loss": 0.198,
      "step": 44570
    },
    {
      "epoch": 133.87,
      "grad_norm": 8.693832397460938,
      "learning_rate": 5.537537537537538e-06,
      "loss": 0.2726,
      "step": 44580
    },
    {
      "epoch": 133.9,
      "grad_norm": 9.460439682006836,
      "learning_rate": 5.536536536536537e-06,
      "loss": 0.2485,
      "step": 44590
    },
    {
      "epoch": 133.93,
      "grad_norm": 7.853692531585693,
      "learning_rate": 5.535535535535535e-06,
      "loss": 0.2206,
      "step": 44600
    },
    {
      "epoch": 133.96,
      "grad_norm": 9.05345344543457,
      "learning_rate": 5.534534534534535e-06,
      "loss": 0.2473,
      "step": 44610
    },
    {
      "epoch": 133.99,
      "grad_norm": 15.882962226867676,
      "learning_rate": 5.533533533533534e-06,
      "loss": 0.2637,
      "step": 44620
    },
    {
      "epoch": 134.0,
      "eval_accuracy": 0.9249,
      "eval_loss": 0.3232244551181793,
      "eval_runtime": 30.355,
      "eval_samples_per_second": 329.435,
      "eval_steps_per_second": 1.318,
      "step": 44622
    },
    {
      "epoch": 134.02,
      "grad_norm": 16.30777359008789,
      "learning_rate": 5.532532532532533e-06,
      "loss": 0.3783,
      "step": 44630
    },
    {
      "epoch": 134.05,
      "grad_norm": 8.987543106079102,
      "learning_rate": 5.531531531531532e-06,
      "loss": 0.2561,
      "step": 44640
    },
    {
      "epoch": 134.08,
      "grad_norm": 6.125888347625732,
      "learning_rate": 5.530530530530531e-06,
      "loss": 0.2336,
      "step": 44650
    },
    {
      "epoch": 134.11,
      "grad_norm": 9.027758598327637,
      "learning_rate": 5.52952952952953e-06,
      "loss": 0.2581,
      "step": 44660
    },
    {
      "epoch": 134.14,
      "grad_norm": 22.544755935668945,
      "learning_rate": 5.52852852852853e-06,
      "loss": 0.2314,
      "step": 44670
    },
    {
      "epoch": 134.17,
      "grad_norm": 46.19795608520508,
      "learning_rate": 5.5275275275275275e-06,
      "loss": 0.2305,
      "step": 44680
    },
    {
      "epoch": 134.2,
      "grad_norm": 8.20695686340332,
      "learning_rate": 5.526526526526527e-06,
      "loss": 0.2297,
      "step": 44690
    },
    {
      "epoch": 134.23,
      "grad_norm": 8.515909194946289,
      "learning_rate": 5.5255255255255255e-06,
      "loss": 0.2225,
      "step": 44700
    },
    {
      "epoch": 134.26,
      "grad_norm": 11.009695053100586,
      "learning_rate": 5.524524524524525e-06,
      "loss": 0.2796,
      "step": 44710
    },
    {
      "epoch": 134.29,
      "grad_norm": 13.820937156677246,
      "learning_rate": 5.523523523523524e-06,
      "loss": 0.2094,
      "step": 44720
    },
    {
      "epoch": 134.32,
      "grad_norm": 16.41875648498535,
      "learning_rate": 5.522522522522523e-06,
      "loss": 0.2656,
      "step": 44730
    },
    {
      "epoch": 134.35,
      "grad_norm": 2.8153207302093506,
      "learning_rate": 5.5215215215215224e-06,
      "loss": 0.2297,
      "step": 44740
    },
    {
      "epoch": 134.38,
      "grad_norm": 11.189840316772461,
      "learning_rate": 5.520520520520521e-06,
      "loss": 0.2478,
      "step": 44750
    },
    {
      "epoch": 134.41,
      "grad_norm": 6.737630367279053,
      "learning_rate": 5.51951951951952e-06,
      "loss": 0.2009,
      "step": 44760
    },
    {
      "epoch": 134.44,
      "grad_norm": 15.346623420715332,
      "learning_rate": 5.518518518518518e-06,
      "loss": 0.2407,
      "step": 44770
    },
    {
      "epoch": 134.47,
      "grad_norm": 6.877048015594482,
      "learning_rate": 5.517517517517518e-06,
      "loss": 0.2206,
      "step": 44780
    },
    {
      "epoch": 134.5,
      "grad_norm": 7.614386081695557,
      "learning_rate": 5.516516516516517e-06,
      "loss": 0.3129,
      "step": 44790
    },
    {
      "epoch": 134.53,
      "grad_norm": 8.978729248046875,
      "learning_rate": 5.515515515515516e-06,
      "loss": 0.282,
      "step": 44800
    },
    {
      "epoch": 134.56,
      "grad_norm": 5.372732639312744,
      "learning_rate": 5.514514514514515e-06,
      "loss": 0.2676,
      "step": 44810
    },
    {
      "epoch": 134.59,
      "grad_norm": 4.174554347991943,
      "learning_rate": 5.513513513513515e-06,
      "loss": 0.2414,
      "step": 44820
    },
    {
      "epoch": 134.62,
      "grad_norm": 16.217344284057617,
      "learning_rate": 5.512512512512513e-06,
      "loss": 0.1985,
      "step": 44830
    },
    {
      "epoch": 134.65,
      "grad_norm": 9.987842559814453,
      "learning_rate": 5.511511511511512e-06,
      "loss": 0.2855,
      "step": 44840
    },
    {
      "epoch": 134.68,
      "grad_norm": 12.135832786560059,
      "learning_rate": 5.51051051051051e-06,
      "loss": 0.2753,
      "step": 44850
    },
    {
      "epoch": 134.71,
      "grad_norm": 4.635865211486816,
      "learning_rate": 5.50950950950951e-06,
      "loss": 0.2207,
      "step": 44860
    },
    {
      "epoch": 134.74,
      "grad_norm": 8.996200561523438,
      "learning_rate": 5.508508508508508e-06,
      "loss": 0.2621,
      "step": 44870
    },
    {
      "epoch": 134.77,
      "grad_norm": 9.730462074279785,
      "learning_rate": 5.507507507507508e-06,
      "loss": 0.231,
      "step": 44880
    },
    {
      "epoch": 134.8,
      "grad_norm": 10.786104202270508,
      "learning_rate": 5.506506506506507e-06,
      "loss": 0.2652,
      "step": 44890
    },
    {
      "epoch": 134.83,
      "grad_norm": 20.312292098999023,
      "learning_rate": 5.505505505505506e-06,
      "loss": 0.2958,
      "step": 44900
    },
    {
      "epoch": 134.86,
      "grad_norm": 15.948713302612305,
      "learning_rate": 5.504504504504505e-06,
      "loss": 0.2268,
      "step": 44910
    },
    {
      "epoch": 134.89,
      "grad_norm": 5.259818077087402,
      "learning_rate": 5.503503503503505e-06,
      "loss": 0.198,
      "step": 44920
    },
    {
      "epoch": 134.92,
      "grad_norm": 12.203185081481934,
      "learning_rate": 5.5025025025025025e-06,
      "loss": 0.2775,
      "step": 44930
    },
    {
      "epoch": 134.95,
      "grad_norm": 11.964247703552246,
      "learning_rate": 5.501501501501502e-06,
      "loss": 0.234,
      "step": 44940
    },
    {
      "epoch": 134.98,
      "grad_norm": 11.36174201965332,
      "learning_rate": 5.5005005005005006e-06,
      "loss": 0.2211,
      "step": 44950
    },
    {
      "epoch": 135.0,
      "eval_accuracy": 0.9256,
      "eval_loss": 0.3190864622592926,
      "eval_runtime": 30.2168,
      "eval_samples_per_second": 330.942,
      "eval_steps_per_second": 1.324,
      "step": 44955
    },
    {
      "epoch": 135.02,
      "grad_norm": 7.5355916023254395,
      "learning_rate": 5.4994994994995e-06,
      "loss": 0.285,
      "step": 44960
    },
    {
      "epoch": 135.05,
      "grad_norm": 7.511192798614502,
      "learning_rate": 5.4984984984984994e-06,
      "loss": 0.2546,
      "step": 44970
    },
    {
      "epoch": 135.08,
      "grad_norm": 25.871912002563477,
      "learning_rate": 5.497497497497498e-06,
      "loss": 0.2179,
      "step": 44980
    },
    {
      "epoch": 135.11,
      "grad_norm": 10.217187881469727,
      "learning_rate": 5.4964964964964975e-06,
      "loss": 0.2522,
      "step": 44990
    },
    {
      "epoch": 135.14,
      "grad_norm": 5.681333065032959,
      "learning_rate": 5.495495495495496e-06,
      "loss": 0.241,
      "step": 45000
    },
    {
      "epoch": 135.17,
      "grad_norm": 10.835148811340332,
      "learning_rate": 5.494494494494495e-06,
      "loss": 0.2616,
      "step": 45010
    },
    {
      "epoch": 135.2,
      "grad_norm": 7.708338737487793,
      "learning_rate": 5.493493493493493e-06,
      "loss": 0.2349,
      "step": 45020
    },
    {
      "epoch": 135.23,
      "grad_norm": 11.318270683288574,
      "learning_rate": 5.492492492492493e-06,
      "loss": 0.2409,
      "step": 45030
    },
    {
      "epoch": 135.26,
      "grad_norm": 12.514248847961426,
      "learning_rate": 5.491491491491492e-06,
      "loss": 0.2991,
      "step": 45040
    },
    {
      "epoch": 135.29,
      "grad_norm": 7.413797855377197,
      "learning_rate": 5.490490490490491e-06,
      "loss": 0.2458,
      "step": 45050
    },
    {
      "epoch": 135.32,
      "grad_norm": 37.649818420410156,
      "learning_rate": 5.48948948948949e-06,
      "loss": 0.2394,
      "step": 45060
    },
    {
      "epoch": 135.35,
      "grad_norm": 9.723489761352539,
      "learning_rate": 5.48848848848849e-06,
      "loss": 0.2458,
      "step": 45070
    },
    {
      "epoch": 135.38,
      "grad_norm": 11.0021333694458,
      "learning_rate": 5.487487487487488e-06,
      "loss": 0.2129,
      "step": 45080
    },
    {
      "epoch": 135.41,
      "grad_norm": 9.306757926940918,
      "learning_rate": 5.486486486486487e-06,
      "loss": 0.2361,
      "step": 45090
    },
    {
      "epoch": 135.44,
      "grad_norm": 10.364835739135742,
      "learning_rate": 5.485485485485485e-06,
      "loss": 0.2401,
      "step": 45100
    },
    {
      "epoch": 135.47,
      "grad_norm": 10.803581237792969,
      "learning_rate": 5.484484484484485e-06,
      "loss": 0.2494,
      "step": 45110
    },
    {
      "epoch": 135.5,
      "grad_norm": 12.841328620910645,
      "learning_rate": 5.4834834834834834e-06,
      "loss": 0.2011,
      "step": 45120
    },
    {
      "epoch": 135.53,
      "grad_norm": 8.190869331359863,
      "learning_rate": 5.482482482482483e-06,
      "loss": 0.2697,
      "step": 45130
    },
    {
      "epoch": 135.56,
      "grad_norm": 10.402886390686035,
      "learning_rate": 5.481481481481482e-06,
      "loss": 0.2488,
      "step": 45140
    },
    {
      "epoch": 135.59,
      "grad_norm": 10.352134704589844,
      "learning_rate": 5.480480480480481e-06,
      "loss": 0.2233,
      "step": 45150
    },
    {
      "epoch": 135.62,
      "grad_norm": 8.505707740783691,
      "learning_rate": 5.47947947947948e-06,
      "loss": 0.2033,
      "step": 45160
    },
    {
      "epoch": 135.65,
      "grad_norm": 11.534820556640625,
      "learning_rate": 5.47847847847848e-06,
      "loss": 0.2557,
      "step": 45170
    },
    {
      "epoch": 135.68,
      "grad_norm": 6.031650543212891,
      "learning_rate": 5.4774774774774776e-06,
      "loss": 0.2237,
      "step": 45180
    },
    {
      "epoch": 135.71,
      "grad_norm": 6.637165546417236,
      "learning_rate": 5.476476476476477e-06,
      "loss": 0.2112,
      "step": 45190
    },
    {
      "epoch": 135.74,
      "grad_norm": 8.140636444091797,
      "learning_rate": 5.475475475475476e-06,
      "loss": 0.2197,
      "step": 45200
    },
    {
      "epoch": 135.77,
      "grad_norm": 6.4671406745910645,
      "learning_rate": 5.474474474474475e-06,
      "loss": 0.2243,
      "step": 45210
    },
    {
      "epoch": 135.8,
      "grad_norm": 8.851451873779297,
      "learning_rate": 5.473473473473474e-06,
      "loss": 0.279,
      "step": 45220
    },
    {
      "epoch": 135.83,
      "grad_norm": 6.162501335144043,
      "learning_rate": 5.472472472472473e-06,
      "loss": 0.2207,
      "step": 45230
    },
    {
      "epoch": 135.86,
      "grad_norm": 8.416167259216309,
      "learning_rate": 5.4714714714714725e-06,
      "loss": 0.2089,
      "step": 45240
    },
    {
      "epoch": 135.89,
      "grad_norm": 17.90583610534668,
      "learning_rate": 5.470470470470471e-06,
      "loss": 0.245,
      "step": 45250
    },
    {
      "epoch": 135.92,
      "grad_norm": 7.159897804260254,
      "learning_rate": 5.46946946946947e-06,
      "loss": 0.2474,
      "step": 45260
    },
    {
      "epoch": 135.95,
      "grad_norm": 4.7758684158325195,
      "learning_rate": 5.468468468468468e-06,
      "loss": 0.2366,
      "step": 45270
    },
    {
      "epoch": 135.98,
      "grad_norm": 9.069210052490234,
      "learning_rate": 5.467467467467468e-06,
      "loss": 0.2498,
      "step": 45280
    },
    {
      "epoch": 136.0,
      "eval_accuracy": 0.9251,
      "eval_loss": 0.3189968168735504,
      "eval_runtime": 30.5321,
      "eval_samples_per_second": 327.524,
      "eval_steps_per_second": 1.31,
      "step": 45288
    },
    {
      "epoch": 136.01,
      "grad_norm": 8.290960311889648,
      "learning_rate": 5.466466466466467e-06,
      "loss": 0.2318,
      "step": 45290
    },
    {
      "epoch": 136.04,
      "grad_norm": 7.4315900802612305,
      "learning_rate": 5.465465465465466e-06,
      "loss": 0.257,
      "step": 45300
    },
    {
      "epoch": 136.07,
      "grad_norm": 8.918089866638184,
      "learning_rate": 5.464464464464465e-06,
      "loss": 0.2785,
      "step": 45310
    },
    {
      "epoch": 136.1,
      "grad_norm": 5.771815299987793,
      "learning_rate": 5.463463463463464e-06,
      "loss": 0.2404,
      "step": 45320
    },
    {
      "epoch": 136.13,
      "grad_norm": 7.275793075561523,
      "learning_rate": 5.462462462462463e-06,
      "loss": 0.277,
      "step": 45330
    },
    {
      "epoch": 136.16,
      "grad_norm": 5.079878330230713,
      "learning_rate": 5.461461461461461e-06,
      "loss": 0.2499,
      "step": 45340
    },
    {
      "epoch": 136.19,
      "grad_norm": 17.487592697143555,
      "learning_rate": 5.4604604604604604e-06,
      "loss": 0.2576,
      "step": 45350
    },
    {
      "epoch": 136.22,
      "grad_norm": 52.99030303955078,
      "learning_rate": 5.45945945945946e-06,
      "loss": 0.2707,
      "step": 45360
    },
    {
      "epoch": 136.25,
      "grad_norm": 6.085946083068848,
      "learning_rate": 5.4584584584584585e-06,
      "loss": 0.3019,
      "step": 45370
    },
    {
      "epoch": 136.28,
      "grad_norm": 10.987302780151367,
      "learning_rate": 5.457457457457458e-06,
      "loss": 0.2529,
      "step": 45380
    },
    {
      "epoch": 136.31,
      "grad_norm": 14.07805061340332,
      "learning_rate": 5.456456456456457e-06,
      "loss": 0.2176,
      "step": 45390
    },
    {
      "epoch": 136.34,
      "grad_norm": 9.089659690856934,
      "learning_rate": 5.455455455455456e-06,
      "loss": 0.2578,
      "step": 45400
    },
    {
      "epoch": 136.37,
      "grad_norm": 10.241616249084473,
      "learning_rate": 5.454454454454455e-06,
      "loss": 0.2776,
      "step": 45410
    },
    {
      "epoch": 136.4,
      "grad_norm": 11.789583206176758,
      "learning_rate": 5.453453453453455e-06,
      "loss": 0.298,
      "step": 45420
    },
    {
      "epoch": 136.43,
      "grad_norm": 7.2094902992248535,
      "learning_rate": 5.452452452452453e-06,
      "loss": 0.2817,
      "step": 45430
    },
    {
      "epoch": 136.46,
      "grad_norm": 8.58770751953125,
      "learning_rate": 5.451451451451451e-06,
      "loss": 0.2431,
      "step": 45440
    },
    {
      "epoch": 136.49,
      "grad_norm": 7.973793983459473,
      "learning_rate": 5.450450450450451e-06,
      "loss": 0.2256,
      "step": 45450
    },
    {
      "epoch": 136.52,
      "grad_norm": 8.360941886901855,
      "learning_rate": 5.44944944944945e-06,
      "loss": 0.2508,
      "step": 45460
    },
    {
      "epoch": 136.55,
      "grad_norm": 6.563233852386475,
      "learning_rate": 5.448448448448449e-06,
      "loss": 0.2118,
      "step": 45470
    },
    {
      "epoch": 136.58,
      "grad_norm": 10.416021347045898,
      "learning_rate": 5.447447447447448e-06,
      "loss": 0.2264,
      "step": 45480
    },
    {
      "epoch": 136.61,
      "grad_norm": 11.556928634643555,
      "learning_rate": 5.4464464464464475e-06,
      "loss": 0.2537,
      "step": 45490
    },
    {
      "epoch": 136.64,
      "grad_norm": 6.735903739929199,
      "learning_rate": 5.445445445445446e-06,
      "loss": 0.174,
      "step": 45500
    },
    {
      "epoch": 136.67,
      "grad_norm": 7.522817134857178,
      "learning_rate": 5.444444444444445e-06,
      "loss": 0.2194,
      "step": 45510
    },
    {
      "epoch": 136.7,
      "grad_norm": 15.136622428894043,
      "learning_rate": 5.443443443443443e-06,
      "loss": 0.2319,
      "step": 45520
    },
    {
      "epoch": 136.73,
      "grad_norm": 18.7313289642334,
      "learning_rate": 5.442442442442443e-06,
      "loss": 0.2291,
      "step": 45530
    },
    {
      "epoch": 136.76,
      "grad_norm": 8.771825790405273,
      "learning_rate": 5.441441441441442e-06,
      "loss": 0.2776,
      "step": 45540
    },
    {
      "epoch": 136.79,
      "grad_norm": 9.186455726623535,
      "learning_rate": 5.440440440440441e-06,
      "loss": 0.2395,
      "step": 45550
    },
    {
      "epoch": 136.82,
      "grad_norm": 9.30461597442627,
      "learning_rate": 5.43943943943944e-06,
      "loss": 0.2602,
      "step": 45560
    },
    {
      "epoch": 136.85,
      "grad_norm": 7.936672210693359,
      "learning_rate": 5.438438438438439e-06,
      "loss": 0.2607,
      "step": 45570
    },
    {
      "epoch": 136.88,
      "grad_norm": 12.845617294311523,
      "learning_rate": 5.437437437437438e-06,
      "loss": 0.2552,
      "step": 45580
    },
    {
      "epoch": 136.91,
      "grad_norm": 10.183942794799805,
      "learning_rate": 5.436436436436436e-06,
      "loss": 0.2536,
      "step": 45590
    },
    {
      "epoch": 136.94,
      "grad_norm": 8.346667289733887,
      "learning_rate": 5.4354354354354355e-06,
      "loss": 0.2648,
      "step": 45600
    },
    {
      "epoch": 136.97,
      "grad_norm": 7.783048629760742,
      "learning_rate": 5.434434434434435e-06,
      "loss": 0.2483,
      "step": 45610
    },
    {
      "epoch": 137.0,
      "grad_norm": 7.687425136566162,
      "learning_rate": 5.4334334334334335e-06,
      "loss": 0.2331,
      "step": 45620
    },
    {
      "epoch": 137.0,
      "eval_accuracy": 0.9245,
      "eval_loss": 0.32260504364967346,
      "eval_runtime": 29.9729,
      "eval_samples_per_second": 333.635,
      "eval_steps_per_second": 1.335,
      "step": 45621
    },
    {
      "epoch": 137.03,
      "grad_norm": 10.067251205444336,
      "learning_rate": 5.432432432432433e-06,
      "loss": 0.1944,
      "step": 45630
    },
    {
      "epoch": 137.06,
      "grad_norm": 10.504684448242188,
      "learning_rate": 5.431431431431432e-06,
      "loss": 0.2362,
      "step": 45640
    },
    {
      "epoch": 137.09,
      "grad_norm": 152.62623596191406,
      "learning_rate": 5.430430430430431e-06,
      "loss": 0.2767,
      "step": 45650
    },
    {
      "epoch": 137.12,
      "grad_norm": 10.280267715454102,
      "learning_rate": 5.4294294294294304e-06,
      "loss": 0.2488,
      "step": 45660
    },
    {
      "epoch": 137.15,
      "grad_norm": 7.609910011291504,
      "learning_rate": 5.428428428428428e-06,
      "loss": 0.2006,
      "step": 45670
    },
    {
      "epoch": 137.18,
      "grad_norm": 15.386307716369629,
      "learning_rate": 5.427427427427428e-06,
      "loss": 0.2698,
      "step": 45680
    },
    {
      "epoch": 137.21,
      "grad_norm": 6.285325050354004,
      "learning_rate": 5.426426426426426e-06,
      "loss": 0.203,
      "step": 45690
    },
    {
      "epoch": 137.24,
      "grad_norm": 7.101779460906982,
      "learning_rate": 5.425425425425426e-06,
      "loss": 0.2233,
      "step": 45700
    },
    {
      "epoch": 137.27,
      "grad_norm": 8.388566017150879,
      "learning_rate": 5.424424424424425e-06,
      "loss": 0.2273,
      "step": 45710
    },
    {
      "epoch": 137.3,
      "grad_norm": 15.108400344848633,
      "learning_rate": 5.423423423423424e-06,
      "loss": 0.2343,
      "step": 45720
    },
    {
      "epoch": 137.33,
      "grad_norm": 9.277636528015137,
      "learning_rate": 5.422422422422423e-06,
      "loss": 0.2071,
      "step": 45730
    },
    {
      "epoch": 137.36,
      "grad_norm": 31.78534698486328,
      "learning_rate": 5.421421421421423e-06,
      "loss": 0.268,
      "step": 45740
    },
    {
      "epoch": 137.39,
      "grad_norm": 11.533870697021484,
      "learning_rate": 5.420420420420421e-06,
      "loss": 0.2267,
      "step": 45750
    },
    {
      "epoch": 137.42,
      "grad_norm": 8.073833465576172,
      "learning_rate": 5.41941941941942e-06,
      "loss": 0.1819,
      "step": 45760
    },
    {
      "epoch": 137.45,
      "grad_norm": 12.339815139770508,
      "learning_rate": 5.418418418418418e-06,
      "loss": 0.2617,
      "step": 45770
    },
    {
      "epoch": 137.48,
      "grad_norm": 11.914682388305664,
      "learning_rate": 5.417417417417418e-06,
      "loss": 0.228,
      "step": 45780
    },
    {
      "epoch": 137.51,
      "grad_norm": 11.551218032836914,
      "learning_rate": 5.416416416416416e-06,
      "loss": 0.2238,
      "step": 45790
    },
    {
      "epoch": 137.54,
      "grad_norm": 6.966918468475342,
      "learning_rate": 5.415415415415416e-06,
      "loss": 0.2655,
      "step": 45800
    },
    {
      "epoch": 137.57,
      "grad_norm": 8.1865873336792,
      "learning_rate": 5.414414414414415e-06,
      "loss": 0.2483,
      "step": 45810
    },
    {
      "epoch": 137.6,
      "grad_norm": 10.602045059204102,
      "learning_rate": 5.413413413413414e-06,
      "loss": 0.2262,
      "step": 45820
    },
    {
      "epoch": 137.63,
      "grad_norm": 21.922863006591797,
      "learning_rate": 5.412412412412413e-06,
      "loss": 0.2615,
      "step": 45830
    },
    {
      "epoch": 137.66,
      "grad_norm": 9.755892753601074,
      "learning_rate": 5.411411411411411e-06,
      "loss": 0.248,
      "step": 45840
    },
    {
      "epoch": 137.69,
      "grad_norm": 22.35479164123535,
      "learning_rate": 5.4104104104104105e-06,
      "loss": 0.2087,
      "step": 45850
    },
    {
      "epoch": 137.72,
      "grad_norm": 16.8187198638916,
      "learning_rate": 5.40940940940941e-06,
      "loss": 0.2215,
      "step": 45860
    },
    {
      "epoch": 137.75,
      "grad_norm": 8.787782669067383,
      "learning_rate": 5.4084084084084086e-06,
      "loss": 0.2463,
      "step": 45870
    },
    {
      "epoch": 137.78,
      "grad_norm": 7.477694511413574,
      "learning_rate": 5.407407407407408e-06,
      "loss": 0.245,
      "step": 45880
    },
    {
      "epoch": 137.81,
      "grad_norm": 10.226480484008789,
      "learning_rate": 5.4064064064064074e-06,
      "loss": 0.1825,
      "step": 45890
    },
    {
      "epoch": 137.84,
      "grad_norm": 7.717316150665283,
      "learning_rate": 5.405405405405406e-06,
      "loss": 0.2053,
      "step": 45900
    },
    {
      "epoch": 137.87,
      "grad_norm": 13.233034133911133,
      "learning_rate": 5.4044044044044055e-06,
      "loss": 0.2574,
      "step": 45910
    },
    {
      "epoch": 137.9,
      "grad_norm": 7.806542873382568,
      "learning_rate": 5.403403403403403e-06,
      "loss": 0.2407,
      "step": 45920
    },
    {
      "epoch": 137.93,
      "grad_norm": 11.736202239990234,
      "learning_rate": 5.402402402402403e-06,
      "loss": 0.2662,
      "step": 45930
    },
    {
      "epoch": 137.96,
      "grad_norm": 25.964811325073242,
      "learning_rate": 5.401401401401401e-06,
      "loss": 0.2813,
      "step": 45940
    },
    {
      "epoch": 137.99,
      "grad_norm": 8.336063385009766,
      "learning_rate": 5.400400400400401e-06,
      "loss": 0.2247,
      "step": 45950
    },
    {
      "epoch": 138.0,
      "eval_accuracy": 0.9228,
      "eval_loss": 0.32409927248954773,
      "eval_runtime": 30.2848,
      "eval_samples_per_second": 330.199,
      "eval_steps_per_second": 1.321,
      "step": 45954
    },
    {
      "epoch": 138.02,
      "grad_norm": 11.210139274597168,
      "learning_rate": 5.3993993993994e-06,
      "loss": 0.2214,
      "step": 45960
    },
    {
      "epoch": 138.05,
      "grad_norm": 14.128129959106445,
      "learning_rate": 5.398398398398399e-06,
      "loss": 0.2685,
      "step": 45970
    },
    {
      "epoch": 138.08,
      "grad_norm": 7.492579936981201,
      "learning_rate": 5.397397397397398e-06,
      "loss": 0.25,
      "step": 45980
    },
    {
      "epoch": 138.11,
      "grad_norm": 9.643213272094727,
      "learning_rate": 5.396396396396398e-06,
      "loss": 0.2383,
      "step": 45990
    },
    {
      "epoch": 138.14,
      "grad_norm": 11.82809829711914,
      "learning_rate": 5.395395395395396e-06,
      "loss": 0.282,
      "step": 46000
    },
    {
      "epoch": 138.17,
      "grad_norm": 12.797986030578613,
      "learning_rate": 5.394394394394394e-06,
      "loss": 0.1953,
      "step": 46010
    },
    {
      "epoch": 138.2,
      "grad_norm": 11.942522048950195,
      "learning_rate": 5.393393393393393e-06,
      "loss": 0.2729,
      "step": 46020
    },
    {
      "epoch": 138.23,
      "grad_norm": 18.473020553588867,
      "learning_rate": 5.392392392392393e-06,
      "loss": 0.2236,
      "step": 46030
    },
    {
      "epoch": 138.26,
      "grad_norm": 6.587696552276611,
      "learning_rate": 5.3913913913913914e-06,
      "loss": 0.2473,
      "step": 46040
    },
    {
      "epoch": 138.29,
      "grad_norm": 8.512557029724121,
      "learning_rate": 5.390390390390391e-06,
      "loss": 0.2268,
      "step": 46050
    },
    {
      "epoch": 138.32,
      "grad_norm": 7.603056907653809,
      "learning_rate": 5.38938938938939e-06,
      "loss": 0.2016,
      "step": 46060
    },
    {
      "epoch": 138.35,
      "grad_norm": 8.400796890258789,
      "learning_rate": 5.388388388388389e-06,
      "loss": 0.2364,
      "step": 46070
    },
    {
      "epoch": 138.38,
      "grad_norm": 10.453791618347168,
      "learning_rate": 5.387387387387388e-06,
      "loss": 0.2452,
      "step": 46080
    },
    {
      "epoch": 138.41,
      "grad_norm": 8.776647567749023,
      "learning_rate": 5.386386386386386e-06,
      "loss": 0.2279,
      "step": 46090
    },
    {
      "epoch": 138.44,
      "grad_norm": 7.556663990020752,
      "learning_rate": 5.3853853853853856e-06,
      "loss": 0.272,
      "step": 46100
    },
    {
      "epoch": 138.47,
      "grad_norm": 8.310675621032715,
      "learning_rate": 5.384384384384385e-06,
      "loss": 0.2197,
      "step": 46110
    },
    {
      "epoch": 138.5,
      "grad_norm": 11.751174926757812,
      "learning_rate": 5.383383383383384e-06,
      "loss": 0.2992,
      "step": 46120
    },
    {
      "epoch": 138.53,
      "grad_norm": 5.595171928405762,
      "learning_rate": 5.382382382382383e-06,
      "loss": 0.2484,
      "step": 46130
    },
    {
      "epoch": 138.56,
      "grad_norm": 20.0321044921875,
      "learning_rate": 5.381381381381382e-06,
      "loss": 0.2426,
      "step": 46140
    },
    {
      "epoch": 138.59,
      "grad_norm": 8.295636177062988,
      "learning_rate": 5.380380380380381e-06,
      "loss": 0.2638,
      "step": 46150
    },
    {
      "epoch": 138.62,
      "grad_norm": 13.655635833740234,
      "learning_rate": 5.3793793793793805e-06,
      "loss": 0.1969,
      "step": 46160
    },
    {
      "epoch": 138.65,
      "grad_norm": 8.479827880859375,
      "learning_rate": 5.378378378378378e-06,
      "loss": 0.2377,
      "step": 46170
    },
    {
      "epoch": 138.68,
      "grad_norm": 6.864954948425293,
      "learning_rate": 5.377377377377378e-06,
      "loss": 0.2697,
      "step": 46180
    },
    {
      "epoch": 138.71,
      "grad_norm": 10.18932056427002,
      "learning_rate": 5.376376376376376e-06,
      "loss": 0.2303,
      "step": 46190
    },
    {
      "epoch": 138.74,
      "grad_norm": 7.930304050445557,
      "learning_rate": 5.375375375375376e-06,
      "loss": 0.2245,
      "step": 46200
    },
    {
      "epoch": 138.77,
      "grad_norm": 8.014949798583984,
      "learning_rate": 5.374374374374375e-06,
      "loss": 0.2682,
      "step": 46210
    },
    {
      "epoch": 138.8,
      "grad_norm": 6.8341474533081055,
      "learning_rate": 5.373373373373374e-06,
      "loss": 0.2452,
      "step": 46220
    },
    {
      "epoch": 138.83,
      "grad_norm": 12.058950424194336,
      "learning_rate": 5.372372372372373e-06,
      "loss": 0.237,
      "step": 46230
    },
    {
      "epoch": 138.86,
      "grad_norm": 9.720832824707031,
      "learning_rate": 5.371371371371372e-06,
      "loss": 0.2405,
      "step": 46240
    },
    {
      "epoch": 138.89,
      "grad_norm": 14.327349662780762,
      "learning_rate": 5.370370370370371e-06,
      "loss": 0.2566,
      "step": 46250
    },
    {
      "epoch": 138.92,
      "grad_norm": 6.611641883850098,
      "learning_rate": 5.369369369369369e-06,
      "loss": 0.2469,
      "step": 46260
    },
    {
      "epoch": 138.95,
      "grad_norm": 9.903032302856445,
      "learning_rate": 5.3683683683683684e-06,
      "loss": 0.25,
      "step": 46270
    },
    {
      "epoch": 138.98,
      "grad_norm": 6.986834526062012,
      "learning_rate": 5.367367367367368e-06,
      "loss": 0.2555,
      "step": 46280
    },
    {
      "epoch": 139.0,
      "eval_accuracy": 0.9224,
      "eval_loss": 0.32689929008483887,
      "eval_runtime": 30.3088,
      "eval_samples_per_second": 329.937,
      "eval_steps_per_second": 1.32,
      "step": 46287
    },
    {
      "epoch": 139.01,
      "grad_norm": 7.098176002502441,
      "learning_rate": 5.3663663663663665e-06,
      "loss": 0.2219,
      "step": 46290
    },
    {
      "epoch": 139.04,
      "grad_norm": 10.282215118408203,
      "learning_rate": 5.365365365365366e-06,
      "loss": 0.2355,
      "step": 46300
    },
    {
      "epoch": 139.07,
      "grad_norm": 6.48627233505249,
      "learning_rate": 5.364364364364365e-06,
      "loss": 0.2683,
      "step": 46310
    },
    {
      "epoch": 139.1,
      "grad_norm": 11.124475479125977,
      "learning_rate": 5.363363363363364e-06,
      "loss": 0.2519,
      "step": 46320
    },
    {
      "epoch": 139.13,
      "grad_norm": 11.010887145996094,
      "learning_rate": 5.362362362362363e-06,
      "loss": 0.2316,
      "step": 46330
    },
    {
      "epoch": 139.16,
      "grad_norm": 3.5160062313079834,
      "learning_rate": 5.361361361361361e-06,
      "loss": 0.2539,
      "step": 46340
    },
    {
      "epoch": 139.19,
      "grad_norm": 5.877514362335205,
      "learning_rate": 5.360360360360361e-06,
      "loss": 0.2257,
      "step": 46350
    },
    {
      "epoch": 139.22,
      "grad_norm": 10.49616527557373,
      "learning_rate": 5.359359359359359e-06,
      "loss": 0.2492,
      "step": 46360
    },
    {
      "epoch": 139.25,
      "grad_norm": 9.404972076416016,
      "learning_rate": 5.358358358358359e-06,
      "loss": 0.2428,
      "step": 46370
    },
    {
      "epoch": 139.28,
      "grad_norm": 6.3773298263549805,
      "learning_rate": 5.357357357357358e-06,
      "loss": 0.2423,
      "step": 46380
    },
    {
      "epoch": 139.31,
      "grad_norm": 11.587224006652832,
      "learning_rate": 5.356356356356357e-06,
      "loss": 0.2697,
      "step": 46390
    },
    {
      "epoch": 139.34,
      "grad_norm": 5.8942670822143555,
      "learning_rate": 5.355355355355356e-06,
      "loss": 0.2483,
      "step": 46400
    },
    {
      "epoch": 139.37,
      "grad_norm": 8.876152038574219,
      "learning_rate": 5.3543543543543555e-06,
      "loss": 0.2631,
      "step": 46410
    },
    {
      "epoch": 139.4,
      "grad_norm": 10.354926109313965,
      "learning_rate": 5.353353353353353e-06,
      "loss": 0.237,
      "step": 46420
    },
    {
      "epoch": 139.43,
      "grad_norm": 11.198017120361328,
      "learning_rate": 5.352352352352353e-06,
      "loss": 0.2231,
      "step": 46430
    },
    {
      "epoch": 139.46,
      "grad_norm": 6.230679988861084,
      "learning_rate": 5.351351351351351e-06,
      "loss": 0.1904,
      "step": 46440
    },
    {
      "epoch": 139.49,
      "grad_norm": 15.67978572845459,
      "learning_rate": 5.350350350350351e-06,
      "loss": 0.2169,
      "step": 46450
    },
    {
      "epoch": 139.52,
      "grad_norm": 7.356552600860596,
      "learning_rate": 5.34934934934935e-06,
      "loss": 0.2503,
      "step": 46460
    },
    {
      "epoch": 139.55,
      "grad_norm": 8.558950424194336,
      "learning_rate": 5.348348348348349e-06,
      "loss": 0.2555,
      "step": 46470
    },
    {
      "epoch": 139.58,
      "grad_norm": 4.161464214324951,
      "learning_rate": 5.347347347347348e-06,
      "loss": 0.2488,
      "step": 46480
    },
    {
      "epoch": 139.61,
      "grad_norm": 9.714371681213379,
      "learning_rate": 5.346346346346347e-06,
      "loss": 0.2101,
      "step": 46490
    },
    {
      "epoch": 139.64,
      "grad_norm": 13.800326347351074,
      "learning_rate": 5.345345345345346e-06,
      "loss": 0.2468,
      "step": 46500
    },
    {
      "epoch": 139.67,
      "grad_norm": 8.63749885559082,
      "learning_rate": 5.344344344344344e-06,
      "loss": 0.2733,
      "step": 46510
    },
    {
      "epoch": 139.7,
      "grad_norm": 5.117772102355957,
      "learning_rate": 5.3433433433433435e-06,
      "loss": 0.2327,
      "step": 46520
    },
    {
      "epoch": 139.73,
      "grad_norm": 11.16925048828125,
      "learning_rate": 5.342342342342343e-06,
      "loss": 0.2593,
      "step": 46530
    },
    {
      "epoch": 139.76,
      "grad_norm": 8.51315689086914,
      "learning_rate": 5.3413413413413415e-06,
      "loss": 0.3038,
      "step": 46540
    },
    {
      "epoch": 139.79,
      "grad_norm": 12.505570411682129,
      "learning_rate": 5.340340340340341e-06,
      "loss": 0.2224,
      "step": 46550
    },
    {
      "epoch": 139.82,
      "grad_norm": 9.381546020507812,
      "learning_rate": 5.33933933933934e-06,
      "loss": 0.2804,
      "step": 46560
    },
    {
      "epoch": 139.85,
      "grad_norm": 6.569847106933594,
      "learning_rate": 5.338338338338339e-06,
      "loss": 0.1841,
      "step": 46570
    },
    {
      "epoch": 139.88,
      "grad_norm": 11.254072189331055,
      "learning_rate": 5.3373373373373384e-06,
      "loss": 0.2805,
      "step": 46580
    },
    {
      "epoch": 139.91,
      "grad_norm": 10.038164138793945,
      "learning_rate": 5.336336336336336e-06,
      "loss": 0.1884,
      "step": 46590
    },
    {
      "epoch": 139.94,
      "grad_norm": 6.995599269866943,
      "learning_rate": 5.335335335335336e-06,
      "loss": 0.2461,
      "step": 46600
    },
    {
      "epoch": 139.97,
      "grad_norm": 9.704051971435547,
      "learning_rate": 5.334334334334334e-06,
      "loss": 0.2355,
      "step": 46610
    },
    {
      "epoch": 140.0,
      "grad_norm": 5.8002238273620605,
      "learning_rate": 5.333333333333334e-06,
      "loss": 0.2255,
      "step": 46620
    },
    {
      "epoch": 140.0,
      "eval_accuracy": 0.9247,
      "eval_loss": 0.322856068611145,
      "eval_runtime": 29.9944,
      "eval_samples_per_second": 333.396,
      "eval_steps_per_second": 1.334,
      "step": 46620
    },
    {
      "epoch": 140.03,
      "grad_norm": 8.391783714294434,
      "learning_rate": 5.332332332332333e-06,
      "loss": 0.206,
      "step": 46630
    },
    {
      "epoch": 140.06,
      "grad_norm": 8.640477180480957,
      "learning_rate": 5.331331331331332e-06,
      "loss": 0.2609,
      "step": 46640
    },
    {
      "epoch": 140.09,
      "grad_norm": 8.709698677062988,
      "learning_rate": 5.330330330330331e-06,
      "loss": 0.2384,
      "step": 46650
    },
    {
      "epoch": 140.12,
      "grad_norm": 11.256769180297852,
      "learning_rate": 5.329329329329331e-06,
      "loss": 0.2521,
      "step": 46660
    },
    {
      "epoch": 140.15,
      "grad_norm": 7.220466613769531,
      "learning_rate": 5.328328328328328e-06,
      "loss": 0.2251,
      "step": 46670
    },
    {
      "epoch": 140.18,
      "grad_norm": 7.095966339111328,
      "learning_rate": 5.327327327327328e-06,
      "loss": 0.2446,
      "step": 46680
    },
    {
      "epoch": 140.21,
      "grad_norm": 5.221657752990723,
      "learning_rate": 5.326326326326326e-06,
      "loss": 0.272,
      "step": 46690
    },
    {
      "epoch": 140.24,
      "grad_norm": 8.691357612609863,
      "learning_rate": 5.325325325325326e-06,
      "loss": 0.2325,
      "step": 46700
    },
    {
      "epoch": 140.27,
      "grad_norm": 23.04254913330078,
      "learning_rate": 5.324324324324324e-06,
      "loss": 0.2541,
      "step": 46710
    },
    {
      "epoch": 140.3,
      "grad_norm": 5.981679439544678,
      "learning_rate": 5.323323323323324e-06,
      "loss": 0.2376,
      "step": 46720
    },
    {
      "epoch": 140.33,
      "grad_norm": 7.4382548332214355,
      "learning_rate": 5.322322322322323e-06,
      "loss": 0.2391,
      "step": 46730
    },
    {
      "epoch": 140.36,
      "grad_norm": 20.322782516479492,
      "learning_rate": 5.321321321321322e-06,
      "loss": 0.2195,
      "step": 46740
    },
    {
      "epoch": 140.39,
      "grad_norm": 14.029951095581055,
      "learning_rate": 5.320320320320321e-06,
      "loss": 0.2304,
      "step": 46750
    },
    {
      "epoch": 140.42,
      "grad_norm": 12.639704704284668,
      "learning_rate": 5.319319319319319e-06,
      "loss": 0.2059,
      "step": 46760
    },
    {
      "epoch": 140.45,
      "grad_norm": 12.205340385437012,
      "learning_rate": 5.3183183183183185e-06,
      "loss": 0.3036,
      "step": 46770
    },
    {
      "epoch": 140.48,
      "grad_norm": 7.953347206115723,
      "learning_rate": 5.317317317317318e-06,
      "loss": 0.2521,
      "step": 46780
    },
    {
      "epoch": 140.51,
      "grad_norm": 15.255284309387207,
      "learning_rate": 5.3163163163163165e-06,
      "loss": 0.2448,
      "step": 46790
    },
    {
      "epoch": 140.54,
      "grad_norm": 7.993159770965576,
      "learning_rate": 5.315315315315316e-06,
      "loss": 0.2272,
      "step": 46800
    },
    {
      "epoch": 140.57,
      "grad_norm": 9.021060943603516,
      "learning_rate": 5.314314314314315e-06,
      "loss": 0.2031,
      "step": 46810
    },
    {
      "epoch": 140.6,
      "grad_norm": 5.981189250946045,
      "learning_rate": 5.313313313313314e-06,
      "loss": 0.2425,
      "step": 46820
    },
    {
      "epoch": 140.63,
      "grad_norm": 5.416562557220459,
      "learning_rate": 5.3123123123123135e-06,
      "loss": 0.2201,
      "step": 46830
    },
    {
      "epoch": 140.66,
      "grad_norm": 6.0045952796936035,
      "learning_rate": 5.311311311311311e-06,
      "loss": 0.2248,
      "step": 46840
    },
    {
      "epoch": 140.69,
      "grad_norm": 10.377646446228027,
      "learning_rate": 5.310310310310311e-06,
      "loss": 0.2723,
      "step": 46850
    },
    {
      "epoch": 140.72,
      "grad_norm": 10.395584106445312,
      "learning_rate": 5.309309309309309e-06,
      "loss": 0.2419,
      "step": 46860
    },
    {
      "epoch": 140.75,
      "grad_norm": 13.102848052978516,
      "learning_rate": 5.308308308308309e-06,
      "loss": 0.2316,
      "step": 46870
    },
    {
      "epoch": 140.78,
      "grad_norm": 7.9880218505859375,
      "learning_rate": 5.307307307307308e-06,
      "loss": 0.2695,
      "step": 46880
    },
    {
      "epoch": 140.81,
      "grad_norm": 98.62823486328125,
      "learning_rate": 5.306306306306307e-06,
      "loss": 0.2625,
      "step": 46890
    },
    {
      "epoch": 140.84,
      "grad_norm": 5.536327362060547,
      "learning_rate": 5.305305305305306e-06,
      "loss": 0.2432,
      "step": 46900
    },
    {
      "epoch": 140.87,
      "grad_norm": 7.112703323364258,
      "learning_rate": 5.304304304304306e-06,
      "loss": 0.2297,
      "step": 46910
    },
    {
      "epoch": 140.9,
      "grad_norm": 9.480280876159668,
      "learning_rate": 5.303303303303303e-06,
      "loss": 0.2305,
      "step": 46920
    },
    {
      "epoch": 140.93,
      "grad_norm": 46.267269134521484,
      "learning_rate": 5.302302302302302e-06,
      "loss": 0.2021,
      "step": 46930
    },
    {
      "epoch": 140.96,
      "grad_norm": 12.784379005432129,
      "learning_rate": 5.301301301301301e-06,
      "loss": 0.2172,
      "step": 46940
    },
    {
      "epoch": 140.99,
      "grad_norm": 7.856723785400391,
      "learning_rate": 5.300300300300301e-06,
      "loss": 0.2909,
      "step": 46950
    },
    {
      "epoch": 141.0,
      "eval_accuracy": 0.9256,
      "eval_loss": 0.32122135162353516,
      "eval_runtime": 30.2084,
      "eval_samples_per_second": 331.034,
      "eval_steps_per_second": 1.324,
      "step": 46953
    },
    {
      "epoch": 141.02,
      "grad_norm": 11.68538761138916,
      "learning_rate": 5.2992992992992994e-06,
      "loss": 0.2679,
      "step": 46960
    },
    {
      "epoch": 141.05,
      "grad_norm": 6.334212303161621,
      "learning_rate": 5.298298298298299e-06,
      "loss": 0.2571,
      "step": 46970
    },
    {
      "epoch": 141.08,
      "grad_norm": 8.599886894226074,
      "learning_rate": 5.297297297297298e-06,
      "loss": 0.2538,
      "step": 46980
    },
    {
      "epoch": 141.11,
      "grad_norm": 7.783298969268799,
      "learning_rate": 5.296296296296297e-06,
      "loss": 0.2304,
      "step": 46990
    },
    {
      "epoch": 141.14,
      "grad_norm": 10.632462501525879,
      "learning_rate": 5.2952952952952955e-06,
      "loss": 0.2879,
      "step": 47000
    },
    {
      "epoch": 141.17,
      "grad_norm": 8.382453918457031,
      "learning_rate": 5.294294294294294e-06,
      "loss": 0.2217,
      "step": 47010
    },
    {
      "epoch": 141.2,
      "grad_norm": 5.767722129821777,
      "learning_rate": 5.2932932932932935e-06,
      "loss": 0.252,
      "step": 47020
    },
    {
      "epoch": 141.23,
      "grad_norm": 9.252038955688477,
      "learning_rate": 5.292292292292293e-06,
      "loss": 0.2266,
      "step": 47030
    },
    {
      "epoch": 141.26,
      "grad_norm": 13.684170722961426,
      "learning_rate": 5.291291291291292e-06,
      "loss": 0.2595,
      "step": 47040
    },
    {
      "epoch": 141.29,
      "grad_norm": 13.407173156738281,
      "learning_rate": 5.290290290290291e-06,
      "loss": 0.2185,
      "step": 47050
    },
    {
      "epoch": 141.32,
      "grad_norm": 68.52151489257812,
      "learning_rate": 5.28928928928929e-06,
      "loss": 0.1945,
      "step": 47060
    },
    {
      "epoch": 141.35,
      "grad_norm": 8.563043594360352,
      "learning_rate": 5.288288288288289e-06,
      "loss": 0.2477,
      "step": 47070
    },
    {
      "epoch": 141.38,
      "grad_norm": 8.71263313293457,
      "learning_rate": 5.2872872872872885e-06,
      "loss": 0.2762,
      "step": 47080
    },
    {
      "epoch": 141.41,
      "grad_norm": 10.0582275390625,
      "learning_rate": 5.286286286286286e-06,
      "loss": 0.2505,
      "step": 47090
    },
    {
      "epoch": 141.44,
      "grad_norm": 13.355328559875488,
      "learning_rate": 5.285285285285286e-06,
      "loss": 0.2744,
      "step": 47100
    },
    {
      "epoch": 141.47,
      "grad_norm": 7.227279186248779,
      "learning_rate": 5.284284284284284e-06,
      "loss": 0.2379,
      "step": 47110
    },
    {
      "epoch": 141.5,
      "grad_norm": 6.138449192047119,
      "learning_rate": 5.283283283283284e-06,
      "loss": 0.2364,
      "step": 47120
    },
    {
      "epoch": 141.53,
      "grad_norm": 6.36877965927124,
      "learning_rate": 5.282282282282283e-06,
      "loss": 0.2195,
      "step": 47130
    },
    {
      "epoch": 141.56,
      "grad_norm": 10.331047058105469,
      "learning_rate": 5.281281281281282e-06,
      "loss": 0.2094,
      "step": 47140
    },
    {
      "epoch": 141.59,
      "grad_norm": 4.286640644073486,
      "learning_rate": 5.280280280280281e-06,
      "loss": 0.2327,
      "step": 47150
    },
    {
      "epoch": 141.62,
      "grad_norm": 16.241350173950195,
      "learning_rate": 5.27927927927928e-06,
      "loss": 0.2521,
      "step": 47160
    },
    {
      "epoch": 141.65,
      "grad_norm": 7.600894927978516,
      "learning_rate": 5.278278278278278e-06,
      "loss": 0.1915,
      "step": 47170
    },
    {
      "epoch": 141.68,
      "grad_norm": 7.80196475982666,
      "learning_rate": 5.277277277277277e-06,
      "loss": 0.2605,
      "step": 47180
    },
    {
      "epoch": 141.71,
      "grad_norm": 16.618003845214844,
      "learning_rate": 5.2762762762762764e-06,
      "loss": 0.2256,
      "step": 47190
    },
    {
      "epoch": 141.74,
      "grad_norm": 6.237436771392822,
      "learning_rate": 5.275275275275276e-06,
      "loss": 0.2148,
      "step": 47200
    },
    {
      "epoch": 141.77,
      "grad_norm": 10.072383880615234,
      "learning_rate": 5.2742742742742745e-06,
      "loss": 0.2636,
      "step": 47210
    },
    {
      "epoch": 141.8,
      "grad_norm": 7.003973007202148,
      "learning_rate": 5.273273273273274e-06,
      "loss": 0.2435,
      "step": 47220
    },
    {
      "epoch": 141.83,
      "grad_norm": 7.360803604125977,
      "learning_rate": 5.272272272272273e-06,
      "loss": 0.2005,
      "step": 47230
    },
    {
      "epoch": 141.86,
      "grad_norm": 12.454647064208984,
      "learning_rate": 5.271271271271272e-06,
      "loss": 0.285,
      "step": 47240
    },
    {
      "epoch": 141.89,
      "grad_norm": 7.729889392852783,
      "learning_rate": 5.2702702702702705e-06,
      "loss": 0.1981,
      "step": 47250
    },
    {
      "epoch": 141.92,
      "grad_norm": 9.975906372070312,
      "learning_rate": 5.269269269269269e-06,
      "loss": 0.2064,
      "step": 47260
    },
    {
      "epoch": 141.95,
      "grad_norm": 13.165088653564453,
      "learning_rate": 5.268268268268269e-06,
      "loss": 0.2507,
      "step": 47270
    },
    {
      "epoch": 141.98,
      "grad_norm": 13.073992729187012,
      "learning_rate": 5.267267267267267e-06,
      "loss": 0.2902,
      "step": 47280
    },
    {
      "epoch": 142.0,
      "eval_accuracy": 0.9231,
      "eval_loss": 0.3215266466140747,
      "eval_runtime": 29.6603,
      "eval_samples_per_second": 337.151,
      "eval_steps_per_second": 1.349,
      "step": 47286
    },
    {
      "epoch": 142.01,
      "grad_norm": 8.8219633102417,
      "learning_rate": 5.266266266266267e-06,
      "loss": 0.213,
      "step": 47290
    },
    {
      "epoch": 142.04,
      "grad_norm": 6.248149871826172,
      "learning_rate": 5.265265265265266e-06,
      "loss": 0.2553,
      "step": 47300
    },
    {
      "epoch": 142.07,
      "grad_norm": 11.465243339538574,
      "learning_rate": 5.264264264264265e-06,
      "loss": 0.3168,
      "step": 47310
    },
    {
      "epoch": 142.1,
      "grad_norm": 7.49376106262207,
      "learning_rate": 5.263263263263264e-06,
      "loss": 0.2263,
      "step": 47320
    },
    {
      "epoch": 142.13,
      "grad_norm": 7.8180832862854,
      "learning_rate": 5.2622622622622635e-06,
      "loss": 0.2111,
      "step": 47330
    },
    {
      "epoch": 142.16,
      "grad_norm": 8.77082347869873,
      "learning_rate": 5.261261261261261e-06,
      "loss": 0.2347,
      "step": 47340
    },
    {
      "epoch": 142.19,
      "grad_norm": 7.617552757263184,
      "learning_rate": 5.260260260260261e-06,
      "loss": 0.2375,
      "step": 47350
    },
    {
      "epoch": 142.22,
      "grad_norm": 8.517900466918945,
      "learning_rate": 5.259259259259259e-06,
      "loss": 0.2424,
      "step": 47360
    },
    {
      "epoch": 142.25,
      "grad_norm": 11.907402038574219,
      "learning_rate": 5.258258258258259e-06,
      "loss": 0.211,
      "step": 47370
    },
    {
      "epoch": 142.28,
      "grad_norm": 17.105802536010742,
      "learning_rate": 5.257257257257257e-06,
      "loss": 0.1942,
      "step": 47380
    },
    {
      "epoch": 142.31,
      "grad_norm": 12.52717399597168,
      "learning_rate": 5.256256256256257e-06,
      "loss": 0.2307,
      "step": 47390
    },
    {
      "epoch": 142.34,
      "grad_norm": 23.57975196838379,
      "learning_rate": 5.255255255255256e-06,
      "loss": 0.2239,
      "step": 47400
    },
    {
      "epoch": 142.37,
      "grad_norm": 10.67141056060791,
      "learning_rate": 5.254254254254255e-06,
      "loss": 0.2548,
      "step": 47410
    },
    {
      "epoch": 142.4,
      "grad_norm": 10.104243278503418,
      "learning_rate": 5.2532532532532534e-06,
      "loss": 0.2567,
      "step": 47420
    },
    {
      "epoch": 142.43,
      "grad_norm": 8.960025787353516,
      "learning_rate": 5.252252252252252e-06,
      "loss": 0.2149,
      "step": 47430
    },
    {
      "epoch": 142.46,
      "grad_norm": 6.438316822052002,
      "learning_rate": 5.2512512512512515e-06,
      "loss": 0.2322,
      "step": 47440
    },
    {
      "epoch": 142.49,
      "grad_norm": 8.826530456542969,
      "learning_rate": 5.250250250250251e-06,
      "loss": 0.2472,
      "step": 47450
    },
    {
      "epoch": 142.52,
      "grad_norm": 20.218299865722656,
      "learning_rate": 5.2492492492492495e-06,
      "loss": 0.1671,
      "step": 47460
    },
    {
      "epoch": 142.55,
      "grad_norm": 6.032350063323975,
      "learning_rate": 5.248248248248249e-06,
      "loss": 0.2475,
      "step": 47470
    },
    {
      "epoch": 142.58,
      "grad_norm": 8.201412200927734,
      "learning_rate": 5.247247247247248e-06,
      "loss": 0.2341,
      "step": 47480
    },
    {
      "epoch": 142.61,
      "grad_norm": 10.273181915283203,
      "learning_rate": 5.246246246246247e-06,
      "loss": 0.2517,
      "step": 47490
    },
    {
      "epoch": 142.64,
      "grad_norm": 9.436448097229004,
      "learning_rate": 5.245245245245245e-06,
      "loss": 0.2624,
      "step": 47500
    },
    {
      "epoch": 142.67,
      "grad_norm": 11.081579208374023,
      "learning_rate": 5.244244244244244e-06,
      "loss": 0.2255,
      "step": 47510
    },
    {
      "epoch": 142.7,
      "grad_norm": 10.488553047180176,
      "learning_rate": 5.243243243243244e-06,
      "loss": 0.2518,
      "step": 47520
    },
    {
      "epoch": 142.73,
      "grad_norm": 8.229935646057129,
      "learning_rate": 5.242242242242242e-06,
      "loss": 0.2352,
      "step": 47530
    },
    {
      "epoch": 142.76,
      "grad_norm": 9.72096061706543,
      "learning_rate": 5.241241241241242e-06,
      "loss": 0.2096,
      "step": 47540
    },
    {
      "epoch": 142.79,
      "grad_norm": 8.15075397491455,
      "learning_rate": 5.240240240240241e-06,
      "loss": 0.2348,
      "step": 47550
    },
    {
      "epoch": 142.82,
      "grad_norm": 7.155173301696777,
      "learning_rate": 5.23923923923924e-06,
      "loss": 0.1743,
      "step": 47560
    },
    {
      "epoch": 142.85,
      "grad_norm": 7.295491695404053,
      "learning_rate": 5.238238238238239e-06,
      "loss": 0.2359,
      "step": 47570
    },
    {
      "epoch": 142.88,
      "grad_norm": 21.072111129760742,
      "learning_rate": 5.2372372372372386e-06,
      "loss": 0.2713,
      "step": 47580
    },
    {
      "epoch": 142.91,
      "grad_norm": 11.862898826599121,
      "learning_rate": 5.236236236236236e-06,
      "loss": 0.2422,
      "step": 47590
    },
    {
      "epoch": 142.94,
      "grad_norm": 12.222519874572754,
      "learning_rate": 5.235235235235236e-06,
      "loss": 0.2379,
      "step": 47600
    },
    {
      "epoch": 142.97,
      "grad_norm": 7.6510701179504395,
      "learning_rate": 5.234234234234234e-06,
      "loss": 0.2384,
      "step": 47610
    },
    {
      "epoch": 143.0,
      "eval_accuracy": 0.9233,
      "eval_loss": 0.32955697178840637,
      "eval_runtime": 30.0979,
      "eval_samples_per_second": 332.249,
      "eval_steps_per_second": 1.329,
      "step": 47619
    },
    {
      "epoch": 143.0,
      "grad_norm": 10.170333862304688,
      "learning_rate": 5.233233233233234e-06,
      "loss": 0.2359,
      "step": 47620
    },
    {
      "epoch": 143.03,
      "grad_norm": 5.511844158172607,
      "learning_rate": 5.232232232232232e-06,
      "loss": 0.2229,
      "step": 47630
    },
    {
      "epoch": 143.06,
      "grad_norm": 8.183457374572754,
      "learning_rate": 5.231231231231232e-06,
      "loss": 0.2562,
      "step": 47640
    },
    {
      "epoch": 143.09,
      "grad_norm": 8.251331329345703,
      "learning_rate": 5.230230230230231e-06,
      "loss": 0.2349,
      "step": 47650
    },
    {
      "epoch": 143.12,
      "grad_norm": 9.304648399353027,
      "learning_rate": 5.22922922922923e-06,
      "loss": 0.2682,
      "step": 47660
    },
    {
      "epoch": 143.15,
      "grad_norm": 6.972588062286377,
      "learning_rate": 5.2282282282282285e-06,
      "loss": 0.2176,
      "step": 47670
    },
    {
      "epoch": 143.18,
      "grad_norm": 10.083733558654785,
      "learning_rate": 5.227227227227227e-06,
      "loss": 0.1915,
      "step": 47680
    },
    {
      "epoch": 143.21,
      "grad_norm": 9.263602256774902,
      "learning_rate": 5.2262262262262265e-06,
      "loss": 0.2191,
      "step": 47690
    },
    {
      "epoch": 143.24,
      "grad_norm": 14.893209457397461,
      "learning_rate": 5.225225225225226e-06,
      "loss": 0.2042,
      "step": 47700
    },
    {
      "epoch": 143.27,
      "grad_norm": 9.135733604431152,
      "learning_rate": 5.2242242242242245e-06,
      "loss": 0.2452,
      "step": 47710
    },
    {
      "epoch": 143.3,
      "grad_norm": 6.861645221710205,
      "learning_rate": 5.223223223223224e-06,
      "loss": 0.2576,
      "step": 47720
    },
    {
      "epoch": 143.33,
      "grad_norm": 8.588603973388672,
      "learning_rate": 5.2222222222222226e-06,
      "loss": 0.2518,
      "step": 47730
    },
    {
      "epoch": 143.36,
      "grad_norm": 15.78881549835205,
      "learning_rate": 5.221221221221222e-06,
      "loss": 0.1877,
      "step": 47740
    },
    {
      "epoch": 143.39,
      "grad_norm": 12.773731231689453,
      "learning_rate": 5.22022022022022e-06,
      "loss": 0.2019,
      "step": 47750
    },
    {
      "epoch": 143.42,
      "grad_norm": 11.98815631866455,
      "learning_rate": 5.219219219219219e-06,
      "loss": 0.2201,
      "step": 47760
    },
    {
      "epoch": 143.45,
      "grad_norm": 10.011098861694336,
      "learning_rate": 5.218218218218219e-06,
      "loss": 0.2439,
      "step": 47770
    },
    {
      "epoch": 143.48,
      "grad_norm": 9.228232383728027,
      "learning_rate": 5.217217217217217e-06,
      "loss": 0.2321,
      "step": 47780
    },
    {
      "epoch": 143.51,
      "grad_norm": 8.998093605041504,
      "learning_rate": 5.216216216216217e-06,
      "loss": 0.2387,
      "step": 47790
    },
    {
      "epoch": 143.54,
      "grad_norm": 9.000285148620605,
      "learning_rate": 5.215215215215216e-06,
      "loss": 0.2218,
      "step": 47800
    },
    {
      "epoch": 143.57,
      "grad_norm": 10.928627014160156,
      "learning_rate": 5.214214214214215e-06,
      "loss": 0.2702,
      "step": 47810
    },
    {
      "epoch": 143.6,
      "grad_norm": 6.31014347076416,
      "learning_rate": 5.213213213213214e-06,
      "loss": 0.2768,
      "step": 47820
    },
    {
      "epoch": 143.63,
      "grad_norm": 7.614251136779785,
      "learning_rate": 5.212212212212214e-06,
      "loss": 0.2518,
      "step": 47830
    },
    {
      "epoch": 143.66,
      "grad_norm": 9.763751983642578,
      "learning_rate": 5.211211211211211e-06,
      "loss": 0.2253,
      "step": 47840
    },
    {
      "epoch": 143.69,
      "grad_norm": 21.81978988647461,
      "learning_rate": 5.21021021021021e-06,
      "loss": 0.228,
      "step": 47850
    },
    {
      "epoch": 143.72,
      "grad_norm": 6.009259223937988,
      "learning_rate": 5.209209209209209e-06,
      "loss": 0.1911,
      "step": 47860
    },
    {
      "epoch": 143.75,
      "grad_norm": 13.918183326721191,
      "learning_rate": 5.208208208208209e-06,
      "loss": 0.2368,
      "step": 47870
    },
    {
      "epoch": 143.78,
      "grad_norm": 7.170630931854248,
      "learning_rate": 5.2072072072072074e-06,
      "loss": 0.2208,
      "step": 47880
    },
    {
      "epoch": 143.81,
      "grad_norm": 11.45240306854248,
      "learning_rate": 5.206206206206207e-06,
      "loss": 0.2286,
      "step": 47890
    },
    {
      "epoch": 143.84,
      "grad_norm": 9.47824764251709,
      "learning_rate": 5.205205205205206e-06,
      "loss": 0.2504,
      "step": 47900
    },
    {
      "epoch": 143.87,
      "grad_norm": 8.820111274719238,
      "learning_rate": 5.204204204204205e-06,
      "loss": 0.2156,
      "step": 47910
    },
    {
      "epoch": 143.9,
      "grad_norm": 5.7362141609191895,
      "learning_rate": 5.2032032032032035e-06,
      "loss": 0.2087,
      "step": 47920
    },
    {
      "epoch": 143.93,
      "grad_norm": 7.854790687561035,
      "learning_rate": 5.202202202202202e-06,
      "loss": 0.2006,
      "step": 47930
    },
    {
      "epoch": 143.96,
      "grad_norm": 6.853731155395508,
      "learning_rate": 5.2012012012012015e-06,
      "loss": 0.2205,
      "step": 47940
    },
    {
      "epoch": 143.99,
      "grad_norm": 149.41139221191406,
      "learning_rate": 5.200200200200201e-06,
      "loss": 0.2538,
      "step": 47950
    },
    {
      "epoch": 144.0,
      "eval_accuracy": 0.9255,
      "eval_loss": 0.32695648074150085,
      "eval_runtime": 30.0921,
      "eval_samples_per_second": 332.313,
      "eval_steps_per_second": 1.329,
      "step": 47952
    },
    {
      "epoch": 144.02,
      "grad_norm": 5.806835174560547,
      "learning_rate": 5.1991991991991996e-06,
      "loss": 0.1927,
      "step": 47960
    },
    {
      "epoch": 144.05,
      "grad_norm": 9.17219352722168,
      "learning_rate": 5.198198198198199e-06,
      "loss": 0.2462,
      "step": 47970
    },
    {
      "epoch": 144.08,
      "grad_norm": 9.759930610656738,
      "learning_rate": 5.197197197197198e-06,
      "loss": 0.2097,
      "step": 47980
    },
    {
      "epoch": 144.11,
      "grad_norm": 10.345853805541992,
      "learning_rate": 5.196196196196197e-06,
      "loss": 0.3005,
      "step": 47990
    },
    {
      "epoch": 144.14,
      "grad_norm": 7.822539806365967,
      "learning_rate": 5.195195195195195e-06,
      "loss": 0.2648,
      "step": 48000
    },
    {
      "epoch": 144.17,
      "grad_norm": 22.131187438964844,
      "learning_rate": 5.194194194194194e-06,
      "loss": 0.2424,
      "step": 48010
    },
    {
      "epoch": 144.2,
      "grad_norm": 7.232821464538574,
      "learning_rate": 5.193193193193194e-06,
      "loss": 0.2238,
      "step": 48020
    },
    {
      "epoch": 144.23,
      "grad_norm": 18.432348251342773,
      "learning_rate": 5.192192192192192e-06,
      "loss": 0.2258,
      "step": 48030
    },
    {
      "epoch": 144.26,
      "grad_norm": 8.351350784301758,
      "learning_rate": 5.191191191191192e-06,
      "loss": 0.285,
      "step": 48040
    },
    {
      "epoch": 144.29,
      "grad_norm": 10.389461517333984,
      "learning_rate": 5.190190190190191e-06,
      "loss": 0.2359,
      "step": 48050
    },
    {
      "epoch": 144.32,
      "grad_norm": 13.135757446289062,
      "learning_rate": 5.18918918918919e-06,
      "loss": 0.213,
      "step": 48060
    },
    {
      "epoch": 144.35,
      "grad_norm": 10.767936706542969,
      "learning_rate": 5.188188188188189e-06,
      "loss": 0.2254,
      "step": 48070
    },
    {
      "epoch": 144.38,
      "grad_norm": 14.9078950881958,
      "learning_rate": 5.187187187187187e-06,
      "loss": 0.2024,
      "step": 48080
    },
    {
      "epoch": 144.41,
      "grad_norm": 6.879306316375732,
      "learning_rate": 5.186186186186186e-06,
      "loss": 0.2032,
      "step": 48090
    },
    {
      "epoch": 144.44,
      "grad_norm": 7.593876361846924,
      "learning_rate": 5.185185185185185e-06,
      "loss": 0.2277,
      "step": 48100
    },
    {
      "epoch": 144.47,
      "grad_norm": 11.30392074584961,
      "learning_rate": 5.1841841841841844e-06,
      "loss": 0.2672,
      "step": 48110
    },
    {
      "epoch": 144.5,
      "grad_norm": 7.801263809204102,
      "learning_rate": 5.183183183183184e-06,
      "loss": 0.2208,
      "step": 48120
    },
    {
      "epoch": 144.53,
      "grad_norm": 5.613268852233887,
      "learning_rate": 5.1821821821821825e-06,
      "loss": 0.2435,
      "step": 48130
    },
    {
      "epoch": 144.56,
      "grad_norm": 10.130819320678711,
      "learning_rate": 5.181181181181182e-06,
      "loss": 0.2836,
      "step": 48140
    },
    {
      "epoch": 144.59,
      "grad_norm": 8.544219017028809,
      "learning_rate": 5.180180180180181e-06,
      "loss": 0.2719,
      "step": 48150
    },
    {
      "epoch": 144.62,
      "grad_norm": 8.937192916870117,
      "learning_rate": 5.17917917917918e-06,
      "loss": 0.2136,
      "step": 48160
    },
    {
      "epoch": 144.65,
      "grad_norm": 10.559365272521973,
      "learning_rate": 5.1781781781781785e-06,
      "loss": 0.2607,
      "step": 48170
    },
    {
      "epoch": 144.68,
      "grad_norm": 10.737166404724121,
      "learning_rate": 5.177177177177177e-06,
      "loss": 0.2566,
      "step": 48180
    },
    {
      "epoch": 144.71,
      "grad_norm": 7.074259281158447,
      "learning_rate": 5.1761761761761766e-06,
      "loss": 0.2794,
      "step": 48190
    },
    {
      "epoch": 144.74,
      "grad_norm": 13.997941970825195,
      "learning_rate": 5.175175175175175e-06,
      "loss": 0.2572,
      "step": 48200
    },
    {
      "epoch": 144.77,
      "grad_norm": 13.471281051635742,
      "learning_rate": 5.174174174174175e-06,
      "loss": 0.2422,
      "step": 48210
    },
    {
      "epoch": 144.8,
      "grad_norm": 10.00949478149414,
      "learning_rate": 5.173173173173174e-06,
      "loss": 0.2437,
      "step": 48220
    },
    {
      "epoch": 144.83,
      "grad_norm": 8.105002403259277,
      "learning_rate": 5.172172172172173e-06,
      "loss": 0.2403,
      "step": 48230
    },
    {
      "epoch": 144.86,
      "grad_norm": 15.698277473449707,
      "learning_rate": 5.171171171171172e-06,
      "loss": 0.2431,
      "step": 48240
    },
    {
      "epoch": 144.89,
      "grad_norm": 5.500518798828125,
      "learning_rate": 5.17017017017017e-06,
      "loss": 0.2376,
      "step": 48250
    },
    {
      "epoch": 144.92,
      "grad_norm": 14.979453086853027,
      "learning_rate": 5.169169169169169e-06,
      "loss": 0.2332,
      "step": 48260
    },
    {
      "epoch": 144.95,
      "grad_norm": 9.583667755126953,
      "learning_rate": 5.168168168168169e-06,
      "loss": 0.1974,
      "step": 48270
    },
    {
      "epoch": 144.98,
      "grad_norm": 11.668166160583496,
      "learning_rate": 5.167167167167167e-06,
      "loss": 0.2174,
      "step": 48280
    },
    {
      "epoch": 145.0,
      "eval_accuracy": 0.9249,
      "eval_loss": 0.3264999985694885,
      "eval_runtime": 29.9389,
      "eval_samples_per_second": 334.014,
      "eval_steps_per_second": 1.336,
      "step": 48285
    },
    {
      "epoch": 145.02,
      "grad_norm": 11.185776710510254,
      "learning_rate": 5.166166166166167e-06,
      "loss": 0.2204,
      "step": 48290
    },
    {
      "epoch": 145.05,
      "grad_norm": 5.479007720947266,
      "learning_rate": 5.165165165165165e-06,
      "loss": 0.2355,
      "step": 48300
    },
    {
      "epoch": 145.08,
      "grad_norm": 9.456266403198242,
      "learning_rate": 5.164164164164165e-06,
      "loss": 0.2592,
      "step": 48310
    },
    {
      "epoch": 145.11,
      "grad_norm": 9.123007774353027,
      "learning_rate": 5.163163163163164e-06,
      "loss": 0.27,
      "step": 48320
    },
    {
      "epoch": 145.14,
      "grad_norm": 7.980206489562988,
      "learning_rate": 5.162162162162162e-06,
      "loss": 0.2672,
      "step": 48330
    },
    {
      "epoch": 145.17,
      "grad_norm": 10.95048713684082,
      "learning_rate": 5.161161161161161e-06,
      "loss": 0.2823,
      "step": 48340
    },
    {
      "epoch": 145.2,
      "grad_norm": 10.173186302185059,
      "learning_rate": 5.16016016016016e-06,
      "loss": 0.1886,
      "step": 48350
    },
    {
      "epoch": 145.23,
      "grad_norm": 9.958852767944336,
      "learning_rate": 5.1591591591591595e-06,
      "loss": 0.2489,
      "step": 48360
    },
    {
      "epoch": 145.26,
      "grad_norm": 7.454949378967285,
      "learning_rate": 5.158158158158159e-06,
      "loss": 0.2284,
      "step": 48370
    },
    {
      "epoch": 145.29,
      "grad_norm": 14.592534065246582,
      "learning_rate": 5.1571571571571575e-06,
      "loss": 0.2579,
      "step": 48380
    },
    {
      "epoch": 145.32,
      "grad_norm": 7.57045841217041,
      "learning_rate": 5.156156156156157e-06,
      "loss": 0.203,
      "step": 48390
    },
    {
      "epoch": 145.35,
      "grad_norm": 9.589801788330078,
      "learning_rate": 5.155155155155156e-06,
      "loss": 0.269,
      "step": 48400
    },
    {
      "epoch": 145.38,
      "grad_norm": 10.910606384277344,
      "learning_rate": 5.154154154154155e-06,
      "loss": 0.2605,
      "step": 48410
    },
    {
      "epoch": 145.41,
      "grad_norm": 6.10244607925415,
      "learning_rate": 5.153153153153153e-06,
      "loss": 0.2053,
      "step": 48420
    },
    {
      "epoch": 145.44,
      "grad_norm": 17.459083557128906,
      "learning_rate": 5.152152152152152e-06,
      "loss": 0.2166,
      "step": 48430
    },
    {
      "epoch": 145.47,
      "grad_norm": 8.135537147521973,
      "learning_rate": 5.151151151151152e-06,
      "loss": 0.2506,
      "step": 48440
    },
    {
      "epoch": 145.5,
      "grad_norm": 9.386465072631836,
      "learning_rate": 5.15015015015015e-06,
      "loss": 0.2507,
      "step": 48450
    },
    {
      "epoch": 145.53,
      "grad_norm": 10.162070274353027,
      "learning_rate": 5.14914914914915e-06,
      "loss": 0.221,
      "step": 48460
    },
    {
      "epoch": 145.56,
      "grad_norm": 8.40754222869873,
      "learning_rate": 5.148148148148149e-06,
      "loss": 0.2441,
      "step": 48470
    },
    {
      "epoch": 145.59,
      "grad_norm": 5.971411228179932,
      "learning_rate": 5.147147147147148e-06,
      "loss": 0.2386,
      "step": 48480
    },
    {
      "epoch": 145.62,
      "grad_norm": 16.369892120361328,
      "learning_rate": 5.146146146146147e-06,
      "loss": 0.2563,
      "step": 48490
    },
    {
      "epoch": 145.65,
      "grad_norm": 4.956251621246338,
      "learning_rate": 5.145145145145145e-06,
      "loss": 0.2406,
      "step": 48500
    },
    {
      "epoch": 145.68,
      "grad_norm": 10.247551918029785,
      "learning_rate": 5.144144144144144e-06,
      "loss": 0.2719,
      "step": 48510
    },
    {
      "epoch": 145.71,
      "grad_norm": 5.316470623016357,
      "learning_rate": 5.143143143143144e-06,
      "loss": 0.2751,
      "step": 48520
    },
    {
      "epoch": 145.74,
      "grad_norm": 10.846887588500977,
      "learning_rate": 5.142142142142142e-06,
      "loss": 0.259,
      "step": 48530
    },
    {
      "epoch": 145.77,
      "grad_norm": 8.434002876281738,
      "learning_rate": 5.141141141141142e-06,
      "loss": 0.2358,
      "step": 48540
    },
    {
      "epoch": 145.8,
      "grad_norm": 5.785188674926758,
      "learning_rate": 5.14014014014014e-06,
      "loss": 0.2265,
      "step": 48550
    },
    {
      "epoch": 145.83,
      "grad_norm": 14.330278396606445,
      "learning_rate": 5.13913913913914e-06,
      "loss": 0.2373,
      "step": 48560
    },
    {
      "epoch": 145.86,
      "grad_norm": 16.056798934936523,
      "learning_rate": 5.138138138138139e-06,
      "loss": 0.2302,
      "step": 48570
    },
    {
      "epoch": 145.89,
      "grad_norm": 14.734687805175781,
      "learning_rate": 5.137137137137137e-06,
      "loss": 0.2251,
      "step": 48580
    },
    {
      "epoch": 145.92,
      "grad_norm": 10.299134254455566,
      "learning_rate": 5.1361361361361365e-06,
      "loss": 0.2604,
      "step": 48590
    },
    {
      "epoch": 145.95,
      "grad_norm": 5.451448917388916,
      "learning_rate": 5.135135135135135e-06,
      "loss": 0.24,
      "step": 48600
    },
    {
      "epoch": 145.98,
      "grad_norm": 11.634796142578125,
      "learning_rate": 5.1341341341341345e-06,
      "loss": 0.2274,
      "step": 48610
    },
    {
      "epoch": 146.0,
      "eval_accuracy": 0.9235,
      "eval_loss": 0.33024531602859497,
      "eval_runtime": 30.162,
      "eval_samples_per_second": 331.544,
      "eval_steps_per_second": 1.326,
      "step": 48618
    },
    {
      "epoch": 146.01,
      "grad_norm": 7.050867557525635,
      "learning_rate": 5.133133133133134e-06,
      "loss": 0.177,
      "step": 48620
    },
    {
      "epoch": 146.04,
      "grad_norm": 10.219776153564453,
      "learning_rate": 5.1321321321321325e-06,
      "loss": 0.2764,
      "step": 48630
    },
    {
      "epoch": 146.07,
      "grad_norm": 43.757442474365234,
      "learning_rate": 5.131131131131132e-06,
      "loss": 0.272,
      "step": 48640
    },
    {
      "epoch": 146.1,
      "grad_norm": 7.640416145324707,
      "learning_rate": 5.1301301301301306e-06,
      "loss": 0.2286,
      "step": 48650
    },
    {
      "epoch": 146.13,
      "grad_norm": 13.768400192260742,
      "learning_rate": 5.12912912912913e-06,
      "loss": 0.2388,
      "step": 48660
    },
    {
      "epoch": 146.16,
      "grad_norm": 9.577308654785156,
      "learning_rate": 5.128128128128128e-06,
      "loss": 0.2197,
      "step": 48670
    },
    {
      "epoch": 146.19,
      "grad_norm": 8.072738647460938,
      "learning_rate": 5.127127127127127e-06,
      "loss": 0.2381,
      "step": 48680
    },
    {
      "epoch": 146.22,
      "grad_norm": 7.099912643432617,
      "learning_rate": 5.126126126126127e-06,
      "loss": 0.2421,
      "step": 48690
    },
    {
      "epoch": 146.25,
      "grad_norm": 13.630492210388184,
      "learning_rate": 5.125125125125125e-06,
      "loss": 0.2353,
      "step": 48700
    },
    {
      "epoch": 146.28,
      "grad_norm": 7.355169296264648,
      "learning_rate": 5.124124124124125e-06,
      "loss": 0.219,
      "step": 48710
    },
    {
      "epoch": 146.31,
      "grad_norm": 8.798535346984863,
      "learning_rate": 5.123123123123124e-06,
      "loss": 0.3015,
      "step": 48720
    },
    {
      "epoch": 146.34,
      "grad_norm": 7.992117881774902,
      "learning_rate": 5.122122122122123e-06,
      "loss": 0.2146,
      "step": 48730
    },
    {
      "epoch": 146.37,
      "grad_norm": 7.921085834503174,
      "learning_rate": 5.121121121121122e-06,
      "loss": 0.2365,
      "step": 48740
    },
    {
      "epoch": 146.4,
      "grad_norm": 8.803024291992188,
      "learning_rate": 5.12012012012012e-06,
      "loss": 0.2472,
      "step": 48750
    },
    {
      "epoch": 146.43,
      "grad_norm": 7.738502025604248,
      "learning_rate": 5.119119119119119e-06,
      "loss": 0.2226,
      "step": 48760
    },
    {
      "epoch": 146.46,
      "grad_norm": 13.700687408447266,
      "learning_rate": 5.118118118118118e-06,
      "loss": 0.2123,
      "step": 48770
    },
    {
      "epoch": 146.49,
      "grad_norm": 10.757940292358398,
      "learning_rate": 5.117117117117117e-06,
      "loss": 0.2533,
      "step": 48780
    },
    {
      "epoch": 146.52,
      "grad_norm": 10.62624454498291,
      "learning_rate": 5.116116116116117e-06,
      "loss": 0.247,
      "step": 48790
    },
    {
      "epoch": 146.55,
      "grad_norm": 9.737347602844238,
      "learning_rate": 5.115115115115115e-06,
      "loss": 0.2535,
      "step": 48800
    },
    {
      "epoch": 146.58,
      "grad_norm": 12.586771965026855,
      "learning_rate": 5.114114114114115e-06,
      "loss": 0.2451,
      "step": 48810
    },
    {
      "epoch": 146.61,
      "grad_norm": 10.345483779907227,
      "learning_rate": 5.113113113113114e-06,
      "loss": 0.3,
      "step": 48820
    },
    {
      "epoch": 146.64,
      "grad_norm": 8.725634574890137,
      "learning_rate": 5.112112112112112e-06,
      "loss": 0.2355,
      "step": 48830
    },
    {
      "epoch": 146.67,
      "grad_norm": 10.017885208129883,
      "learning_rate": 5.1111111111111115e-06,
      "loss": 0.2583,
      "step": 48840
    },
    {
      "epoch": 146.7,
      "grad_norm": 4.91624116897583,
      "learning_rate": 5.11011011011011e-06,
      "loss": 0.1984,
      "step": 48850
    },
    {
      "epoch": 146.73,
      "grad_norm": 5.546725273132324,
      "learning_rate": 5.1091091091091095e-06,
      "loss": 0.2404,
      "step": 48860
    },
    {
      "epoch": 146.76,
      "grad_norm": 8.626056671142578,
      "learning_rate": 5.108108108108108e-06,
      "loss": 0.2663,
      "step": 48870
    },
    {
      "epoch": 146.79,
      "grad_norm": 7.942873477935791,
      "learning_rate": 5.1071071071071076e-06,
      "loss": 0.2518,
      "step": 48880
    },
    {
      "epoch": 146.82,
      "grad_norm": 12.471729278564453,
      "learning_rate": 5.106106106106107e-06,
      "loss": 0.3087,
      "step": 48890
    },
    {
      "epoch": 146.85,
      "grad_norm": 6.467403411865234,
      "learning_rate": 5.105105105105106e-06,
      "loss": 0.2464,
      "step": 48900
    },
    {
      "epoch": 146.88,
      "grad_norm": 11.525655746459961,
      "learning_rate": 5.104104104104105e-06,
      "loss": 0.215,
      "step": 48910
    },
    {
      "epoch": 146.91,
      "grad_norm": 8.88946533203125,
      "learning_rate": 5.103103103103103e-06,
      "loss": 0.2497,
      "step": 48920
    },
    {
      "epoch": 146.94,
      "grad_norm": 11.179242134094238,
      "learning_rate": 5.102102102102102e-06,
      "loss": 0.2581,
      "step": 48930
    },
    {
      "epoch": 146.97,
      "grad_norm": 15.216320037841797,
      "learning_rate": 5.101101101101102e-06,
      "loss": 0.2142,
      "step": 48940
    },
    {
      "epoch": 147.0,
      "grad_norm": 12.42892074584961,
      "learning_rate": 5.1001001001001e-06,
      "loss": 0.2354,
      "step": 48950
    },
    {
      "epoch": 147.0,
      "eval_accuracy": 0.9254,
      "eval_loss": 0.32998380064964294,
      "eval_runtime": 30.9555,
      "eval_samples_per_second": 323.044,
      "eval_steps_per_second": 1.292,
      "step": 48951
    },
    {
      "epoch": 147.03,
      "grad_norm": 6.423228740692139,
      "learning_rate": 5.0990990990991e-06,
      "loss": 0.2558,
      "step": 48960
    },
    {
      "epoch": 147.06,
      "grad_norm": 13.780508041381836,
      "learning_rate": 5.098098098098099e-06,
      "loss": 0.2214,
      "step": 48970
    },
    {
      "epoch": 147.09,
      "grad_norm": 6.9629435539245605,
      "learning_rate": 5.097097097097098e-06,
      "loss": 0.241,
      "step": 48980
    },
    {
      "epoch": 147.12,
      "grad_norm": 8.467791557312012,
      "learning_rate": 5.096096096096097e-06,
      "loss": 0.2308,
      "step": 48990
    },
    {
      "epoch": 147.15,
      "grad_norm": 23.57042694091797,
      "learning_rate": 5.095095095095095e-06,
      "loss": 0.2402,
      "step": 49000
    },
    {
      "epoch": 147.18,
      "grad_norm": 71.94815826416016,
      "learning_rate": 5.094094094094094e-06,
      "loss": 0.2674,
      "step": 49010
    },
    {
      "epoch": 147.21,
      "grad_norm": 11.953336715698242,
      "learning_rate": 5.093093093093093e-06,
      "loss": 0.2976,
      "step": 49020
    },
    {
      "epoch": 147.24,
      "grad_norm": 8.680285453796387,
      "learning_rate": 5.092092092092092e-06,
      "loss": 0.2304,
      "step": 49030
    },
    {
      "epoch": 147.27,
      "grad_norm": 7.0171799659729,
      "learning_rate": 5.091091091091092e-06,
      "loss": 0.2402,
      "step": 49040
    },
    {
      "epoch": 147.3,
      "grad_norm": 13.619209289550781,
      "learning_rate": 5.0900900900900905e-06,
      "loss": 0.286,
      "step": 49050
    },
    {
      "epoch": 147.33,
      "grad_norm": 14.42691707611084,
      "learning_rate": 5.08908908908909e-06,
      "loss": 0.2265,
      "step": 49060
    },
    {
      "epoch": 147.36,
      "grad_norm": 7.9618916511535645,
      "learning_rate": 5.088088088088089e-06,
      "loss": 0.2315,
      "step": 49070
    },
    {
      "epoch": 147.39,
      "grad_norm": 10.338654518127441,
      "learning_rate": 5.087087087087087e-06,
      "loss": 0.235,
      "step": 49080
    },
    {
      "epoch": 147.42,
      "grad_norm": 8.249358177185059,
      "learning_rate": 5.0860860860860865e-06,
      "loss": 0.2653,
      "step": 49090
    },
    {
      "epoch": 147.45,
      "grad_norm": 8.354909896850586,
      "learning_rate": 5.085085085085085e-06,
      "loss": 0.2541,
      "step": 49100
    },
    {
      "epoch": 147.48,
      "grad_norm": 9.522316932678223,
      "learning_rate": 5.0840840840840846e-06,
      "loss": 0.2383,
      "step": 49110
    },
    {
      "epoch": 147.51,
      "grad_norm": 6.77536678314209,
      "learning_rate": 5.083083083083083e-06,
      "loss": 0.1885,
      "step": 49120
    },
    {
      "epoch": 147.54,
      "grad_norm": 8.02919864654541,
      "learning_rate": 5.082082082082083e-06,
      "loss": 0.1945,
      "step": 49130
    },
    {
      "epoch": 147.57,
      "grad_norm": 15.834362983703613,
      "learning_rate": 5.081081081081082e-06,
      "loss": 0.2361,
      "step": 49140
    },
    {
      "epoch": 147.6,
      "grad_norm": 7.476492404937744,
      "learning_rate": 5.080080080080081e-06,
      "loss": 0.2385,
      "step": 49150
    },
    {
      "epoch": 147.63,
      "grad_norm": 7.329262733459473,
      "learning_rate": 5.07907907907908e-06,
      "loss": 0.2238,
      "step": 49160
    },
    {
      "epoch": 147.66,
      "grad_norm": 8.222497940063477,
      "learning_rate": 5.078078078078078e-06,
      "loss": 0.2525,
      "step": 49170
    },
    {
      "epoch": 147.69,
      "grad_norm": 7.929909706115723,
      "learning_rate": 5.077077077077077e-06,
      "loss": 0.2502,
      "step": 49180
    },
    {
      "epoch": 147.72,
      "grad_norm": 11.457579612731934,
      "learning_rate": 5.076076076076077e-06,
      "loss": 0.2469,
      "step": 49190
    },
    {
      "epoch": 147.75,
      "grad_norm": 12.695802688598633,
      "learning_rate": 5.075075075075075e-06,
      "loss": 0.2105,
      "step": 49200
    },
    {
      "epoch": 147.78,
      "grad_norm": 9.358962059020996,
      "learning_rate": 5.074074074074075e-06,
      "loss": 0.2336,
      "step": 49210
    },
    {
      "epoch": 147.81,
      "grad_norm": 5.821318626403809,
      "learning_rate": 5.073073073073073e-06,
      "loss": 0.2642,
      "step": 49220
    },
    {
      "epoch": 147.84,
      "grad_norm": 10.369251251220703,
      "learning_rate": 5.072072072072073e-06,
      "loss": 0.2754,
      "step": 49230
    },
    {
      "epoch": 147.87,
      "grad_norm": 28.8354434967041,
      "learning_rate": 5.071071071071072e-06,
      "loss": 0.2639,
      "step": 49240
    },
    {
      "epoch": 147.9,
      "grad_norm": 12.660977363586426,
      "learning_rate": 5.07007007007007e-06,
      "loss": 0.226,
      "step": 49250
    },
    {
      "epoch": 147.93,
      "grad_norm": 9.609078407287598,
      "learning_rate": 5.069069069069069e-06,
      "loss": 0.2769,
      "step": 49260
    },
    {
      "epoch": 147.96,
      "grad_norm": 11.024402618408203,
      "learning_rate": 5.068068068068068e-06,
      "loss": 0.2701,
      "step": 49270
    },
    {
      "epoch": 147.99,
      "grad_norm": 7.1358323097229,
      "learning_rate": 5.0670670670670675e-06,
      "loss": 0.2555,
      "step": 49280
    },
    {
      "epoch": 148.0,
      "eval_accuracy": 0.9236,
      "eval_loss": 0.32654035091400146,
      "eval_runtime": 30.0692,
      "eval_samples_per_second": 332.566,
      "eval_steps_per_second": 1.33,
      "step": 49284
    },
    {
      "epoch": 148.02,
      "grad_norm": 9.342644691467285,
      "learning_rate": 5.066066066066067e-06,
      "loss": 0.1962,
      "step": 49290
    },
    {
      "epoch": 148.05,
      "grad_norm": 8.90712833404541,
      "learning_rate": 5.0650650650650655e-06,
      "loss": 0.2167,
      "step": 49300
    },
    {
      "epoch": 148.08,
      "grad_norm": 5.820209503173828,
      "learning_rate": 5.064064064064065e-06,
      "loss": 0.2343,
      "step": 49310
    },
    {
      "epoch": 148.11,
      "grad_norm": 10.376343727111816,
      "learning_rate": 5.0630630630630635e-06,
      "loss": 0.2305,
      "step": 49320
    },
    {
      "epoch": 148.14,
      "grad_norm": 4.752234935760498,
      "learning_rate": 5.062062062062062e-06,
      "loss": 0.1889,
      "step": 49330
    },
    {
      "epoch": 148.17,
      "grad_norm": 9.236347198486328,
      "learning_rate": 5.061061061061061e-06,
      "loss": 0.2915,
      "step": 49340
    },
    {
      "epoch": 148.2,
      "grad_norm": 6.034448146820068,
      "learning_rate": 5.06006006006006e-06,
      "loss": 0.2616,
      "step": 49350
    },
    {
      "epoch": 148.23,
      "grad_norm": 9.50649642944336,
      "learning_rate": 5.05905905905906e-06,
      "loss": 0.2262,
      "step": 49360
    },
    {
      "epoch": 148.26,
      "grad_norm": 9.511129379272461,
      "learning_rate": 5.058058058058058e-06,
      "loss": 0.2463,
      "step": 49370
    },
    {
      "epoch": 148.29,
      "grad_norm": 7.671658039093018,
      "learning_rate": 5.057057057057058e-06,
      "loss": 0.2557,
      "step": 49380
    },
    {
      "epoch": 148.32,
      "grad_norm": 8.313021659851074,
      "learning_rate": 5.056056056056057e-06,
      "loss": 0.2425,
      "step": 49390
    },
    {
      "epoch": 148.35,
      "grad_norm": 17.637683868408203,
      "learning_rate": 5.055055055055056e-06,
      "loss": 0.2101,
      "step": 49400
    },
    {
      "epoch": 148.38,
      "grad_norm": 13.083399772644043,
      "learning_rate": 5.054054054054054e-06,
      "loss": 0.2189,
      "step": 49410
    },
    {
      "epoch": 148.41,
      "grad_norm": 7.656839370727539,
      "learning_rate": 5.053053053053053e-06,
      "loss": 0.2371,
      "step": 49420
    },
    {
      "epoch": 148.44,
      "grad_norm": 6.677846431732178,
      "learning_rate": 5.052052052052052e-06,
      "loss": 0.2129,
      "step": 49430
    },
    {
      "epoch": 148.47,
      "grad_norm": 17.035531997680664,
      "learning_rate": 5.051051051051051e-06,
      "loss": 0.2728,
      "step": 49440
    },
    {
      "epoch": 148.5,
      "grad_norm": 22.835166931152344,
      "learning_rate": 5.05005005005005e-06,
      "loss": 0.2347,
      "step": 49450
    },
    {
      "epoch": 148.53,
      "grad_norm": 8.110856056213379,
      "learning_rate": 5.04904904904905e-06,
      "loss": 0.245,
      "step": 49460
    },
    {
      "epoch": 148.56,
      "grad_norm": 8.6338529586792,
      "learning_rate": 5.048048048048048e-06,
      "loss": 0.2076,
      "step": 49470
    },
    {
      "epoch": 148.59,
      "grad_norm": 6.641716480255127,
      "learning_rate": 5.047047047047048e-06,
      "loss": 0.2409,
      "step": 49480
    },
    {
      "epoch": 148.62,
      "grad_norm": 6.18310022354126,
      "learning_rate": 5.046046046046047e-06,
      "loss": 0.2953,
      "step": 49490
    },
    {
      "epoch": 148.65,
      "grad_norm": 8.047459602355957,
      "learning_rate": 5.045045045045045e-06,
      "loss": 0.233,
      "step": 49500
    },
    {
      "epoch": 148.68,
      "grad_norm": 29.734088897705078,
      "learning_rate": 5.0440440440440444e-06,
      "loss": 0.2311,
      "step": 49510
    },
    {
      "epoch": 148.71,
      "grad_norm": 10.40032958984375,
      "learning_rate": 5.043043043043043e-06,
      "loss": 0.1819,
      "step": 49520
    },
    {
      "epoch": 148.74,
      "grad_norm": 11.605247497558594,
      "learning_rate": 5.0420420420420425e-06,
      "loss": 0.2913,
      "step": 49530
    },
    {
      "epoch": 148.77,
      "grad_norm": 8.537494659423828,
      "learning_rate": 5.041041041041042e-06,
      "loss": 0.2468,
      "step": 49540
    },
    {
      "epoch": 148.8,
      "grad_norm": 5.692840099334717,
      "learning_rate": 5.0400400400400405e-06,
      "loss": 0.2081,
      "step": 49550
    },
    {
      "epoch": 148.83,
      "grad_norm": 5.905738353729248,
      "learning_rate": 5.03903903903904e-06,
      "loss": 0.2394,
      "step": 49560
    },
    {
      "epoch": 148.86,
      "grad_norm": 11.151970863342285,
      "learning_rate": 5.0380380380380386e-06,
      "loss": 0.2294,
      "step": 49570
    },
    {
      "epoch": 148.89,
      "grad_norm": 7.9914116859436035,
      "learning_rate": 5.037037037037037e-06,
      "loss": 0.2759,
      "step": 49580
    },
    {
      "epoch": 148.92,
      "grad_norm": 7.827892303466797,
      "learning_rate": 5.036036036036036e-06,
      "loss": 0.1982,
      "step": 49590
    },
    {
      "epoch": 148.95,
      "grad_norm": 6.978936672210693,
      "learning_rate": 5.035035035035035e-06,
      "loss": 0.234,
      "step": 49600
    },
    {
      "epoch": 148.98,
      "grad_norm": 24.396188735961914,
      "learning_rate": 5.034034034034035e-06,
      "loss": 0.2389,
      "step": 49610
    },
    {
      "epoch": 149.0,
      "eval_accuracy": 0.9235,
      "eval_loss": 0.3329337239265442,
      "eval_runtime": 30.1447,
      "eval_samples_per_second": 331.734,
      "eval_steps_per_second": 1.327,
      "step": 49617
    },
    {
      "epoch": 149.01,
      "grad_norm": 9.483830451965332,
      "learning_rate": 5.033033033033033e-06,
      "loss": 0.3218,
      "step": 49620
    },
    {
      "epoch": 149.04,
      "grad_norm": 10.066933631896973,
      "learning_rate": 5.032032032032033e-06,
      "loss": 0.2531,
      "step": 49630
    },
    {
      "epoch": 149.07,
      "grad_norm": 7.273628234863281,
      "learning_rate": 5.031031031031032e-06,
      "loss": 0.2118,
      "step": 49640
    },
    {
      "epoch": 149.1,
      "grad_norm": 13.651021957397461,
      "learning_rate": 5.030030030030031e-06,
      "loss": 0.2366,
      "step": 49650
    },
    {
      "epoch": 149.13,
      "grad_norm": 6.743479251861572,
      "learning_rate": 5.029029029029029e-06,
      "loss": 0.2435,
      "step": 49660
    },
    {
      "epoch": 149.16,
      "grad_norm": 14.561511993408203,
      "learning_rate": 5.028028028028028e-06,
      "loss": 0.2136,
      "step": 49670
    },
    {
      "epoch": 149.19,
      "grad_norm": 10.303435325622559,
      "learning_rate": 5.027027027027027e-06,
      "loss": 0.2459,
      "step": 49680
    },
    {
      "epoch": 149.22,
      "grad_norm": 6.597631931304932,
      "learning_rate": 5.026026026026026e-06,
      "loss": 0.2708,
      "step": 49690
    },
    {
      "epoch": 149.25,
      "grad_norm": 7.308007717132568,
      "learning_rate": 5.025025025025025e-06,
      "loss": 0.2079,
      "step": 49700
    },
    {
      "epoch": 149.28,
      "grad_norm": 6.979787826538086,
      "learning_rate": 5.024024024024025e-06,
      "loss": 0.2264,
      "step": 49710
    },
    {
      "epoch": 149.31,
      "grad_norm": 13.906595230102539,
      "learning_rate": 5.023023023023023e-06,
      "loss": 0.2673,
      "step": 49720
    },
    {
      "epoch": 149.34,
      "grad_norm": 14.35170841217041,
      "learning_rate": 5.022022022022023e-06,
      "loss": 0.2238,
      "step": 49730
    },
    {
      "epoch": 149.37,
      "grad_norm": 9.655450820922852,
      "learning_rate": 5.021021021021022e-06,
      "loss": 0.195,
      "step": 49740
    },
    {
      "epoch": 149.4,
      "grad_norm": 14.673532485961914,
      "learning_rate": 5.02002002002002e-06,
      "loss": 0.1986,
      "step": 49750
    },
    {
      "epoch": 149.43,
      "grad_norm": 9.177469253540039,
      "learning_rate": 5.0190190190190195e-06,
      "loss": 0.2133,
      "step": 49760
    },
    {
      "epoch": 149.46,
      "grad_norm": 4.796307563781738,
      "learning_rate": 5.018018018018018e-06,
      "loss": 0.1961,
      "step": 49770
    },
    {
      "epoch": 149.49,
      "grad_norm": 8.587081909179688,
      "learning_rate": 5.0170170170170175e-06,
      "loss": 0.2309,
      "step": 49780
    },
    {
      "epoch": 149.52,
      "grad_norm": 10.213457107543945,
      "learning_rate": 5.016016016016016e-06,
      "loss": 0.2421,
      "step": 49790
    },
    {
      "epoch": 149.55,
      "grad_norm": 12.847905158996582,
      "learning_rate": 5.0150150150150156e-06,
      "loss": 0.2221,
      "step": 49800
    },
    {
      "epoch": 149.58,
      "grad_norm": 11.52994155883789,
      "learning_rate": 5.014014014014015e-06,
      "loss": 0.2664,
      "step": 49810
    },
    {
      "epoch": 149.61,
      "grad_norm": 8.524340629577637,
      "learning_rate": 5.013013013013014e-06,
      "loss": 0.2276,
      "step": 49820
    },
    {
      "epoch": 149.64,
      "grad_norm": 9.458572387695312,
      "learning_rate": 5.012012012012012e-06,
      "loss": 0.2214,
      "step": 49830
    },
    {
      "epoch": 149.67,
      "grad_norm": 7.292261123657227,
      "learning_rate": 5.011011011011011e-06,
      "loss": 0.2409,
      "step": 49840
    },
    {
      "epoch": 149.7,
      "grad_norm": 9.84286880493164,
      "learning_rate": 5.01001001001001e-06,
      "loss": 0.2908,
      "step": 49850
    },
    {
      "epoch": 149.73,
      "grad_norm": 12.440919876098633,
      "learning_rate": 5.00900900900901e-06,
      "loss": 0.2319,
      "step": 49860
    },
    {
      "epoch": 149.76,
      "grad_norm": 14.211118698120117,
      "learning_rate": 5.008008008008008e-06,
      "loss": 0.2442,
      "step": 49870
    },
    {
      "epoch": 149.79,
      "grad_norm": 16.026323318481445,
      "learning_rate": 5.007007007007008e-06,
      "loss": 0.2065,
      "step": 49880
    },
    {
      "epoch": 149.82,
      "grad_norm": 17.31975555419922,
      "learning_rate": 5.006006006006007e-06,
      "loss": 0.2369,
      "step": 49890
    },
    {
      "epoch": 149.85,
      "grad_norm": 7.937825679779053,
      "learning_rate": 5.005005005005006e-06,
      "loss": 0.2218,
      "step": 49900
    },
    {
      "epoch": 149.88,
      "grad_norm": 8.57983684539795,
      "learning_rate": 5.0040040040040035e-06,
      "loss": 0.2263,
      "step": 49910
    },
    {
      "epoch": 149.91,
      "grad_norm": 6.879357814788818,
      "learning_rate": 5.003003003003003e-06,
      "loss": 0.251,
      "step": 49920
    },
    {
      "epoch": 149.94,
      "grad_norm": 7.593476295471191,
      "learning_rate": 5.002002002002002e-06,
      "loss": 0.2024,
      "step": 49930
    },
    {
      "epoch": 149.97,
      "grad_norm": 9.840690612792969,
      "learning_rate": 5.001001001001001e-06,
      "loss": 0.2189,
      "step": 49940
    },
    {
      "epoch": 150.0,
      "grad_norm": 1.2699086666107178,
      "learning_rate": 5e-06,
      "loss": 0.2441,
      "step": 49950
    },
    {
      "epoch": 150.0,
      "eval_accuracy": 0.924,
      "eval_loss": 0.32222384214401245,
      "eval_runtime": 29.7652,
      "eval_samples_per_second": 335.962,
      "eval_steps_per_second": 1.344,
      "step": 49950
    },
    {
      "epoch": 150.03,
      "grad_norm": 6.871353626251221,
      "learning_rate": 4.998998998999e-06,
      "loss": 0.2117,
      "step": 49960
    },
    {
      "epoch": 150.06,
      "grad_norm": 14.333979606628418,
      "learning_rate": 4.9979979979979984e-06,
      "loss": 0.2088,
      "step": 49970
    },
    {
      "epoch": 150.09,
      "grad_norm": 6.679909706115723,
      "learning_rate": 4.996996996996997e-06,
      "loss": 0.2714,
      "step": 49980
    },
    {
      "epoch": 150.12,
      "grad_norm": 6.596663475036621,
      "learning_rate": 4.9959959959959965e-06,
      "loss": 0.2432,
      "step": 49990
    },
    {
      "epoch": 150.15,
      "grad_norm": 9.277660369873047,
      "learning_rate": 4.994994994994996e-06,
      "loss": 0.1969,
      "step": 50000
    },
    {
      "epoch": 150.18,
      "grad_norm": 9.219948768615723,
      "learning_rate": 4.9939939939939945e-06,
      "loss": 0.2504,
      "step": 50010
    },
    {
      "epoch": 150.21,
      "grad_norm": 9.945173263549805,
      "learning_rate": 4.992992992992993e-06,
      "loss": 0.2511,
      "step": 50020
    },
    {
      "epoch": 150.24,
      "grad_norm": 14.701431274414062,
      "learning_rate": 4.9919919919919926e-06,
      "loss": 0.2097,
      "step": 50030
    },
    {
      "epoch": 150.27,
      "grad_norm": 5.916761875152588,
      "learning_rate": 4.990990990990991e-06,
      "loss": 0.2359,
      "step": 50040
    },
    {
      "epoch": 150.3,
      "grad_norm": 11.579957962036133,
      "learning_rate": 4.989989989989991e-06,
      "loss": 0.2352,
      "step": 50050
    },
    {
      "epoch": 150.33,
      "grad_norm": 9.476393699645996,
      "learning_rate": 4.988988988988989e-06,
      "loss": 0.2578,
      "step": 50060
    },
    {
      "epoch": 150.36,
      "grad_norm": 11.668774604797363,
      "learning_rate": 4.987987987987989e-06,
      "loss": 0.2605,
      "step": 50070
    },
    {
      "epoch": 150.39,
      "grad_norm": 10.833105087280273,
      "learning_rate": 4.986986986986987e-06,
      "loss": 0.2156,
      "step": 50080
    },
    {
      "epoch": 150.42,
      "grad_norm": 8.86446475982666,
      "learning_rate": 4.985985985985987e-06,
      "loss": 0.1935,
      "step": 50090
    },
    {
      "epoch": 150.45,
      "grad_norm": 8.010031700134277,
      "learning_rate": 4.984984984984985e-06,
      "loss": 0.2343,
      "step": 50100
    },
    {
      "epoch": 150.48,
      "grad_norm": 7.511141777038574,
      "learning_rate": 4.983983983983985e-06,
      "loss": 0.2369,
      "step": 50110
    },
    {
      "epoch": 150.51,
      "grad_norm": 12.473719596862793,
      "learning_rate": 4.982982982982983e-06,
      "loss": 0.2304,
      "step": 50120
    },
    {
      "epoch": 150.54,
      "grad_norm": 12.684759140014648,
      "learning_rate": 4.981981981981983e-06,
      "loss": 0.2103,
      "step": 50130
    },
    {
      "epoch": 150.57,
      "grad_norm": 9.042656898498535,
      "learning_rate": 4.980980980980981e-06,
      "loss": 0.2467,
      "step": 50140
    },
    {
      "epoch": 150.6,
      "grad_norm": 5.764816761016846,
      "learning_rate": 4.97997997997998e-06,
      "loss": 0.2248,
      "step": 50150
    },
    {
      "epoch": 150.63,
      "grad_norm": 8.748346328735352,
      "learning_rate": 4.978978978978979e-06,
      "loss": 0.2775,
      "step": 50160
    },
    {
      "epoch": 150.66,
      "grad_norm": 7.7015252113342285,
      "learning_rate": 4.977977977977979e-06,
      "loss": 0.252,
      "step": 50170
    },
    {
      "epoch": 150.69,
      "grad_norm": 9.170125961303711,
      "learning_rate": 4.976976976976977e-06,
      "loss": 0.2128,
      "step": 50180
    },
    {
      "epoch": 150.72,
      "grad_norm": 13.892009735107422,
      "learning_rate": 4.975975975975976e-06,
      "loss": 0.2494,
      "step": 50190
    },
    {
      "epoch": 150.75,
      "grad_norm": 7.737305164337158,
      "learning_rate": 4.9749749749749754e-06,
      "loss": 0.3093,
      "step": 50200
    },
    {
      "epoch": 150.78,
      "grad_norm": 9.453953742980957,
      "learning_rate": 4.973973973973975e-06,
      "loss": 0.2236,
      "step": 50210
    },
    {
      "epoch": 150.81,
      "grad_norm": 16.679332733154297,
      "learning_rate": 4.9729729729729735e-06,
      "loss": 0.2338,
      "step": 50220
    },
    {
      "epoch": 150.84,
      "grad_norm": 7.992870330810547,
      "learning_rate": 4.971971971971972e-06,
      "loss": 0.2341,
      "step": 50230
    },
    {
      "epoch": 150.87,
      "grad_norm": 11.4469633102417,
      "learning_rate": 4.9709709709709715e-06,
      "loss": 0.2262,
      "step": 50240
    },
    {
      "epoch": 150.9,
      "grad_norm": 57.37582015991211,
      "learning_rate": 4.96996996996997e-06,
      "loss": 0.2519,
      "step": 50250
    },
    {
      "epoch": 150.93,
      "grad_norm": 11.643623352050781,
      "learning_rate": 4.9689689689689696e-06,
      "loss": 0.242,
      "step": 50260
    },
    {
      "epoch": 150.96,
      "grad_norm": 7.651035308837891,
      "learning_rate": 4.967967967967968e-06,
      "loss": 0.2035,
      "step": 50270
    },
    {
      "epoch": 150.99,
      "grad_norm": 9.384330749511719,
      "learning_rate": 4.966966966966968e-06,
      "loss": 0.2419,
      "step": 50280
    },
    {
      "epoch": 151.0,
      "eval_accuracy": 0.9235,
      "eval_loss": 0.32870054244995117,
      "eval_runtime": 30.035,
      "eval_samples_per_second": 332.945,
      "eval_steps_per_second": 1.332,
      "step": 50283
    },
    {
      "epoch": 151.02,
      "grad_norm": 7.480780124664307,
      "learning_rate": 4.965965965965966e-06,
      "loss": 0.2101,
      "step": 50290
    },
    {
      "epoch": 151.05,
      "grad_norm": 6.316562652587891,
      "learning_rate": 4.964964964964966e-06,
      "loss": 0.2485,
      "step": 50300
    },
    {
      "epoch": 151.08,
      "grad_norm": 5.209827423095703,
      "learning_rate": 4.963963963963964e-06,
      "loss": 0.2216,
      "step": 50310
    },
    {
      "epoch": 151.11,
      "grad_norm": 10.14749813079834,
      "learning_rate": 4.962962962962964e-06,
      "loss": 0.2334,
      "step": 50320
    },
    {
      "epoch": 151.14,
      "grad_norm": 10.345035552978516,
      "learning_rate": 4.961961961961962e-06,
      "loss": 0.2415,
      "step": 50330
    },
    {
      "epoch": 151.17,
      "grad_norm": 35.78424072265625,
      "learning_rate": 4.960960960960962e-06,
      "loss": 0.273,
      "step": 50340
    },
    {
      "epoch": 151.2,
      "grad_norm": 9.070600509643555,
      "learning_rate": 4.95995995995996e-06,
      "loss": 0.2354,
      "step": 50350
    },
    {
      "epoch": 151.23,
      "grad_norm": 17.64927864074707,
      "learning_rate": 4.958958958958959e-06,
      "loss": 0.2546,
      "step": 50360
    },
    {
      "epoch": 151.26,
      "grad_norm": 8.714930534362793,
      "learning_rate": 4.957957957957958e-06,
      "loss": 0.2106,
      "step": 50370
    },
    {
      "epoch": 151.29,
      "grad_norm": 6.82290506362915,
      "learning_rate": 4.956956956956958e-06,
      "loss": 0.2579,
      "step": 50380
    },
    {
      "epoch": 151.32,
      "grad_norm": 13.768546104431152,
      "learning_rate": 4.955955955955956e-06,
      "loss": 0.2047,
      "step": 50390
    },
    {
      "epoch": 151.35,
      "grad_norm": 9.172098159790039,
      "learning_rate": 4.954954954954955e-06,
      "loss": 0.2094,
      "step": 50400
    },
    {
      "epoch": 151.38,
      "grad_norm": 12.838061332702637,
      "learning_rate": 4.953953953953954e-06,
      "loss": 0.2258,
      "step": 50410
    },
    {
      "epoch": 151.41,
      "grad_norm": 12.923576354980469,
      "learning_rate": 4.952952952952954e-06,
      "loss": 0.2724,
      "step": 50420
    },
    {
      "epoch": 151.44,
      "grad_norm": 11.85954475402832,
      "learning_rate": 4.9519519519519524e-06,
      "loss": 0.2549,
      "step": 50430
    },
    {
      "epoch": 151.47,
      "grad_norm": 11.876255989074707,
      "learning_rate": 4.950950950950951e-06,
      "loss": 0.2082,
      "step": 50440
    },
    {
      "epoch": 151.5,
      "grad_norm": 12.164875030517578,
      "learning_rate": 4.9499499499499505e-06,
      "loss": 0.2229,
      "step": 50450
    },
    {
      "epoch": 151.53,
      "grad_norm": 6.513701438903809,
      "learning_rate": 4.94894894894895e-06,
      "loss": 0.2477,
      "step": 50460
    },
    {
      "epoch": 151.56,
      "grad_norm": 8.207629203796387,
      "learning_rate": 4.947947947947948e-06,
      "loss": 0.28,
      "step": 50470
    },
    {
      "epoch": 151.59,
      "grad_norm": 12.85261344909668,
      "learning_rate": 4.946946946946947e-06,
      "loss": 0.2646,
      "step": 50480
    },
    {
      "epoch": 151.62,
      "grad_norm": 12.279462814331055,
      "learning_rate": 4.9459459459459466e-06,
      "loss": 0.2208,
      "step": 50490
    },
    {
      "epoch": 151.65,
      "grad_norm": 32.35654067993164,
      "learning_rate": 4.944944944944945e-06,
      "loss": 0.2404,
      "step": 50500
    },
    {
      "epoch": 151.68,
      "grad_norm": 12.754739761352539,
      "learning_rate": 4.943943943943945e-06,
      "loss": 0.2192,
      "step": 50510
    },
    {
      "epoch": 151.71,
      "grad_norm": 5.905869007110596,
      "learning_rate": 4.942942942942943e-06,
      "loss": 0.2503,
      "step": 50520
    },
    {
      "epoch": 151.74,
      "grad_norm": 8.1692533493042,
      "learning_rate": 4.941941941941943e-06,
      "loss": 0.2769,
      "step": 50530
    },
    {
      "epoch": 151.77,
      "grad_norm": 6.464527606964111,
      "learning_rate": 4.940940940940941e-06,
      "loss": 0.2096,
      "step": 50540
    },
    {
      "epoch": 151.8,
      "grad_norm": 6.255577564239502,
      "learning_rate": 4.939939939939941e-06,
      "loss": 0.2458,
      "step": 50550
    },
    {
      "epoch": 151.83,
      "grad_norm": 8.056628227233887,
      "learning_rate": 4.938938938938939e-06,
      "loss": 0.2579,
      "step": 50560
    },
    {
      "epoch": 151.86,
      "grad_norm": 10.514535903930664,
      "learning_rate": 4.937937937937939e-06,
      "loss": 0.2069,
      "step": 50570
    },
    {
      "epoch": 151.89,
      "grad_norm": 7.967158794403076,
      "learning_rate": 4.936936936936937e-06,
      "loss": 0.2222,
      "step": 50580
    },
    {
      "epoch": 151.92,
      "grad_norm": 5.999773025512695,
      "learning_rate": 4.935935935935937e-06,
      "loss": 0.2317,
      "step": 50590
    },
    {
      "epoch": 151.95,
      "grad_norm": 8.459094047546387,
      "learning_rate": 4.934934934934935e-06,
      "loss": 0.2088,
      "step": 50600
    },
    {
      "epoch": 151.98,
      "grad_norm": 16.82901382446289,
      "learning_rate": 4.933933933933934e-06,
      "loss": 0.2293,
      "step": 50610
    },
    {
      "epoch": 152.0,
      "eval_accuracy": 0.9221,
      "eval_loss": 0.3310253322124481,
      "eval_runtime": 30.2848,
      "eval_samples_per_second": 330.198,
      "eval_steps_per_second": 1.321,
      "step": 50616
    },
    {
      "epoch": 152.01,
      "grad_norm": 8.360386848449707,
      "learning_rate": 4.932932932932933e-06,
      "loss": 0.2462,
      "step": 50620
    },
    {
      "epoch": 152.04,
      "grad_norm": 9.696908950805664,
      "learning_rate": 4.931931931931933e-06,
      "loss": 0.2136,
      "step": 50630
    },
    {
      "epoch": 152.07,
      "grad_norm": 6.3286542892456055,
      "learning_rate": 4.930930930930931e-06,
      "loss": 0.2422,
      "step": 50640
    },
    {
      "epoch": 152.1,
      "grad_norm": 5.575071334838867,
      "learning_rate": 4.92992992992993e-06,
      "loss": 0.2279,
      "step": 50650
    },
    {
      "epoch": 152.13,
      "grad_norm": 12.49942398071289,
      "learning_rate": 4.9289289289289294e-06,
      "loss": 0.218,
      "step": 50660
    },
    {
      "epoch": 152.16,
      "grad_norm": 62.22034454345703,
      "learning_rate": 4.927927927927929e-06,
      "loss": 0.1984,
      "step": 50670
    },
    {
      "epoch": 152.19,
      "grad_norm": 8.51760196685791,
      "learning_rate": 4.9269269269269275e-06,
      "loss": 0.2263,
      "step": 50680
    },
    {
      "epoch": 152.22,
      "grad_norm": 15.988168716430664,
      "learning_rate": 4.925925925925926e-06,
      "loss": 0.2162,
      "step": 50690
    },
    {
      "epoch": 152.25,
      "grad_norm": 10.93830680847168,
      "learning_rate": 4.9249249249249255e-06,
      "loss": 0.199,
      "step": 50700
    },
    {
      "epoch": 152.28,
      "grad_norm": 6.854659080505371,
      "learning_rate": 4.923923923923924e-06,
      "loss": 0.2199,
      "step": 50710
    },
    {
      "epoch": 152.31,
      "grad_norm": 14.560139656066895,
      "learning_rate": 4.922922922922923e-06,
      "loss": 0.2443,
      "step": 50720
    },
    {
      "epoch": 152.34,
      "grad_norm": 10.161324501037598,
      "learning_rate": 4.921921921921922e-06,
      "loss": 0.2612,
      "step": 50730
    },
    {
      "epoch": 152.37,
      "grad_norm": 10.49738883972168,
      "learning_rate": 4.920920920920922e-06,
      "loss": 0.1866,
      "step": 50740
    },
    {
      "epoch": 152.4,
      "grad_norm": 9.785523414611816,
      "learning_rate": 4.91991991991992e-06,
      "loss": 0.2312,
      "step": 50750
    },
    {
      "epoch": 152.43,
      "grad_norm": 12.919539451599121,
      "learning_rate": 4.91891891891892e-06,
      "loss": 0.2389,
      "step": 50760
    },
    {
      "epoch": 152.46,
      "grad_norm": 6.534263610839844,
      "learning_rate": 4.917917917917918e-06,
      "loss": 0.248,
      "step": 50770
    },
    {
      "epoch": 152.49,
      "grad_norm": 6.204430103302002,
      "learning_rate": 4.916916916916918e-06,
      "loss": 0.1724,
      "step": 50780
    },
    {
      "epoch": 152.52,
      "grad_norm": 7.975450038909912,
      "learning_rate": 4.915915915915916e-06,
      "loss": 0.2245,
      "step": 50790
    },
    {
      "epoch": 152.55,
      "grad_norm": 7.731273651123047,
      "learning_rate": 4.914914914914916e-06,
      "loss": 0.2285,
      "step": 50800
    },
    {
      "epoch": 152.58,
      "grad_norm": 8.033792495727539,
      "learning_rate": 4.913913913913914e-06,
      "loss": 0.2402,
      "step": 50810
    },
    {
      "epoch": 152.61,
      "grad_norm": 6.591159343719482,
      "learning_rate": 4.912912912912913e-06,
      "loss": 0.2474,
      "step": 50820
    },
    {
      "epoch": 152.64,
      "grad_norm": 7.779619216918945,
      "learning_rate": 4.911911911911912e-06,
      "loss": 0.1948,
      "step": 50830
    },
    {
      "epoch": 152.67,
      "grad_norm": 22.785364151000977,
      "learning_rate": 4.910910910910912e-06,
      "loss": 0.2505,
      "step": 50840
    },
    {
      "epoch": 152.7,
      "grad_norm": 6.532778263092041,
      "learning_rate": 4.90990990990991e-06,
      "loss": 0.2422,
      "step": 50850
    },
    {
      "epoch": 152.73,
      "grad_norm": 9.579747200012207,
      "learning_rate": 4.908908908908909e-06,
      "loss": 0.2437,
      "step": 50860
    },
    {
      "epoch": 152.76,
      "grad_norm": 10.946322441101074,
      "learning_rate": 4.907907907907908e-06,
      "loss": 0.2484,
      "step": 50870
    },
    {
      "epoch": 152.79,
      "grad_norm": 8.377988815307617,
      "learning_rate": 4.906906906906908e-06,
      "loss": 0.2449,
      "step": 50880
    },
    {
      "epoch": 152.82,
      "grad_norm": 13.010729789733887,
      "learning_rate": 4.9059059059059064e-06,
      "loss": 0.2285,
      "step": 50890
    },
    {
      "epoch": 152.85,
      "grad_norm": 6.976149082183838,
      "learning_rate": 4.904904904904905e-06,
      "loss": 0.2426,
      "step": 50900
    },
    {
      "epoch": 152.88,
      "grad_norm": 17.81083106994629,
      "learning_rate": 4.9039039039039045e-06,
      "loss": 0.2314,
      "step": 50910
    },
    {
      "epoch": 152.91,
      "grad_norm": 6.334768772125244,
      "learning_rate": 4.902902902902904e-06,
      "loss": 0.2086,
      "step": 50920
    },
    {
      "epoch": 152.94,
      "grad_norm": 10.830709457397461,
      "learning_rate": 4.901901901901902e-06,
      "loss": 0.2429,
      "step": 50930
    },
    {
      "epoch": 152.97,
      "grad_norm": 7.8575029373168945,
      "learning_rate": 4.900900900900901e-06,
      "loss": 0.2432,
      "step": 50940
    },
    {
      "epoch": 153.0,
      "eval_accuracy": 0.9237,
      "eval_loss": 0.32934698462486267,
      "eval_runtime": 30.1549,
      "eval_samples_per_second": 331.621,
      "eval_steps_per_second": 1.326,
      "step": 50949
    },
    {
      "epoch": 153.0,
      "grad_norm": 9.523240089416504,
      "learning_rate": 4.8998998998999005e-06,
      "loss": 0.2084,
      "step": 50950
    },
    {
      "epoch": 153.03,
      "grad_norm": 14.13718032836914,
      "learning_rate": 4.898898898898899e-06,
      "loss": 0.2544,
      "step": 50960
    },
    {
      "epoch": 153.06,
      "grad_norm": 5.4915771484375,
      "learning_rate": 4.897897897897898e-06,
      "loss": 0.2229,
      "step": 50970
    },
    {
      "epoch": 153.09,
      "grad_norm": 11.770608901977539,
      "learning_rate": 4.896896896896897e-06,
      "loss": 0.1975,
      "step": 50980
    },
    {
      "epoch": 153.12,
      "grad_norm": 15.267172813415527,
      "learning_rate": 4.895895895895897e-06,
      "loss": 0.2245,
      "step": 50990
    },
    {
      "epoch": 153.15,
      "grad_norm": 9.833891868591309,
      "learning_rate": 4.894894894894895e-06,
      "loss": 0.2584,
      "step": 51000
    },
    {
      "epoch": 153.18,
      "grad_norm": 5.989986419677734,
      "learning_rate": 4.893893893893894e-06,
      "loss": 0.2607,
      "step": 51010
    },
    {
      "epoch": 153.21,
      "grad_norm": 11.517068862915039,
      "learning_rate": 4.892892892892893e-06,
      "loss": 0.2261,
      "step": 51020
    },
    {
      "epoch": 153.24,
      "grad_norm": 9.926610946655273,
      "learning_rate": 4.891891891891893e-06,
      "loss": 0.2196,
      "step": 51030
    },
    {
      "epoch": 153.27,
      "grad_norm": 9.067619323730469,
      "learning_rate": 4.890890890890891e-06,
      "loss": 0.2194,
      "step": 51040
    },
    {
      "epoch": 153.3,
      "grad_norm": 7.116804599761963,
      "learning_rate": 4.889889889889891e-06,
      "loss": 0.2468,
      "step": 51050
    },
    {
      "epoch": 153.33,
      "grad_norm": 7.517953872680664,
      "learning_rate": 4.888888888888889e-06,
      "loss": 0.256,
      "step": 51060
    },
    {
      "epoch": 153.36,
      "grad_norm": 7.633245944976807,
      "learning_rate": 4.887887887887888e-06,
      "loss": 0.217,
      "step": 51070
    },
    {
      "epoch": 153.39,
      "grad_norm": 13.708901405334473,
      "learning_rate": 4.886886886886887e-06,
      "loss": 0.2391,
      "step": 51080
    },
    {
      "epoch": 153.42,
      "grad_norm": 24.10340690612793,
      "learning_rate": 4.885885885885887e-06,
      "loss": 0.2197,
      "step": 51090
    },
    {
      "epoch": 153.45,
      "grad_norm": 11.549978256225586,
      "learning_rate": 4.884884884884885e-06,
      "loss": 0.2177,
      "step": 51100
    },
    {
      "epoch": 153.48,
      "grad_norm": 6.874415874481201,
      "learning_rate": 4.883883883883884e-06,
      "loss": 0.2126,
      "step": 51110
    },
    {
      "epoch": 153.51,
      "grad_norm": 8.026674270629883,
      "learning_rate": 4.8828828828828834e-06,
      "loss": 0.2625,
      "step": 51120
    },
    {
      "epoch": 153.54,
      "grad_norm": 6.055148124694824,
      "learning_rate": 4.881881881881883e-06,
      "loss": 0.2445,
      "step": 51130
    },
    {
      "epoch": 153.57,
      "grad_norm": 8.034958839416504,
      "learning_rate": 4.8808808808808815e-06,
      "loss": 0.2131,
      "step": 51140
    },
    {
      "epoch": 153.6,
      "grad_norm": 5.177096366882324,
      "learning_rate": 4.87987987987988e-06,
      "loss": 0.2102,
      "step": 51150
    },
    {
      "epoch": 153.63,
      "grad_norm": 8.35079288482666,
      "learning_rate": 4.8788788788788795e-06,
      "loss": 0.2001,
      "step": 51160
    },
    {
      "epoch": 153.66,
      "grad_norm": 7.677814960479736,
      "learning_rate": 4.877877877877878e-06,
      "loss": 0.284,
      "step": 51170
    },
    {
      "epoch": 153.69,
      "grad_norm": 6.635144233703613,
      "learning_rate": 4.876876876876877e-06,
      "loss": 0.2238,
      "step": 51180
    },
    {
      "epoch": 153.72,
      "grad_norm": 6.808417320251465,
      "learning_rate": 4.875875875875876e-06,
      "loss": 0.2282,
      "step": 51190
    },
    {
      "epoch": 153.75,
      "grad_norm": 6.606178283691406,
      "learning_rate": 4.874874874874876e-06,
      "loss": 0.217,
      "step": 51200
    },
    {
      "epoch": 153.78,
      "grad_norm": 22.319948196411133,
      "learning_rate": 4.873873873873874e-06,
      "loss": 0.2416,
      "step": 51210
    },
    {
      "epoch": 153.81,
      "grad_norm": 9.870071411132812,
      "learning_rate": 4.872872872872873e-06,
      "loss": 0.2123,
      "step": 51220
    },
    {
      "epoch": 153.84,
      "grad_norm": 9.52846622467041,
      "learning_rate": 4.871871871871872e-06,
      "loss": 0.2187,
      "step": 51230
    },
    {
      "epoch": 153.87,
      "grad_norm": 21.089841842651367,
      "learning_rate": 4.870870870870872e-06,
      "loss": 0.2647,
      "step": 51240
    },
    {
      "epoch": 153.9,
      "grad_norm": 28.91900062561035,
      "learning_rate": 4.86986986986987e-06,
      "loss": 0.2564,
      "step": 51250
    },
    {
      "epoch": 153.93,
      "grad_norm": 7.906915664672852,
      "learning_rate": 4.868868868868869e-06,
      "loss": 0.2255,
      "step": 51260
    },
    {
      "epoch": 153.96,
      "grad_norm": 8.276420593261719,
      "learning_rate": 4.867867867867868e-06,
      "loss": 0.2252,
      "step": 51270
    },
    {
      "epoch": 153.99,
      "grad_norm": 6.902585029602051,
      "learning_rate": 4.866866866866867e-06,
      "loss": 0.2255,
      "step": 51280
    },
    {
      "epoch": 154.0,
      "eval_accuracy": 0.9257,
      "eval_loss": 0.33379286527633667,
      "eval_runtime": 29.8745,
      "eval_samples_per_second": 334.734,
      "eval_steps_per_second": 1.339,
      "step": 51282
    },
    {
      "epoch": 154.02,
      "grad_norm": 8.701339721679688,
      "learning_rate": 4.865865865865866e-06,
      "loss": 0.256,
      "step": 51290
    },
    {
      "epoch": 154.05,
      "grad_norm": 4.921034336090088,
      "learning_rate": 4.864864864864866e-06,
      "loss": 0.2742,
      "step": 51300
    },
    {
      "epoch": 154.08,
      "grad_norm": 7.366501331329346,
      "learning_rate": 4.863863863863864e-06,
      "loss": 0.2528,
      "step": 51310
    },
    {
      "epoch": 154.11,
      "grad_norm": 9.616310119628906,
      "learning_rate": 4.862862862862863e-06,
      "loss": 0.2411,
      "step": 51320
    },
    {
      "epoch": 154.14,
      "grad_norm": 3.902371883392334,
      "learning_rate": 4.861861861861862e-06,
      "loss": 0.1888,
      "step": 51330
    },
    {
      "epoch": 154.17,
      "grad_norm": 8.122920989990234,
      "learning_rate": 4.860860860860862e-06,
      "loss": 0.2375,
      "step": 51340
    },
    {
      "epoch": 154.2,
      "grad_norm": 9.019819259643555,
      "learning_rate": 4.8598598598598604e-06,
      "loss": 0.2582,
      "step": 51350
    },
    {
      "epoch": 154.23,
      "grad_norm": 10.810879707336426,
      "learning_rate": 4.858858858858859e-06,
      "loss": 0.1884,
      "step": 51360
    },
    {
      "epoch": 154.26,
      "grad_norm": 11.42614459991455,
      "learning_rate": 4.8578578578578585e-06,
      "loss": 0.2501,
      "step": 51370
    },
    {
      "epoch": 154.29,
      "grad_norm": 11.070305824279785,
      "learning_rate": 4.856856856856857e-06,
      "loss": 0.2006,
      "step": 51380
    },
    {
      "epoch": 154.32,
      "grad_norm": 11.003349304199219,
      "learning_rate": 4.855855855855856e-06,
      "loss": 0.2413,
      "step": 51390
    },
    {
      "epoch": 154.35,
      "grad_norm": 7.455862045288086,
      "learning_rate": 4.854854854854855e-06,
      "loss": 0.2311,
      "step": 51400
    },
    {
      "epoch": 154.38,
      "grad_norm": 4.766420841217041,
      "learning_rate": 4.8538538538538545e-06,
      "loss": 0.2417,
      "step": 51410
    },
    {
      "epoch": 154.41,
      "grad_norm": 7.618348121643066,
      "learning_rate": 4.852852852852853e-06,
      "loss": 0.2135,
      "step": 51420
    },
    {
      "epoch": 154.44,
      "grad_norm": 4.981983184814453,
      "learning_rate": 4.851851851851852e-06,
      "loss": 0.2252,
      "step": 51430
    },
    {
      "epoch": 154.47,
      "grad_norm": 6.795830726623535,
      "learning_rate": 4.850850850850851e-06,
      "loss": 0.2243,
      "step": 51440
    },
    {
      "epoch": 154.5,
      "grad_norm": 14.961501121520996,
      "learning_rate": 4.849849849849851e-06,
      "loss": 0.1806,
      "step": 51450
    },
    {
      "epoch": 154.53,
      "grad_norm": 20.594196319580078,
      "learning_rate": 4.848848848848849e-06,
      "loss": 0.2243,
      "step": 51460
    },
    {
      "epoch": 154.56,
      "grad_norm": 9.672661781311035,
      "learning_rate": 4.847847847847848e-06,
      "loss": 0.2606,
      "step": 51470
    },
    {
      "epoch": 154.59,
      "grad_norm": 25.354019165039062,
      "learning_rate": 4.846846846846847e-06,
      "loss": 0.2345,
      "step": 51480
    },
    {
      "epoch": 154.62,
      "grad_norm": 9.198265075683594,
      "learning_rate": 4.845845845845847e-06,
      "loss": 0.2601,
      "step": 51490
    },
    {
      "epoch": 154.65,
      "grad_norm": 9.846884727478027,
      "learning_rate": 4.844844844844845e-06,
      "loss": 0.2361,
      "step": 51500
    },
    {
      "epoch": 154.68,
      "grad_norm": 9.307990074157715,
      "learning_rate": 4.843843843843844e-06,
      "loss": 0.2155,
      "step": 51510
    },
    {
      "epoch": 154.71,
      "grad_norm": 21.168720245361328,
      "learning_rate": 4.842842842842843e-06,
      "loss": 0.2358,
      "step": 51520
    },
    {
      "epoch": 154.74,
      "grad_norm": 18.72102928161621,
      "learning_rate": 4.841841841841842e-06,
      "loss": 0.223,
      "step": 51530
    },
    {
      "epoch": 154.77,
      "grad_norm": 5.653415679931641,
      "learning_rate": 4.840840840840841e-06,
      "loss": 0.2086,
      "step": 51540
    },
    {
      "epoch": 154.8,
      "grad_norm": 8.401850700378418,
      "learning_rate": 4.83983983983984e-06,
      "loss": 0.2347,
      "step": 51550
    },
    {
      "epoch": 154.83,
      "grad_norm": 7.434954643249512,
      "learning_rate": 4.838838838838839e-06,
      "loss": 0.2368,
      "step": 51560
    },
    {
      "epoch": 154.86,
      "grad_norm": 13.095830917358398,
      "learning_rate": 4.837837837837838e-06,
      "loss": 0.2182,
      "step": 51570
    },
    {
      "epoch": 154.89,
      "grad_norm": 9.596685409545898,
      "learning_rate": 4.8368368368368374e-06,
      "loss": 0.2432,
      "step": 51580
    },
    {
      "epoch": 154.92,
      "grad_norm": 9.698973655700684,
      "learning_rate": 4.835835835835837e-06,
      "loss": 0.2174,
      "step": 51590
    },
    {
      "epoch": 154.95,
      "grad_norm": 5.622157096862793,
      "learning_rate": 4.8348348348348355e-06,
      "loss": 0.2281,
      "step": 51600
    },
    {
      "epoch": 154.98,
      "grad_norm": 6.662172317504883,
      "learning_rate": 4.833833833833834e-06,
      "loss": 0.2418,
      "step": 51610
    },
    {
      "epoch": 155.0,
      "eval_accuracy": 0.924,
      "eval_loss": 0.33134719729423523,
      "eval_runtime": 30.3803,
      "eval_samples_per_second": 329.16,
      "eval_steps_per_second": 1.317,
      "step": 51615
    },
    {
      "epoch": 155.02,
      "grad_norm": 11.345495223999023,
      "learning_rate": 4.8328328328328335e-06,
      "loss": 0.2242,
      "step": 51620
    },
    {
      "epoch": 155.05,
      "grad_norm": 5.380265235900879,
      "learning_rate": 4.831831831831832e-06,
      "loss": 0.2597,
      "step": 51630
    },
    {
      "epoch": 155.08,
      "grad_norm": 10.997798919677734,
      "learning_rate": 4.830830830830831e-06,
      "loss": 0.1962,
      "step": 51640
    },
    {
      "epoch": 155.11,
      "grad_norm": 8.056928634643555,
      "learning_rate": 4.82982982982983e-06,
      "loss": 0.241,
      "step": 51650
    },
    {
      "epoch": 155.14,
      "grad_norm": 10.545522689819336,
      "learning_rate": 4.82882882882883e-06,
      "loss": 0.246,
      "step": 51660
    },
    {
      "epoch": 155.17,
      "grad_norm": 9.412901878356934,
      "learning_rate": 4.827827827827828e-06,
      "loss": 0.2265,
      "step": 51670
    },
    {
      "epoch": 155.2,
      "grad_norm": 8.811488151550293,
      "learning_rate": 4.826826826826827e-06,
      "loss": 0.2635,
      "step": 51680
    },
    {
      "epoch": 155.23,
      "grad_norm": 6.839724540710449,
      "learning_rate": 4.825825825825826e-06,
      "loss": 0.1995,
      "step": 51690
    },
    {
      "epoch": 155.26,
      "grad_norm": 7.720097064971924,
      "learning_rate": 4.824824824824826e-06,
      "loss": 0.2098,
      "step": 51700
    },
    {
      "epoch": 155.29,
      "grad_norm": 8.505663871765137,
      "learning_rate": 4.823823823823824e-06,
      "loss": 0.2352,
      "step": 51710
    },
    {
      "epoch": 155.32,
      "grad_norm": 14.713737487792969,
      "learning_rate": 4.822822822822823e-06,
      "loss": 0.2581,
      "step": 51720
    },
    {
      "epoch": 155.35,
      "grad_norm": 9.532608985900879,
      "learning_rate": 4.821821821821822e-06,
      "loss": 0.2045,
      "step": 51730
    },
    {
      "epoch": 155.38,
      "grad_norm": 12.383424758911133,
      "learning_rate": 4.820820820820821e-06,
      "loss": 0.2644,
      "step": 51740
    },
    {
      "epoch": 155.41,
      "grad_norm": 11.944461822509766,
      "learning_rate": 4.81981981981982e-06,
      "loss": 0.237,
      "step": 51750
    },
    {
      "epoch": 155.44,
      "grad_norm": 5.512807846069336,
      "learning_rate": 4.818818818818819e-06,
      "loss": 0.2426,
      "step": 51760
    },
    {
      "epoch": 155.47,
      "grad_norm": 9.510177612304688,
      "learning_rate": 4.817817817817818e-06,
      "loss": 0.2158,
      "step": 51770
    },
    {
      "epoch": 155.5,
      "grad_norm": 7.541386127471924,
      "learning_rate": 4.816816816816817e-06,
      "loss": 0.247,
      "step": 51780
    },
    {
      "epoch": 155.53,
      "grad_norm": 10.485660552978516,
      "learning_rate": 4.815815815815816e-06,
      "loss": 0.2429,
      "step": 51790
    },
    {
      "epoch": 155.56,
      "grad_norm": 9.429244995117188,
      "learning_rate": 4.814814814814815e-06,
      "loss": 0.2308,
      "step": 51800
    },
    {
      "epoch": 155.59,
      "grad_norm": 7.828701496124268,
      "learning_rate": 4.8138138138138144e-06,
      "loss": 0.2345,
      "step": 51810
    },
    {
      "epoch": 155.62,
      "grad_norm": 16.525373458862305,
      "learning_rate": 4.812812812812813e-06,
      "loss": 0.1925,
      "step": 51820
    },
    {
      "epoch": 155.65,
      "grad_norm": 5.142930030822754,
      "learning_rate": 4.8118118118118125e-06,
      "loss": 0.2003,
      "step": 51830
    },
    {
      "epoch": 155.68,
      "grad_norm": 18.884838104248047,
      "learning_rate": 4.810810810810811e-06,
      "loss": 0.217,
      "step": 51840
    },
    {
      "epoch": 155.71,
      "grad_norm": 8.799847602844238,
      "learning_rate": 4.80980980980981e-06,
      "loss": 0.2452,
      "step": 51850
    },
    {
      "epoch": 155.74,
      "grad_norm": 12.023438453674316,
      "learning_rate": 4.808808808808809e-06,
      "loss": 0.206,
      "step": 51860
    },
    {
      "epoch": 155.77,
      "grad_norm": 9.900361061096191,
      "learning_rate": 4.8078078078078085e-06,
      "loss": 0.2328,
      "step": 51870
    },
    {
      "epoch": 155.8,
      "grad_norm": 14.788601875305176,
      "learning_rate": 4.806806806806807e-06,
      "loss": 0.2593,
      "step": 51880
    },
    {
      "epoch": 155.83,
      "grad_norm": 7.116164684295654,
      "learning_rate": 4.805805805805806e-06,
      "loss": 0.2035,
      "step": 51890
    },
    {
      "epoch": 155.86,
      "grad_norm": 7.8541035652160645,
      "learning_rate": 4.804804804804805e-06,
      "loss": 0.2146,
      "step": 51900
    },
    {
      "epoch": 155.89,
      "grad_norm": 5.374006271362305,
      "learning_rate": 4.803803803803805e-06,
      "loss": 0.2118,
      "step": 51910
    },
    {
      "epoch": 155.92,
      "grad_norm": 11.057071685791016,
      "learning_rate": 4.802802802802803e-06,
      "loss": 0.2517,
      "step": 51920
    },
    {
      "epoch": 155.95,
      "grad_norm": 5.812248229980469,
      "learning_rate": 4.801801801801802e-06,
      "loss": 0.2223,
      "step": 51930
    },
    {
      "epoch": 155.98,
      "grad_norm": 28.86939239501953,
      "learning_rate": 4.800800800800801e-06,
      "loss": 0.2254,
      "step": 51940
    },
    {
      "epoch": 156.0,
      "eval_accuracy": 0.9245,
      "eval_loss": 0.3326365649700165,
      "eval_runtime": 29.955,
      "eval_samples_per_second": 333.834,
      "eval_steps_per_second": 1.335,
      "step": 51948
    },
    {
      "epoch": 156.01,
      "grad_norm": 10.568303108215332,
      "learning_rate": 4.7997997997998e-06,
      "loss": 0.2126,
      "step": 51950
    },
    {
      "epoch": 156.04,
      "grad_norm": 9.335015296936035,
      "learning_rate": 4.798798798798799e-06,
      "loss": 0.2034,
      "step": 51960
    },
    {
      "epoch": 156.07,
      "grad_norm": 4.129494667053223,
      "learning_rate": 4.797797797797798e-06,
      "loss": 0.1969,
      "step": 51970
    },
    {
      "epoch": 156.1,
      "grad_norm": 14.394609451293945,
      "learning_rate": 4.796796796796797e-06,
      "loss": 0.212,
      "step": 51980
    },
    {
      "epoch": 156.13,
      "grad_norm": 6.003185749053955,
      "learning_rate": 4.795795795795796e-06,
      "loss": 0.2657,
      "step": 51990
    },
    {
      "epoch": 156.16,
      "grad_norm": 8.329994201660156,
      "learning_rate": 4.794794794794795e-06,
      "loss": 0.2213,
      "step": 52000
    },
    {
      "epoch": 156.19,
      "grad_norm": 6.2720136642456055,
      "learning_rate": 4.793793793793794e-06,
      "loss": 0.2036,
      "step": 52010
    },
    {
      "epoch": 156.22,
      "grad_norm": 7.718132972717285,
      "learning_rate": 4.792792792792793e-06,
      "loss": 0.2184,
      "step": 52020
    },
    {
      "epoch": 156.25,
      "grad_norm": 12.24686336517334,
      "learning_rate": 4.791791791791792e-06,
      "loss": 0.2208,
      "step": 52030
    },
    {
      "epoch": 156.28,
      "grad_norm": 7.959453105926514,
      "learning_rate": 4.7907907907907914e-06,
      "loss": 0.26,
      "step": 52040
    },
    {
      "epoch": 156.31,
      "grad_norm": 9.305211067199707,
      "learning_rate": 4.78978978978979e-06,
      "loss": 0.2254,
      "step": 52050
    },
    {
      "epoch": 156.34,
      "grad_norm": 5.315782070159912,
      "learning_rate": 4.7887887887887895e-06,
      "loss": 0.2318,
      "step": 52060
    },
    {
      "epoch": 156.37,
      "grad_norm": 19.112895965576172,
      "learning_rate": 4.787787787787788e-06,
      "loss": 0.2487,
      "step": 52070
    },
    {
      "epoch": 156.4,
      "grad_norm": 9.419197082519531,
      "learning_rate": 4.7867867867867875e-06,
      "loss": 0.2655,
      "step": 52080
    },
    {
      "epoch": 156.43,
      "grad_norm": 7.446198463439941,
      "learning_rate": 4.785785785785786e-06,
      "loss": 0.2231,
      "step": 52090
    },
    {
      "epoch": 156.46,
      "grad_norm": 8.584416389465332,
      "learning_rate": 4.784784784784785e-06,
      "loss": 0.2548,
      "step": 52100
    },
    {
      "epoch": 156.49,
      "grad_norm": 8.505463600158691,
      "learning_rate": 4.783783783783784e-06,
      "loss": 0.2101,
      "step": 52110
    },
    {
      "epoch": 156.52,
      "grad_norm": 8.393235206604004,
      "learning_rate": 4.7827827827827836e-06,
      "loss": 0.2229,
      "step": 52120
    },
    {
      "epoch": 156.55,
      "grad_norm": 9.171605110168457,
      "learning_rate": 4.781781781781782e-06,
      "loss": 0.2435,
      "step": 52130
    },
    {
      "epoch": 156.58,
      "grad_norm": 6.645887851715088,
      "learning_rate": 4.780780780780781e-06,
      "loss": 0.2375,
      "step": 52140
    },
    {
      "epoch": 156.61,
      "grad_norm": 12.583640098571777,
      "learning_rate": 4.77977977977978e-06,
      "loss": 0.2193,
      "step": 52150
    },
    {
      "epoch": 156.64,
      "grad_norm": 12.529520034790039,
      "learning_rate": 4.77877877877878e-06,
      "loss": 0.2071,
      "step": 52160
    },
    {
      "epoch": 156.67,
      "grad_norm": 9.641632080078125,
      "learning_rate": 4.777777777777778e-06,
      "loss": 0.2252,
      "step": 52170
    },
    {
      "epoch": 156.7,
      "grad_norm": 7.480172634124756,
      "learning_rate": 4.776776776776777e-06,
      "loss": 0.2343,
      "step": 52180
    },
    {
      "epoch": 156.73,
      "grad_norm": 8.160656929016113,
      "learning_rate": 4.775775775775776e-06,
      "loss": 0.2219,
      "step": 52190
    },
    {
      "epoch": 156.76,
      "grad_norm": 8.355504035949707,
      "learning_rate": 4.774774774774775e-06,
      "loss": 0.2093,
      "step": 52200
    },
    {
      "epoch": 156.79,
      "grad_norm": 16.198945999145508,
      "learning_rate": 4.773773773773774e-06,
      "loss": 0.2236,
      "step": 52210
    },
    {
      "epoch": 156.82,
      "grad_norm": 20.639619827270508,
      "learning_rate": 4.772772772772773e-06,
      "loss": 0.2222,
      "step": 52220
    },
    {
      "epoch": 156.85,
      "grad_norm": 7.156136989593506,
      "learning_rate": 4.771771771771772e-06,
      "loss": 0.2212,
      "step": 52230
    },
    {
      "epoch": 156.88,
      "grad_norm": 7.677603721618652,
      "learning_rate": 4.770770770770771e-06,
      "loss": 0.2299,
      "step": 52240
    },
    {
      "epoch": 156.91,
      "grad_norm": 8.574952125549316,
      "learning_rate": 4.76976976976977e-06,
      "loss": 0.2209,
      "step": 52250
    },
    {
      "epoch": 156.94,
      "grad_norm": 12.44437026977539,
      "learning_rate": 4.768768768768769e-06,
      "loss": 0.252,
      "step": 52260
    },
    {
      "epoch": 156.97,
      "grad_norm": 8.333195686340332,
      "learning_rate": 4.7677677677677684e-06,
      "loss": 0.2507,
      "step": 52270
    },
    {
      "epoch": 157.0,
      "grad_norm": 8.956460952758789,
      "learning_rate": 4.766766766766767e-06,
      "loss": 0.2549,
      "step": 52280
    },
    {
      "epoch": 157.0,
      "eval_accuracy": 0.924,
      "eval_loss": 0.33654868602752686,
      "eval_runtime": 30.4801,
      "eval_samples_per_second": 328.083,
      "eval_steps_per_second": 1.312,
      "step": 52281
    },
    {
      "epoch": 157.03,
      "grad_norm": 9.178203582763672,
      "learning_rate": 4.7657657657657665e-06,
      "loss": 0.1764,
      "step": 52290
    },
    {
      "epoch": 157.06,
      "grad_norm": 19.72076416015625,
      "learning_rate": 4.764764764764765e-06,
      "loss": 0.2064,
      "step": 52300
    },
    {
      "epoch": 157.09,
      "grad_norm": 21.992027282714844,
      "learning_rate": 4.763763763763764e-06,
      "loss": 0.2015,
      "step": 52310
    },
    {
      "epoch": 157.12,
      "grad_norm": 11.144847869873047,
      "learning_rate": 4.762762762762763e-06,
      "loss": 0.2802,
      "step": 52320
    },
    {
      "epoch": 157.15,
      "grad_norm": 6.325456619262695,
      "learning_rate": 4.7617617617617625e-06,
      "loss": 0.2031,
      "step": 52330
    },
    {
      "epoch": 157.18,
      "grad_norm": 14.852211952209473,
      "learning_rate": 4.760760760760761e-06,
      "loss": 0.2086,
      "step": 52340
    },
    {
      "epoch": 157.21,
      "grad_norm": 7.694889545440674,
      "learning_rate": 4.75975975975976e-06,
      "loss": 0.2327,
      "step": 52350
    },
    {
      "epoch": 157.24,
      "grad_norm": 10.468422889709473,
      "learning_rate": 4.758758758758759e-06,
      "loss": 0.2448,
      "step": 52360
    },
    {
      "epoch": 157.27,
      "grad_norm": 10.598589897155762,
      "learning_rate": 4.757757757757759e-06,
      "loss": 0.2424,
      "step": 52370
    },
    {
      "epoch": 157.3,
      "grad_norm": 10.585576057434082,
      "learning_rate": 4.756756756756757e-06,
      "loss": 0.218,
      "step": 52380
    },
    {
      "epoch": 157.33,
      "grad_norm": 5.100591659545898,
      "learning_rate": 4.755755755755756e-06,
      "loss": 0.207,
      "step": 52390
    },
    {
      "epoch": 157.36,
      "grad_norm": 8.878326416015625,
      "learning_rate": 4.754754754754755e-06,
      "loss": 0.2162,
      "step": 52400
    },
    {
      "epoch": 157.39,
      "grad_norm": 29.28949737548828,
      "learning_rate": 4.753753753753754e-06,
      "loss": 0.2333,
      "step": 52410
    },
    {
      "epoch": 157.42,
      "grad_norm": 6.981842041015625,
      "learning_rate": 4.752752752752753e-06,
      "loss": 0.2601,
      "step": 52420
    },
    {
      "epoch": 157.45,
      "grad_norm": 8.898372650146484,
      "learning_rate": 4.751751751751752e-06,
      "loss": 0.2527,
      "step": 52430
    },
    {
      "epoch": 157.48,
      "grad_norm": 12.406496047973633,
      "learning_rate": 4.750750750750751e-06,
      "loss": 0.2082,
      "step": 52440
    },
    {
      "epoch": 157.51,
      "grad_norm": 5.972841739654541,
      "learning_rate": 4.74974974974975e-06,
      "loss": 0.2773,
      "step": 52450
    },
    {
      "epoch": 157.54,
      "grad_norm": 9.638297080993652,
      "learning_rate": 4.748748748748749e-06,
      "loss": 0.2376,
      "step": 52460
    },
    {
      "epoch": 157.57,
      "grad_norm": 8.366703033447266,
      "learning_rate": 4.747747747747748e-06,
      "loss": 0.1965,
      "step": 52470
    },
    {
      "epoch": 157.6,
      "grad_norm": 9.455042839050293,
      "learning_rate": 4.746746746746747e-06,
      "loss": 0.2168,
      "step": 52480
    },
    {
      "epoch": 157.63,
      "grad_norm": 11.023990631103516,
      "learning_rate": 4.745745745745746e-06,
      "loss": 0.2144,
      "step": 52490
    },
    {
      "epoch": 157.66,
      "grad_norm": 7.698910713195801,
      "learning_rate": 4.7447447447447454e-06,
      "loss": 0.2346,
      "step": 52500
    },
    {
      "epoch": 157.69,
      "grad_norm": 9.206985473632812,
      "learning_rate": 4.743743743743744e-06,
      "loss": 0.2079,
      "step": 52510
    },
    {
      "epoch": 157.72,
      "grad_norm": 8.86409854888916,
      "learning_rate": 4.7427427427427435e-06,
      "loss": 0.2445,
      "step": 52520
    },
    {
      "epoch": 157.75,
      "grad_norm": 5.6387176513671875,
      "learning_rate": 4.741741741741742e-06,
      "loss": 0.221,
      "step": 52530
    },
    {
      "epoch": 157.78,
      "grad_norm": 7.967801094055176,
      "learning_rate": 4.7407407407407415e-06,
      "loss": 0.2611,
      "step": 52540
    },
    {
      "epoch": 157.81,
      "grad_norm": 12.387418746948242,
      "learning_rate": 4.73973973973974e-06,
      "loss": 0.276,
      "step": 52550
    },
    {
      "epoch": 157.84,
      "grad_norm": 6.598055362701416,
      "learning_rate": 4.738738738738739e-06,
      "loss": 0.2073,
      "step": 52560
    },
    {
      "epoch": 157.87,
      "grad_norm": 13.323147773742676,
      "learning_rate": 4.737737737737738e-06,
      "loss": 0.2215,
      "step": 52570
    },
    {
      "epoch": 157.9,
      "grad_norm": 9.873664855957031,
      "learning_rate": 4.7367367367367376e-06,
      "loss": 0.244,
      "step": 52580
    },
    {
      "epoch": 157.93,
      "grad_norm": 5.412178039550781,
      "learning_rate": 4.735735735735736e-06,
      "loss": 0.2476,
      "step": 52590
    },
    {
      "epoch": 157.96,
      "grad_norm": 9.275127410888672,
      "learning_rate": 4.734734734734735e-06,
      "loss": 0.2214,
      "step": 52600
    },
    {
      "epoch": 157.99,
      "grad_norm": 10.711506843566895,
      "learning_rate": 4.733733733733734e-06,
      "loss": 0.213,
      "step": 52610
    },
    {
      "epoch": 158.0,
      "eval_accuracy": 0.9257,
      "eval_loss": 0.32498699426651,
      "eval_runtime": 30.1616,
      "eval_samples_per_second": 331.547,
      "eval_steps_per_second": 1.326,
      "step": 52614
    },
    {
      "epoch": 158.02,
      "grad_norm": 6.636791706085205,
      "learning_rate": 4.732732732732734e-06,
      "loss": 0.2291,
      "step": 52620
    },
    {
      "epoch": 158.05,
      "grad_norm": 9.673050880432129,
      "learning_rate": 4.731731731731732e-06,
      "loss": 0.2398,
      "step": 52630
    },
    {
      "epoch": 158.08,
      "grad_norm": 8.08838939666748,
      "learning_rate": 4.730730730730731e-06,
      "loss": 0.2175,
      "step": 52640
    },
    {
      "epoch": 158.11,
      "grad_norm": 10.658343315124512,
      "learning_rate": 4.72972972972973e-06,
      "loss": 0.2036,
      "step": 52650
    },
    {
      "epoch": 158.14,
      "grad_norm": 9.893966674804688,
      "learning_rate": 4.728728728728729e-06,
      "loss": 0.2011,
      "step": 52660
    },
    {
      "epoch": 158.17,
      "grad_norm": 7.430913925170898,
      "learning_rate": 4.727727727727728e-06,
      "loss": 0.2422,
      "step": 52670
    },
    {
      "epoch": 158.2,
      "grad_norm": 6.359540939331055,
      "learning_rate": 4.726726726726727e-06,
      "loss": 0.2335,
      "step": 52680
    },
    {
      "epoch": 158.23,
      "grad_norm": 9.334575653076172,
      "learning_rate": 4.725725725725726e-06,
      "loss": 0.2387,
      "step": 52690
    },
    {
      "epoch": 158.26,
      "grad_norm": 7.012524127960205,
      "learning_rate": 4.724724724724725e-06,
      "loss": 0.2035,
      "step": 52700
    },
    {
      "epoch": 158.29,
      "grad_norm": 16.826290130615234,
      "learning_rate": 4.723723723723724e-06,
      "loss": 0.2502,
      "step": 52710
    },
    {
      "epoch": 158.32,
      "grad_norm": 8.164862632751465,
      "learning_rate": 4.722722722722723e-06,
      "loss": 0.2692,
      "step": 52720
    },
    {
      "epoch": 158.35,
      "grad_norm": 18.467369079589844,
      "learning_rate": 4.7217217217217224e-06,
      "loss": 0.2399,
      "step": 52730
    },
    {
      "epoch": 158.38,
      "grad_norm": 5.325833797454834,
      "learning_rate": 4.720720720720721e-06,
      "loss": 0.2146,
      "step": 52740
    },
    {
      "epoch": 158.41,
      "grad_norm": 9.523168563842773,
      "learning_rate": 4.7197197197197205e-06,
      "loss": 0.2376,
      "step": 52750
    },
    {
      "epoch": 158.44,
      "grad_norm": 7.524415969848633,
      "learning_rate": 4.718718718718719e-06,
      "loss": 0.2749,
      "step": 52760
    },
    {
      "epoch": 158.47,
      "grad_norm": 6.4482421875,
      "learning_rate": 4.717717717717718e-06,
      "loss": 0.2427,
      "step": 52770
    },
    {
      "epoch": 158.5,
      "grad_norm": 4.475306987762451,
      "learning_rate": 4.716716716716717e-06,
      "loss": 0.2593,
      "step": 52780
    },
    {
      "epoch": 158.53,
      "grad_norm": 9.684050559997559,
      "learning_rate": 4.7157157157157165e-06,
      "loss": 0.1975,
      "step": 52790
    },
    {
      "epoch": 158.56,
      "grad_norm": 15.817584037780762,
      "learning_rate": 4.714714714714715e-06,
      "loss": 0.2353,
      "step": 52800
    },
    {
      "epoch": 158.59,
      "grad_norm": 9.213953018188477,
      "learning_rate": 4.713713713713714e-06,
      "loss": 0.2739,
      "step": 52810
    },
    {
      "epoch": 158.62,
      "grad_norm": 20.491413116455078,
      "learning_rate": 4.712712712712713e-06,
      "loss": 0.1996,
      "step": 52820
    },
    {
      "epoch": 158.65,
      "grad_norm": 7.7160258293151855,
      "learning_rate": 4.711711711711713e-06,
      "loss": 0.1939,
      "step": 52830
    },
    {
      "epoch": 158.68,
      "grad_norm": 8.457836151123047,
      "learning_rate": 4.710710710710711e-06,
      "loss": 0.1978,
      "step": 52840
    },
    {
      "epoch": 158.71,
      "grad_norm": 11.336189270019531,
      "learning_rate": 4.70970970970971e-06,
      "loss": 0.2547,
      "step": 52850
    },
    {
      "epoch": 158.74,
      "grad_norm": 7.6278557777404785,
      "learning_rate": 4.708708708708709e-06,
      "loss": 0.2407,
      "step": 52860
    },
    {
      "epoch": 158.77,
      "grad_norm": 11.82660961151123,
      "learning_rate": 4.707707707707708e-06,
      "loss": 0.22,
      "step": 52870
    },
    {
      "epoch": 158.8,
      "grad_norm": 13.511394500732422,
      "learning_rate": 4.7067067067067064e-06,
      "loss": 0.2049,
      "step": 52880
    },
    {
      "epoch": 158.83,
      "grad_norm": 5.261289596557617,
      "learning_rate": 4.705705705705706e-06,
      "loss": 0.2453,
      "step": 52890
    },
    {
      "epoch": 158.86,
      "grad_norm": 12.96481990814209,
      "learning_rate": 4.704704704704705e-06,
      "loss": 0.2014,
      "step": 52900
    },
    {
      "epoch": 158.89,
      "grad_norm": 7.2584309577941895,
      "learning_rate": 4.703703703703704e-06,
      "loss": 0.2207,
      "step": 52910
    },
    {
      "epoch": 158.92,
      "grad_norm": 14.114686012268066,
      "learning_rate": 4.702702702702703e-06,
      "loss": 0.2363,
      "step": 52920
    },
    {
      "epoch": 158.95,
      "grad_norm": 6.5042500495910645,
      "learning_rate": 4.701701701701702e-06,
      "loss": 0.2034,
      "step": 52930
    },
    {
      "epoch": 158.98,
      "grad_norm": 10.360466957092285,
      "learning_rate": 4.700700700700701e-06,
      "loss": 0.2178,
      "step": 52940
    },
    {
      "epoch": 159.0,
      "eval_accuracy": 0.9252,
      "eval_loss": 0.32592347264289856,
      "eval_runtime": 30.3325,
      "eval_samples_per_second": 329.679,
      "eval_steps_per_second": 1.319,
      "step": 52947
    },
    {
      "epoch": 159.01,
      "grad_norm": 9.38613224029541,
      "learning_rate": 4.6996996996997e-06,
      "loss": 0.2144,
      "step": 52950
    },
    {
      "epoch": 159.04,
      "grad_norm": 3.326748847961426,
      "learning_rate": 4.698698698698699e-06,
      "loss": 0.2059,
      "step": 52960
    },
    {
      "epoch": 159.07,
      "grad_norm": 8.556920051574707,
      "learning_rate": 4.697697697697698e-06,
      "loss": 0.2441,
      "step": 52970
    },
    {
      "epoch": 159.1,
      "grad_norm": 5.185380935668945,
      "learning_rate": 4.696696696696697e-06,
      "loss": 0.2343,
      "step": 52980
    },
    {
      "epoch": 159.13,
      "grad_norm": 10.799127578735352,
      "learning_rate": 4.695695695695696e-06,
      "loss": 0.2702,
      "step": 52990
    },
    {
      "epoch": 159.16,
      "grad_norm": 8.982497215270996,
      "learning_rate": 4.6946946946946955e-06,
      "loss": 0.2133,
      "step": 53000
    },
    {
      "epoch": 159.19,
      "grad_norm": 11.458314895629883,
      "learning_rate": 4.693693693693694e-06,
      "loss": 0.2339,
      "step": 53010
    },
    {
      "epoch": 159.22,
      "grad_norm": 8.909994125366211,
      "learning_rate": 4.692692692692693e-06,
      "loss": 0.2292,
      "step": 53020
    },
    {
      "epoch": 159.25,
      "grad_norm": 9.721319198608398,
      "learning_rate": 4.691691691691692e-06,
      "loss": 0.2016,
      "step": 53030
    },
    {
      "epoch": 159.28,
      "grad_norm": 6.746898174285889,
      "learning_rate": 4.6906906906906916e-06,
      "loss": 0.2826,
      "step": 53040
    },
    {
      "epoch": 159.31,
      "grad_norm": 8.615961074829102,
      "learning_rate": 4.68968968968969e-06,
      "loss": 0.2201,
      "step": 53050
    },
    {
      "epoch": 159.34,
      "grad_norm": 129.66629028320312,
      "learning_rate": 4.688688688688689e-06,
      "loss": 0.2484,
      "step": 53060
    },
    {
      "epoch": 159.37,
      "grad_norm": 17.674978256225586,
      "learning_rate": 4.687687687687688e-06,
      "loss": 0.233,
      "step": 53070
    },
    {
      "epoch": 159.4,
      "grad_norm": 5.240177631378174,
      "learning_rate": 4.686686686686688e-06,
      "loss": 0.2213,
      "step": 53080
    },
    {
      "epoch": 159.43,
      "grad_norm": 16.37641716003418,
      "learning_rate": 4.685685685685686e-06,
      "loss": 0.2484,
      "step": 53090
    },
    {
      "epoch": 159.46,
      "grad_norm": 6.744995594024658,
      "learning_rate": 4.684684684684685e-06,
      "loss": 0.2078,
      "step": 53100
    },
    {
      "epoch": 159.49,
      "grad_norm": 4.318368911743164,
      "learning_rate": 4.683683683683684e-06,
      "loss": 0.2371,
      "step": 53110
    },
    {
      "epoch": 159.52,
      "grad_norm": 9.29955005645752,
      "learning_rate": 4.682682682682683e-06,
      "loss": 0.2401,
      "step": 53120
    },
    {
      "epoch": 159.55,
      "grad_norm": 8.086505889892578,
      "learning_rate": 4.6816816816816815e-06,
      "loss": 0.2791,
      "step": 53130
    },
    {
      "epoch": 159.58,
      "grad_norm": 5.041549205780029,
      "learning_rate": 4.680680680680681e-06,
      "loss": 0.177,
      "step": 53140
    },
    {
      "epoch": 159.61,
      "grad_norm": 6.62031888961792,
      "learning_rate": 4.67967967967968e-06,
      "loss": 0.232,
      "step": 53150
    },
    {
      "epoch": 159.64,
      "grad_norm": 7.937234878540039,
      "learning_rate": 4.678678678678679e-06,
      "loss": 0.248,
      "step": 53160
    },
    {
      "epoch": 159.67,
      "grad_norm": 8.86257266998291,
      "learning_rate": 4.677677677677678e-06,
      "loss": 0.2529,
      "step": 53170
    },
    {
      "epoch": 159.7,
      "grad_norm": 21.545177459716797,
      "learning_rate": 4.676676676676677e-06,
      "loss": 0.2495,
      "step": 53180
    },
    {
      "epoch": 159.73,
      "grad_norm": 6.944525718688965,
      "learning_rate": 4.675675675675676e-06,
      "loss": 0.2052,
      "step": 53190
    },
    {
      "epoch": 159.76,
      "grad_norm": 6.034377574920654,
      "learning_rate": 4.674674674674675e-06,
      "loss": 0.2029,
      "step": 53200
    },
    {
      "epoch": 159.79,
      "grad_norm": 8.637722969055176,
      "learning_rate": 4.6736736736736745e-06,
      "loss": 0.2675,
      "step": 53210
    },
    {
      "epoch": 159.82,
      "grad_norm": 12.653735160827637,
      "learning_rate": 4.672672672672673e-06,
      "loss": 0.1962,
      "step": 53220
    },
    {
      "epoch": 159.85,
      "grad_norm": 7.345870494842529,
      "learning_rate": 4.671671671671672e-06,
      "loss": 0.2057,
      "step": 53230
    },
    {
      "epoch": 159.88,
      "grad_norm": 7.774090766906738,
      "learning_rate": 4.670670670670671e-06,
      "loss": 0.2917,
      "step": 53240
    },
    {
      "epoch": 159.91,
      "grad_norm": 8.27519702911377,
      "learning_rate": 4.6696696696696705e-06,
      "loss": 0.25,
      "step": 53250
    },
    {
      "epoch": 159.94,
      "grad_norm": 4.695452690124512,
      "learning_rate": 4.668668668668669e-06,
      "loss": 0.1916,
      "step": 53260
    },
    {
      "epoch": 159.97,
      "grad_norm": 7.426046371459961,
      "learning_rate": 4.667667667667668e-06,
      "loss": 0.2072,
      "step": 53270
    },
    {
      "epoch": 160.0,
      "grad_norm": 0.026458166539669037,
      "learning_rate": 4.666666666666667e-06,
      "loss": 0.2127,
      "step": 53280
    },
    {
      "epoch": 160.0,
      "eval_accuracy": 0.9241,
      "eval_loss": 0.32941845059394836,
      "eval_runtime": 29.9836,
      "eval_samples_per_second": 333.516,
      "eval_steps_per_second": 1.334,
      "step": 53280
    },
    {
      "epoch": 160.03,
      "grad_norm": 8.392930030822754,
      "learning_rate": 4.665665665665667e-06,
      "loss": 0.2392,
      "step": 53290
    },
    {
      "epoch": 160.06,
      "grad_norm": 14.991545677185059,
      "learning_rate": 4.664664664664665e-06,
      "loss": 0.2372,
      "step": 53300
    },
    {
      "epoch": 160.09,
      "grad_norm": 24.135440826416016,
      "learning_rate": 4.663663663663664e-06,
      "loss": 0.2211,
      "step": 53310
    },
    {
      "epoch": 160.12,
      "grad_norm": 11.29979133605957,
      "learning_rate": 4.662662662662663e-06,
      "loss": 0.2117,
      "step": 53320
    },
    {
      "epoch": 160.15,
      "grad_norm": 7.912114143371582,
      "learning_rate": 4.661661661661662e-06,
      "loss": 0.2278,
      "step": 53330
    },
    {
      "epoch": 160.18,
      "grad_norm": 8.492486953735352,
      "learning_rate": 4.6606606606606604e-06,
      "loss": 0.2388,
      "step": 53340
    },
    {
      "epoch": 160.21,
      "grad_norm": 18.46411895751953,
      "learning_rate": 4.65965965965966e-06,
      "loss": 0.2512,
      "step": 53350
    },
    {
      "epoch": 160.24,
      "grad_norm": 10.792285919189453,
      "learning_rate": 4.658658658658659e-06,
      "loss": 0.2007,
      "step": 53360
    },
    {
      "epoch": 160.27,
      "grad_norm": 14.06588363647461,
      "learning_rate": 4.657657657657658e-06,
      "loss": 0.2233,
      "step": 53370
    },
    {
      "epoch": 160.3,
      "grad_norm": 7.73682165145874,
      "learning_rate": 4.6566566566566565e-06,
      "loss": 0.2185,
      "step": 53380
    },
    {
      "epoch": 160.33,
      "grad_norm": 14.46812629699707,
      "learning_rate": 4.655655655655656e-06,
      "loss": 0.2404,
      "step": 53390
    },
    {
      "epoch": 160.36,
      "grad_norm": 15.967412948608398,
      "learning_rate": 4.654654654654655e-06,
      "loss": 0.215,
      "step": 53400
    },
    {
      "epoch": 160.39,
      "grad_norm": 8.37017822265625,
      "learning_rate": 4.653653653653654e-06,
      "loss": 0.2483,
      "step": 53410
    },
    {
      "epoch": 160.42,
      "grad_norm": 11.099723815917969,
      "learning_rate": 4.6526526526526526e-06,
      "loss": 0.2316,
      "step": 53420
    },
    {
      "epoch": 160.45,
      "grad_norm": 10.787940979003906,
      "learning_rate": 4.651651651651652e-06,
      "loss": 0.2233,
      "step": 53430
    },
    {
      "epoch": 160.48,
      "grad_norm": 11.968506813049316,
      "learning_rate": 4.650650650650651e-06,
      "loss": 0.183,
      "step": 53440
    },
    {
      "epoch": 160.51,
      "grad_norm": 6.652988433837891,
      "learning_rate": 4.64964964964965e-06,
      "loss": 0.2544,
      "step": 53450
    },
    {
      "epoch": 160.54,
      "grad_norm": 12.25078296661377,
      "learning_rate": 4.6486486486486495e-06,
      "loss": 0.2525,
      "step": 53460
    },
    {
      "epoch": 160.57,
      "grad_norm": 17.557573318481445,
      "learning_rate": 4.647647647647648e-06,
      "loss": 0.2377,
      "step": 53470
    },
    {
      "epoch": 160.6,
      "grad_norm": 6.98940896987915,
      "learning_rate": 4.646646646646647e-06,
      "loss": 0.1769,
      "step": 53480
    },
    {
      "epoch": 160.63,
      "grad_norm": 5.710747241973877,
      "learning_rate": 4.645645645645646e-06,
      "loss": 0.2205,
      "step": 53490
    },
    {
      "epoch": 160.66,
      "grad_norm": 8.788612365722656,
      "learning_rate": 4.6446446446446456e-06,
      "loss": 0.1964,
      "step": 53500
    },
    {
      "epoch": 160.69,
      "grad_norm": 7.891353130340576,
      "learning_rate": 4.643643643643644e-06,
      "loss": 0.2301,
      "step": 53510
    },
    {
      "epoch": 160.72,
      "grad_norm": 8.274057388305664,
      "learning_rate": 4.642642642642643e-06,
      "loss": 0.2193,
      "step": 53520
    },
    {
      "epoch": 160.75,
      "grad_norm": 12.055450439453125,
      "learning_rate": 4.641641641641642e-06,
      "loss": 0.2258,
      "step": 53530
    },
    {
      "epoch": 160.78,
      "grad_norm": 10.021851539611816,
      "learning_rate": 4.640640640640642e-06,
      "loss": 0.2141,
      "step": 53540
    },
    {
      "epoch": 160.81,
      "grad_norm": 8.061833381652832,
      "learning_rate": 4.63963963963964e-06,
      "loss": 0.2538,
      "step": 53550
    },
    {
      "epoch": 160.84,
      "grad_norm": 7.97246789932251,
      "learning_rate": 4.638638638638639e-06,
      "loss": 0.2253,
      "step": 53560
    },
    {
      "epoch": 160.87,
      "grad_norm": 15.664198875427246,
      "learning_rate": 4.637637637637638e-06,
      "loss": 0.2123,
      "step": 53570
    },
    {
      "epoch": 160.9,
      "grad_norm": 92.16514587402344,
      "learning_rate": 4.636636636636637e-06,
      "loss": 0.2909,
      "step": 53580
    },
    {
      "epoch": 160.93,
      "grad_norm": 7.430404186248779,
      "learning_rate": 4.6356356356356355e-06,
      "loss": 0.2016,
      "step": 53590
    },
    {
      "epoch": 160.96,
      "grad_norm": 10.438864707946777,
      "learning_rate": 4.634634634634635e-06,
      "loss": 0.2071,
      "step": 53600
    },
    {
      "epoch": 160.99,
      "grad_norm": 6.1503400802612305,
      "learning_rate": 4.633633633633634e-06,
      "loss": 0.2063,
      "step": 53610
    },
    {
      "epoch": 161.0,
      "eval_accuracy": 0.9239,
      "eval_loss": 0.33008116483688354,
      "eval_runtime": 30.3183,
      "eval_samples_per_second": 329.834,
      "eval_steps_per_second": 1.319,
      "step": 53613
    },
    {
      "epoch": 161.02,
      "grad_norm": 14.912995338439941,
      "learning_rate": 4.632632632632633e-06,
      "loss": 0.1675,
      "step": 53620
    },
    {
      "epoch": 161.05,
      "grad_norm": 8.249415397644043,
      "learning_rate": 4.6316316316316315e-06,
      "loss": 0.2086,
      "step": 53630
    },
    {
      "epoch": 161.08,
      "grad_norm": 12.46451473236084,
      "learning_rate": 4.630630630630631e-06,
      "loss": 0.2806,
      "step": 53640
    },
    {
      "epoch": 161.11,
      "grad_norm": 5.655143737792969,
      "learning_rate": 4.62962962962963e-06,
      "loss": 0.254,
      "step": 53650
    },
    {
      "epoch": 161.14,
      "grad_norm": 7.3935980796813965,
      "learning_rate": 4.628628628628629e-06,
      "loss": 0.2237,
      "step": 53660
    },
    {
      "epoch": 161.17,
      "grad_norm": 16.488887786865234,
      "learning_rate": 4.627627627627628e-06,
      "loss": 0.224,
      "step": 53670
    },
    {
      "epoch": 161.2,
      "grad_norm": 8.503061294555664,
      "learning_rate": 4.626626626626627e-06,
      "loss": 0.1905,
      "step": 53680
    },
    {
      "epoch": 161.23,
      "grad_norm": 11.190000534057617,
      "learning_rate": 4.625625625625626e-06,
      "loss": 0.2406,
      "step": 53690
    },
    {
      "epoch": 161.26,
      "grad_norm": 7.0630974769592285,
      "learning_rate": 4.624624624624625e-06,
      "loss": 0.2614,
      "step": 53700
    },
    {
      "epoch": 161.29,
      "grad_norm": 6.3123779296875,
      "learning_rate": 4.6236236236236245e-06,
      "loss": 0.2258,
      "step": 53710
    },
    {
      "epoch": 161.32,
      "grad_norm": 11.281715393066406,
      "learning_rate": 4.622622622622623e-06,
      "loss": 0.2336,
      "step": 53720
    },
    {
      "epoch": 161.35,
      "grad_norm": 14.42401123046875,
      "learning_rate": 4.621621621621622e-06,
      "loss": 0.2136,
      "step": 53730
    },
    {
      "epoch": 161.38,
      "grad_norm": 10.103219032287598,
      "learning_rate": 4.620620620620621e-06,
      "loss": 0.2441,
      "step": 53740
    },
    {
      "epoch": 161.41,
      "grad_norm": 6.633398532867432,
      "learning_rate": 4.619619619619621e-06,
      "loss": 0.1781,
      "step": 53750
    },
    {
      "epoch": 161.44,
      "grad_norm": 7.498027324676514,
      "learning_rate": 4.618618618618619e-06,
      "loss": 0.2271,
      "step": 53760
    },
    {
      "epoch": 161.47,
      "grad_norm": 10.038934707641602,
      "learning_rate": 4.617617617617618e-06,
      "loss": 0.2199,
      "step": 53770
    },
    {
      "epoch": 161.5,
      "grad_norm": 8.569863319396973,
      "learning_rate": 4.616616616616617e-06,
      "loss": 0.229,
      "step": 53780
    },
    {
      "epoch": 161.53,
      "grad_norm": 8.690765380859375,
      "learning_rate": 4.615615615615616e-06,
      "loss": 0.2425,
      "step": 53790
    },
    {
      "epoch": 161.56,
      "grad_norm": 6.183032512664795,
      "learning_rate": 4.614614614614614e-06,
      "loss": 0.2405,
      "step": 53800
    },
    {
      "epoch": 161.59,
      "grad_norm": 10.475062370300293,
      "learning_rate": 4.613613613613614e-06,
      "loss": 0.2318,
      "step": 53810
    },
    {
      "epoch": 161.62,
      "grad_norm": 17.61800193786621,
      "learning_rate": 4.612612612612613e-06,
      "loss": 0.2274,
      "step": 53820
    },
    {
      "epoch": 161.65,
      "grad_norm": 6.997801780700684,
      "learning_rate": 4.611611611611612e-06,
      "loss": 0.2216,
      "step": 53830
    },
    {
      "epoch": 161.68,
      "grad_norm": 11.070558547973633,
      "learning_rate": 4.6106106106106105e-06,
      "loss": 0.2141,
      "step": 53840
    },
    {
      "epoch": 161.71,
      "grad_norm": 8.441399574279785,
      "learning_rate": 4.60960960960961e-06,
      "loss": 0.233,
      "step": 53850
    },
    {
      "epoch": 161.74,
      "grad_norm": 20.18665885925293,
      "learning_rate": 4.608608608608609e-06,
      "loss": 0.1755,
      "step": 53860
    },
    {
      "epoch": 161.77,
      "grad_norm": 10.719461441040039,
      "learning_rate": 4.607607607607608e-06,
      "loss": 0.2332,
      "step": 53870
    },
    {
      "epoch": 161.8,
      "grad_norm": 19.808568954467773,
      "learning_rate": 4.6066066066066066e-06,
      "loss": 0.2136,
      "step": 53880
    },
    {
      "epoch": 161.83,
      "grad_norm": 7.565125465393066,
      "learning_rate": 4.605605605605606e-06,
      "loss": 0.2648,
      "step": 53890
    },
    {
      "epoch": 161.86,
      "grad_norm": 14.4301176071167,
      "learning_rate": 4.604604604604605e-06,
      "loss": 0.2525,
      "step": 53900
    },
    {
      "epoch": 161.89,
      "grad_norm": 10.929415702819824,
      "learning_rate": 4.603603603603604e-06,
      "loss": 0.2208,
      "step": 53910
    },
    {
      "epoch": 161.92,
      "grad_norm": 3.7004024982452393,
      "learning_rate": 4.602602602602603e-06,
      "loss": 0.1962,
      "step": 53920
    },
    {
      "epoch": 161.95,
      "grad_norm": 6.828344345092773,
      "learning_rate": 4.601601601601602e-06,
      "loss": 0.216,
      "step": 53930
    },
    {
      "epoch": 161.98,
      "grad_norm": 10.260102272033691,
      "learning_rate": 4.600600600600601e-06,
      "loss": 0.2401,
      "step": 53940
    },
    {
      "epoch": 162.0,
      "eval_accuracy": 0.9247,
      "eval_loss": 0.33063414692878723,
      "eval_runtime": 29.9296,
      "eval_samples_per_second": 334.117,
      "eval_steps_per_second": 1.336,
      "step": 53946
    },
    {
      "epoch": 162.01,
      "grad_norm": 8.206123352050781,
      "learning_rate": 4.5995995995996e-06,
      "loss": 0.2667,
      "step": 53950
    },
    {
      "epoch": 162.04,
      "grad_norm": 6.101869106292725,
      "learning_rate": 4.5985985985985996e-06,
      "loss": 0.2199,
      "step": 53960
    },
    {
      "epoch": 162.07,
      "grad_norm": 10.512721061706543,
      "learning_rate": 4.597597597597598e-06,
      "loss": 0.2294,
      "step": 53970
    },
    {
      "epoch": 162.1,
      "grad_norm": 5.545969486236572,
      "learning_rate": 4.596596596596597e-06,
      "loss": 0.2116,
      "step": 53980
    },
    {
      "epoch": 162.13,
      "grad_norm": 8.23900318145752,
      "learning_rate": 4.595595595595596e-06,
      "loss": 0.2513,
      "step": 53990
    },
    {
      "epoch": 162.16,
      "grad_norm": 8.523061752319336,
      "learning_rate": 4.594594594594596e-06,
      "loss": 0.2031,
      "step": 54000
    },
    {
      "epoch": 162.19,
      "grad_norm": 7.60551118850708,
      "learning_rate": 4.593593593593593e-06,
      "loss": 0.2316,
      "step": 54010
    },
    {
      "epoch": 162.22,
      "grad_norm": 9.037618637084961,
      "learning_rate": 4.592592592592593e-06,
      "loss": 0.2124,
      "step": 54020
    },
    {
      "epoch": 162.25,
      "grad_norm": 6.6602044105529785,
      "learning_rate": 4.591591591591592e-06,
      "loss": 0.1915,
      "step": 54030
    },
    {
      "epoch": 162.28,
      "grad_norm": 16.6350154876709,
      "learning_rate": 4.590590590590591e-06,
      "loss": 0.2157,
      "step": 54040
    },
    {
      "epoch": 162.31,
      "grad_norm": 9.656782150268555,
      "learning_rate": 4.5895895895895895e-06,
      "loss": 0.243,
      "step": 54050
    },
    {
      "epoch": 162.34,
      "grad_norm": 12.040166854858398,
      "learning_rate": 4.588588588588589e-06,
      "loss": 0.1913,
      "step": 54060
    },
    {
      "epoch": 162.37,
      "grad_norm": 10.18789291381836,
      "learning_rate": 4.587587587587588e-06,
      "loss": 0.2304,
      "step": 54070
    },
    {
      "epoch": 162.4,
      "grad_norm": 9.689175605773926,
      "learning_rate": 4.586586586586587e-06,
      "loss": 0.2432,
      "step": 54080
    },
    {
      "epoch": 162.43,
      "grad_norm": 6.890012741088867,
      "learning_rate": 4.5855855855855855e-06,
      "loss": 0.2278,
      "step": 54090
    },
    {
      "epoch": 162.46,
      "grad_norm": 13.864195823669434,
      "learning_rate": 4.584584584584585e-06,
      "loss": 0.28,
      "step": 54100
    },
    {
      "epoch": 162.49,
      "grad_norm": 9.04413890838623,
      "learning_rate": 4.583583583583584e-06,
      "loss": 0.2035,
      "step": 54110
    },
    {
      "epoch": 162.52,
      "grad_norm": 7.523566722869873,
      "learning_rate": 4.582582582582583e-06,
      "loss": 0.1947,
      "step": 54120
    },
    {
      "epoch": 162.55,
      "grad_norm": 13.783957481384277,
      "learning_rate": 4.581581581581582e-06,
      "loss": 0.2144,
      "step": 54130
    },
    {
      "epoch": 162.58,
      "grad_norm": 11.207551002502441,
      "learning_rate": 4.580580580580581e-06,
      "loss": 0.2192,
      "step": 54140
    },
    {
      "epoch": 162.61,
      "grad_norm": 9.570781707763672,
      "learning_rate": 4.57957957957958e-06,
      "loss": 0.2214,
      "step": 54150
    },
    {
      "epoch": 162.64,
      "grad_norm": 7.458194732666016,
      "learning_rate": 4.578578578578579e-06,
      "loss": 0.2364,
      "step": 54160
    },
    {
      "epoch": 162.67,
      "grad_norm": 8.747142791748047,
      "learning_rate": 4.577577577577578e-06,
      "loss": 0.2195,
      "step": 54170
    },
    {
      "epoch": 162.7,
      "grad_norm": 11.234743118286133,
      "learning_rate": 4.576576576576577e-06,
      "loss": 0.2641,
      "step": 54180
    },
    {
      "epoch": 162.73,
      "grad_norm": 10.67456340789795,
      "learning_rate": 4.575575575575576e-06,
      "loss": 0.2427,
      "step": 54190
    },
    {
      "epoch": 162.76,
      "grad_norm": 13.939278602600098,
      "learning_rate": 4.574574574574575e-06,
      "loss": 0.2162,
      "step": 54200
    },
    {
      "epoch": 162.79,
      "grad_norm": 9.393550872802734,
      "learning_rate": 4.573573573573574e-06,
      "loss": 0.2207,
      "step": 54210
    },
    {
      "epoch": 162.82,
      "grad_norm": 13.784997940063477,
      "learning_rate": 4.572572572572573e-06,
      "loss": 0.2235,
      "step": 54220
    },
    {
      "epoch": 162.85,
      "grad_norm": 5.1563005447387695,
      "learning_rate": 4.571571571571572e-06,
      "loss": 0.1822,
      "step": 54230
    },
    {
      "epoch": 162.88,
      "grad_norm": 8.547918319702148,
      "learning_rate": 4.570570570570571e-06,
      "loss": 0.2778,
      "step": 54240
    },
    {
      "epoch": 162.91,
      "grad_norm": 11.325383186340332,
      "learning_rate": 4.56956956956957e-06,
      "loss": 0.2294,
      "step": 54250
    },
    {
      "epoch": 162.94,
      "grad_norm": 8.60693645477295,
      "learning_rate": 4.568568568568568e-06,
      "loss": 0.2384,
      "step": 54260
    },
    {
      "epoch": 162.97,
      "grad_norm": 8.002211570739746,
      "learning_rate": 4.567567567567568e-06,
      "loss": 0.2198,
      "step": 54270
    },
    {
      "epoch": 163.0,
      "eval_accuracy": 0.9226,
      "eval_loss": 0.33226117491722107,
      "eval_runtime": 30.1741,
      "eval_samples_per_second": 331.41,
      "eval_steps_per_second": 1.326,
      "step": 54279
    },
    {
      "epoch": 163.0,
      "grad_norm": 7.746329307556152,
      "learning_rate": 4.566566566566567e-06,
      "loss": 0.2036,
      "step": 54280
    },
    {
      "epoch": 163.03,
      "grad_norm": 13.083803176879883,
      "learning_rate": 4.565565565565566e-06,
      "loss": 0.2285,
      "step": 54290
    },
    {
      "epoch": 163.06,
      "grad_norm": 8.196017265319824,
      "learning_rate": 4.5645645645645645e-06,
      "loss": 0.2468,
      "step": 54300
    },
    {
      "epoch": 163.09,
      "grad_norm": 8.04019546508789,
      "learning_rate": 4.563563563563564e-06,
      "loss": 0.2378,
      "step": 54310
    },
    {
      "epoch": 163.12,
      "grad_norm": 10.761398315429688,
      "learning_rate": 4.562562562562563e-06,
      "loss": 0.222,
      "step": 54320
    },
    {
      "epoch": 163.15,
      "grad_norm": 8.819869041442871,
      "learning_rate": 4.561561561561562e-06,
      "loss": 0.253,
      "step": 54330
    },
    {
      "epoch": 163.18,
      "grad_norm": 7.956119060516357,
      "learning_rate": 4.5605605605605606e-06,
      "loss": 0.234,
      "step": 54340
    },
    {
      "epoch": 163.21,
      "grad_norm": 7.8786091804504395,
      "learning_rate": 4.55955955955956e-06,
      "loss": 0.2468,
      "step": 54350
    },
    {
      "epoch": 163.24,
      "grad_norm": 15.148385047912598,
      "learning_rate": 4.558558558558559e-06,
      "loss": 0.2296,
      "step": 54360
    },
    {
      "epoch": 163.27,
      "grad_norm": 17.641992568969727,
      "learning_rate": 4.557557557557558e-06,
      "loss": 0.189,
      "step": 54370
    },
    {
      "epoch": 163.3,
      "grad_norm": 8.355741500854492,
      "learning_rate": 4.556556556556557e-06,
      "loss": 0.1902,
      "step": 54380
    },
    {
      "epoch": 163.33,
      "grad_norm": 4.9927167892456055,
      "learning_rate": 4.555555555555556e-06,
      "loss": 0.2,
      "step": 54390
    },
    {
      "epoch": 163.36,
      "grad_norm": 7.3431220054626465,
      "learning_rate": 4.554554554554555e-06,
      "loss": 0.238,
      "step": 54400
    },
    {
      "epoch": 163.39,
      "grad_norm": 9.206598281860352,
      "learning_rate": 4.553553553553554e-06,
      "loss": 0.2319,
      "step": 54410
    },
    {
      "epoch": 163.42,
      "grad_norm": 12.194259643554688,
      "learning_rate": 4.552552552552553e-06,
      "loss": 0.2593,
      "step": 54420
    },
    {
      "epoch": 163.45,
      "grad_norm": 7.9242167472839355,
      "learning_rate": 4.551551551551552e-06,
      "loss": 0.2119,
      "step": 54430
    },
    {
      "epoch": 163.48,
      "grad_norm": 8.548921585083008,
      "learning_rate": 4.550550550550551e-06,
      "loss": 0.2183,
      "step": 54440
    },
    {
      "epoch": 163.51,
      "grad_norm": 11.263555526733398,
      "learning_rate": 4.54954954954955e-06,
      "loss": 0.2225,
      "step": 54450
    },
    {
      "epoch": 163.54,
      "grad_norm": 9.60326862335205,
      "learning_rate": 4.548548548548549e-06,
      "loss": 0.2332,
      "step": 54460
    },
    {
      "epoch": 163.57,
      "grad_norm": 14.083166122436523,
      "learning_rate": 4.547547547547547e-06,
      "loss": 0.2626,
      "step": 54470
    },
    {
      "epoch": 163.6,
      "grad_norm": 5.647646903991699,
      "learning_rate": 4.546546546546547e-06,
      "loss": 0.188,
      "step": 54480
    },
    {
      "epoch": 163.63,
      "grad_norm": 5.143772125244141,
      "learning_rate": 4.545545545545546e-06,
      "loss": 0.1921,
      "step": 54490
    },
    {
      "epoch": 163.66,
      "grad_norm": 11.514725685119629,
      "learning_rate": 4.544544544544545e-06,
      "loss": 0.183,
      "step": 54500
    },
    {
      "epoch": 163.69,
      "grad_norm": 12.917338371276855,
      "learning_rate": 4.5435435435435435e-06,
      "loss": 0.2336,
      "step": 54510
    },
    {
      "epoch": 163.72,
      "grad_norm": 10.116469383239746,
      "learning_rate": 4.542542542542543e-06,
      "loss": 0.2083,
      "step": 54520
    },
    {
      "epoch": 163.75,
      "grad_norm": 12.83767318725586,
      "learning_rate": 4.541541541541542e-06,
      "loss": 0.2025,
      "step": 54530
    },
    {
      "epoch": 163.78,
      "grad_norm": 6.273693084716797,
      "learning_rate": 4.540540540540541e-06,
      "loss": 0.2184,
      "step": 54540
    },
    {
      "epoch": 163.81,
      "grad_norm": 10.288824081420898,
      "learning_rate": 4.5395395395395395e-06,
      "loss": 0.2217,
      "step": 54550
    },
    {
      "epoch": 163.84,
      "grad_norm": 6.753376007080078,
      "learning_rate": 4.538538538538539e-06,
      "loss": 0.2219,
      "step": 54560
    },
    {
      "epoch": 163.87,
      "grad_norm": 5.303664684295654,
      "learning_rate": 4.537537537537538e-06,
      "loss": 0.1852,
      "step": 54570
    },
    {
      "epoch": 163.9,
      "grad_norm": 8.734065055847168,
      "learning_rate": 4.536536536536537e-06,
      "loss": 0.2517,
      "step": 54580
    },
    {
      "epoch": 163.93,
      "grad_norm": 9.335838317871094,
      "learning_rate": 4.535535535535536e-06,
      "loss": 0.2981,
      "step": 54590
    },
    {
      "epoch": 163.96,
      "grad_norm": 6.206470966339111,
      "learning_rate": 4.534534534534535e-06,
      "loss": 0.2352,
      "step": 54600
    },
    {
      "epoch": 163.99,
      "grad_norm": 7.264355659484863,
      "learning_rate": 4.533533533533534e-06,
      "loss": 0.2642,
      "step": 54610
    },
    {
      "epoch": 164.0,
      "eval_accuracy": 0.9239,
      "eval_loss": 0.33240100741386414,
      "eval_runtime": 29.8797,
      "eval_samples_per_second": 334.675,
      "eval_steps_per_second": 1.339,
      "step": 54612
    },
    {
      "epoch": 164.02,
      "grad_norm": 10.803627967834473,
      "learning_rate": 4.532532532532533e-06,
      "loss": 0.2438,
      "step": 54620
    },
    {
      "epoch": 164.05,
      "grad_norm": 9.215641021728516,
      "learning_rate": 4.531531531531532e-06,
      "loss": 0.2543,
      "step": 54630
    },
    {
      "epoch": 164.08,
      "grad_norm": 5.424944877624512,
      "learning_rate": 4.530530530530531e-06,
      "loss": 0.2427,
      "step": 54640
    },
    {
      "epoch": 164.11,
      "grad_norm": 10.076470375061035,
      "learning_rate": 4.52952952952953e-06,
      "loss": 0.2233,
      "step": 54650
    },
    {
      "epoch": 164.14,
      "grad_norm": 7.147024154663086,
      "learning_rate": 4.528528528528529e-06,
      "loss": 0.218,
      "step": 54660
    },
    {
      "epoch": 164.17,
      "grad_norm": 5.5970964431762695,
      "learning_rate": 4.527527527527528e-06,
      "loss": 0.2046,
      "step": 54670
    },
    {
      "epoch": 164.2,
      "grad_norm": 10.613640785217285,
      "learning_rate": 4.526526526526527e-06,
      "loss": 0.2156,
      "step": 54680
    },
    {
      "epoch": 164.23,
      "grad_norm": 10.301372528076172,
      "learning_rate": 4.525525525525526e-06,
      "loss": 0.2077,
      "step": 54690
    },
    {
      "epoch": 164.26,
      "grad_norm": 6.461187362670898,
      "learning_rate": 4.524524524524525e-06,
      "loss": 0.2148,
      "step": 54700
    },
    {
      "epoch": 164.29,
      "grad_norm": 9.407197952270508,
      "learning_rate": 4.523523523523524e-06,
      "loss": 0.2062,
      "step": 54710
    },
    {
      "epoch": 164.32,
      "grad_norm": 5.884110450744629,
      "learning_rate": 4.522522522522522e-06,
      "loss": 0.2114,
      "step": 54720
    },
    {
      "epoch": 164.35,
      "grad_norm": 8.790129661560059,
      "learning_rate": 4.521521521521522e-06,
      "loss": 0.2344,
      "step": 54730
    },
    {
      "epoch": 164.38,
      "grad_norm": 6.080642223358154,
      "learning_rate": 4.520520520520521e-06,
      "loss": 0.2677,
      "step": 54740
    },
    {
      "epoch": 164.41,
      "grad_norm": 7.045116901397705,
      "learning_rate": 4.51951951951952e-06,
      "loss": 0.2009,
      "step": 54750
    },
    {
      "epoch": 164.44,
      "grad_norm": 10.57974624633789,
      "learning_rate": 4.5185185185185185e-06,
      "loss": 0.2345,
      "step": 54760
    },
    {
      "epoch": 164.47,
      "grad_norm": 11.177694320678711,
      "learning_rate": 4.517517517517518e-06,
      "loss": 0.2124,
      "step": 54770
    },
    {
      "epoch": 164.5,
      "grad_norm": 8.120506286621094,
      "learning_rate": 4.516516516516517e-06,
      "loss": 0.2525,
      "step": 54780
    },
    {
      "epoch": 164.53,
      "grad_norm": 12.399333000183105,
      "learning_rate": 4.515515515515516e-06,
      "loss": 0.1892,
      "step": 54790
    },
    {
      "epoch": 164.56,
      "grad_norm": 13.276013374328613,
      "learning_rate": 4.5145145145145146e-06,
      "loss": 0.2331,
      "step": 54800
    },
    {
      "epoch": 164.59,
      "grad_norm": 7.409654140472412,
      "learning_rate": 4.513513513513514e-06,
      "loss": 0.2664,
      "step": 54810
    },
    {
      "epoch": 164.62,
      "grad_norm": 9.604588508605957,
      "learning_rate": 4.512512512512513e-06,
      "loss": 0.2183,
      "step": 54820
    },
    {
      "epoch": 164.65,
      "grad_norm": 14.118377685546875,
      "learning_rate": 4.511511511511512e-06,
      "loss": 0.1813,
      "step": 54830
    },
    {
      "epoch": 164.68,
      "grad_norm": 8.285717010498047,
      "learning_rate": 4.510510510510511e-06,
      "loss": 0.1859,
      "step": 54840
    },
    {
      "epoch": 164.71,
      "grad_norm": 12.778569221496582,
      "learning_rate": 4.50950950950951e-06,
      "loss": 0.2513,
      "step": 54850
    },
    {
      "epoch": 164.74,
      "grad_norm": 11.10633373260498,
      "learning_rate": 4.508508508508509e-06,
      "loss": 0.2212,
      "step": 54860
    },
    {
      "epoch": 164.77,
      "grad_norm": 8.277276039123535,
      "learning_rate": 4.507507507507508e-06,
      "loss": 0.2333,
      "step": 54870
    },
    {
      "epoch": 164.8,
      "grad_norm": 6.997796058654785,
      "learning_rate": 4.506506506506507e-06,
      "loss": 0.2561,
      "step": 54880
    },
    {
      "epoch": 164.83,
      "grad_norm": 9.887724876403809,
      "learning_rate": 4.505505505505506e-06,
      "loss": 0.2438,
      "step": 54890
    },
    {
      "epoch": 164.86,
      "grad_norm": 10.1692476272583,
      "learning_rate": 4.504504504504505e-06,
      "loss": 0.2497,
      "step": 54900
    },
    {
      "epoch": 164.89,
      "grad_norm": 6.974490642547607,
      "learning_rate": 4.503503503503504e-06,
      "loss": 0.2671,
      "step": 54910
    },
    {
      "epoch": 164.92,
      "grad_norm": 20.422269821166992,
      "learning_rate": 4.502502502502503e-06,
      "loss": 0.2418,
      "step": 54920
    },
    {
      "epoch": 164.95,
      "grad_norm": 7.2772016525268555,
      "learning_rate": 4.501501501501501e-06,
      "loss": 0.236,
      "step": 54930
    },
    {
      "epoch": 164.98,
      "grad_norm": 8.569696426391602,
      "learning_rate": 4.500500500500501e-06,
      "loss": 0.2477,
      "step": 54940
    },
    {
      "epoch": 165.0,
      "eval_accuracy": 0.9238,
      "eval_loss": 0.331449031829834,
      "eval_runtime": 30.996,
      "eval_samples_per_second": 322.623,
      "eval_steps_per_second": 1.29,
      "step": 54945
    },
    {
      "epoch": 165.02,
      "grad_norm": 7.242735385894775,
      "learning_rate": 4.4994994994995e-06,
      "loss": 0.1523,
      "step": 54950
    },
    {
      "epoch": 165.05,
      "grad_norm": 13.290122032165527,
      "learning_rate": 4.498498498498499e-06,
      "loss": 0.2693,
      "step": 54960
    },
    {
      "epoch": 165.08,
      "grad_norm": 6.605894565582275,
      "learning_rate": 4.4974974974974974e-06,
      "loss": 0.2509,
      "step": 54970
    },
    {
      "epoch": 165.11,
      "grad_norm": 9.756126403808594,
      "learning_rate": 4.496496496496497e-06,
      "loss": 0.2348,
      "step": 54980
    },
    {
      "epoch": 165.14,
      "grad_norm": 5.319377422332764,
      "learning_rate": 4.495495495495496e-06,
      "loss": 0.2502,
      "step": 54990
    },
    {
      "epoch": 165.17,
      "grad_norm": 6.73790979385376,
      "learning_rate": 4.494494494494495e-06,
      "loss": 0.2246,
      "step": 55000
    },
    {
      "epoch": 165.2,
      "grad_norm": 9.544591903686523,
      "learning_rate": 4.4934934934934935e-06,
      "loss": 0.2675,
      "step": 55010
    },
    {
      "epoch": 165.23,
      "grad_norm": 7.7104172706604,
      "learning_rate": 4.492492492492493e-06,
      "loss": 0.2594,
      "step": 55020
    },
    {
      "epoch": 165.26,
      "grad_norm": 11.659339904785156,
      "learning_rate": 4.491491491491492e-06,
      "loss": 0.2571,
      "step": 55030
    },
    {
      "epoch": 165.29,
      "grad_norm": 6.093270301818848,
      "learning_rate": 4.490490490490491e-06,
      "loss": 0.2577,
      "step": 55040
    },
    {
      "epoch": 165.32,
      "grad_norm": 7.178973197937012,
      "learning_rate": 4.48948948948949e-06,
      "loss": 0.2264,
      "step": 55050
    },
    {
      "epoch": 165.35,
      "grad_norm": 9.118023872375488,
      "learning_rate": 4.488488488488489e-06,
      "loss": 0.2349,
      "step": 55060
    },
    {
      "epoch": 165.38,
      "grad_norm": 13.65462589263916,
      "learning_rate": 4.487487487487488e-06,
      "loss": 0.195,
      "step": 55070
    },
    {
      "epoch": 165.41,
      "grad_norm": 8.139679908752441,
      "learning_rate": 4.486486486486487e-06,
      "loss": 0.2112,
      "step": 55080
    },
    {
      "epoch": 165.44,
      "grad_norm": 8.250259399414062,
      "learning_rate": 4.485485485485486e-06,
      "loss": 0.2001,
      "step": 55090
    },
    {
      "epoch": 165.47,
      "grad_norm": 7.8776397705078125,
      "learning_rate": 4.484484484484485e-06,
      "loss": 0.2292,
      "step": 55100
    },
    {
      "epoch": 165.5,
      "grad_norm": 6.745391845703125,
      "learning_rate": 4.483483483483484e-06,
      "loss": 0.2167,
      "step": 55110
    },
    {
      "epoch": 165.53,
      "grad_norm": 8.784013748168945,
      "learning_rate": 4.482482482482483e-06,
      "loss": 0.2202,
      "step": 55120
    },
    {
      "epoch": 165.56,
      "grad_norm": 9.502866744995117,
      "learning_rate": 4.481481481481482e-06,
      "loss": 0.2184,
      "step": 55130
    },
    {
      "epoch": 165.59,
      "grad_norm": 7.275116443634033,
      "learning_rate": 4.480480480480481e-06,
      "loss": 0.2464,
      "step": 55140
    },
    {
      "epoch": 165.62,
      "grad_norm": 6.574884414672852,
      "learning_rate": 4.47947947947948e-06,
      "loss": 0.224,
      "step": 55150
    },
    {
      "epoch": 165.65,
      "grad_norm": 9.785810470581055,
      "learning_rate": 4.478478478478479e-06,
      "loss": 0.2272,
      "step": 55160
    },
    {
      "epoch": 165.68,
      "grad_norm": 9.137442588806152,
      "learning_rate": 4.477477477477478e-06,
      "loss": 0.211,
      "step": 55170
    },
    {
      "epoch": 165.71,
      "grad_norm": 51.40827178955078,
      "learning_rate": 4.476476476476476e-06,
      "loss": 0.22,
      "step": 55180
    },
    {
      "epoch": 165.74,
      "grad_norm": 9.388598442077637,
      "learning_rate": 4.475475475475476e-06,
      "loss": 0.1838,
      "step": 55190
    },
    {
      "epoch": 165.77,
      "grad_norm": 8.431360244750977,
      "learning_rate": 4.474474474474475e-06,
      "loss": 0.2237,
      "step": 55200
    },
    {
      "epoch": 165.8,
      "grad_norm": 5.80774450302124,
      "learning_rate": 4.473473473473474e-06,
      "loss": 0.2227,
      "step": 55210
    },
    {
      "epoch": 165.83,
      "grad_norm": 6.214824199676514,
      "learning_rate": 4.4724724724724725e-06,
      "loss": 0.212,
      "step": 55220
    },
    {
      "epoch": 165.86,
      "grad_norm": 6.487475395202637,
      "learning_rate": 4.471471471471472e-06,
      "loss": 0.2785,
      "step": 55230
    },
    {
      "epoch": 165.89,
      "grad_norm": 13.325371742248535,
      "learning_rate": 4.470470470470471e-06,
      "loss": 0.2096,
      "step": 55240
    },
    {
      "epoch": 165.92,
      "grad_norm": 6.346006393432617,
      "learning_rate": 4.46946946946947e-06,
      "loss": 0.1964,
      "step": 55250
    },
    {
      "epoch": 165.95,
      "grad_norm": 11.393712043762207,
      "learning_rate": 4.4684684684684686e-06,
      "loss": 0.2332,
      "step": 55260
    },
    {
      "epoch": 165.98,
      "grad_norm": 5.452152729034424,
      "learning_rate": 4.467467467467468e-06,
      "loss": 0.1936,
      "step": 55270
    },
    {
      "epoch": 166.0,
      "eval_accuracy": 0.9233,
      "eval_loss": 0.33244433999061584,
      "eval_runtime": 30.3975,
      "eval_samples_per_second": 328.975,
      "eval_steps_per_second": 1.316,
      "step": 55278
    },
    {
      "epoch": 166.01,
      "grad_norm": 9.989035606384277,
      "learning_rate": 4.466466466466467e-06,
      "loss": 0.2062,
      "step": 55280
    },
    {
      "epoch": 166.04,
      "grad_norm": 7.959853172302246,
      "learning_rate": 4.465465465465465e-06,
      "loss": 0.2317,
      "step": 55290
    },
    {
      "epoch": 166.07,
      "grad_norm": 16.409549713134766,
      "learning_rate": 4.464464464464465e-06,
      "loss": 0.2547,
      "step": 55300
    },
    {
      "epoch": 166.1,
      "grad_norm": 15.112076759338379,
      "learning_rate": 4.463463463463464e-06,
      "loss": 0.2342,
      "step": 55310
    },
    {
      "epoch": 166.13,
      "grad_norm": 5.104234218597412,
      "learning_rate": 4.462462462462463e-06,
      "loss": 0.2028,
      "step": 55320
    },
    {
      "epoch": 166.16,
      "grad_norm": 7.790355205535889,
      "learning_rate": 4.461461461461462e-06,
      "loss": 0.2395,
      "step": 55330
    },
    {
      "epoch": 166.19,
      "grad_norm": 8.20337200164795,
      "learning_rate": 4.460460460460461e-06,
      "loss": 0.2142,
      "step": 55340
    },
    {
      "epoch": 166.22,
      "grad_norm": 7.796027660369873,
      "learning_rate": 4.45945945945946e-06,
      "loss": 0.2109,
      "step": 55350
    },
    {
      "epoch": 166.25,
      "grad_norm": 6.597646236419678,
      "learning_rate": 4.458458458458459e-06,
      "loss": 0.1973,
      "step": 55360
    },
    {
      "epoch": 166.28,
      "grad_norm": 40.9026985168457,
      "learning_rate": 4.457457457457458e-06,
      "loss": 0.255,
      "step": 55370
    },
    {
      "epoch": 166.31,
      "grad_norm": 10.929301261901855,
      "learning_rate": 4.456456456456457e-06,
      "loss": 0.2078,
      "step": 55380
    },
    {
      "epoch": 166.34,
      "grad_norm": 6.582366943359375,
      "learning_rate": 4.455455455455455e-06,
      "loss": 0.2477,
      "step": 55390
    },
    {
      "epoch": 166.37,
      "grad_norm": 14.138811111450195,
      "learning_rate": 4.454454454454455e-06,
      "loss": 0.2699,
      "step": 55400
    },
    {
      "epoch": 166.4,
      "grad_norm": 10.416831016540527,
      "learning_rate": 4.453453453453454e-06,
      "loss": 0.2427,
      "step": 55410
    },
    {
      "epoch": 166.43,
      "grad_norm": 13.27324390411377,
      "learning_rate": 4.452452452452453e-06,
      "loss": 0.2223,
      "step": 55420
    },
    {
      "epoch": 166.46,
      "grad_norm": 12.502567291259766,
      "learning_rate": 4.4514514514514514e-06,
      "loss": 0.216,
      "step": 55430
    },
    {
      "epoch": 166.49,
      "grad_norm": 28.578332901000977,
      "learning_rate": 4.450450450450451e-06,
      "loss": 0.2399,
      "step": 55440
    },
    {
      "epoch": 166.52,
      "grad_norm": 13.876229286193848,
      "learning_rate": 4.44944944944945e-06,
      "loss": 0.192,
      "step": 55450
    },
    {
      "epoch": 166.55,
      "grad_norm": 8.698898315429688,
      "learning_rate": 4.448448448448449e-06,
      "loss": 0.2401,
      "step": 55460
    },
    {
      "epoch": 166.58,
      "grad_norm": 6.726069450378418,
      "learning_rate": 4.4474474474474475e-06,
      "loss": 0.1925,
      "step": 55470
    },
    {
      "epoch": 166.61,
      "grad_norm": 4.6574320793151855,
      "learning_rate": 4.446446446446447e-06,
      "loss": 0.2174,
      "step": 55480
    },
    {
      "epoch": 166.64,
      "grad_norm": 10.131753921508789,
      "learning_rate": 4.445445445445446e-06,
      "loss": 0.2424,
      "step": 55490
    },
    {
      "epoch": 166.67,
      "grad_norm": 8.28680419921875,
      "learning_rate": 4.444444444444444e-06,
      "loss": 0.2457,
      "step": 55500
    },
    {
      "epoch": 166.7,
      "grad_norm": 8.269224166870117,
      "learning_rate": 4.443443443443444e-06,
      "loss": 0.2014,
      "step": 55510
    },
    {
      "epoch": 166.73,
      "grad_norm": 8.145212173461914,
      "learning_rate": 4.442442442442443e-06,
      "loss": 0.1874,
      "step": 55520
    },
    {
      "epoch": 166.76,
      "grad_norm": 10.034188270568848,
      "learning_rate": 4.441441441441442e-06,
      "loss": 0.2367,
      "step": 55530
    },
    {
      "epoch": 166.79,
      "grad_norm": 6.682107448577881,
      "learning_rate": 4.44044044044044e-06,
      "loss": 0.207,
      "step": 55540
    },
    {
      "epoch": 166.82,
      "grad_norm": 7.585057735443115,
      "learning_rate": 4.43943943943944e-06,
      "loss": 0.1787,
      "step": 55550
    },
    {
      "epoch": 166.85,
      "grad_norm": 11.205612182617188,
      "learning_rate": 4.438438438438439e-06,
      "loss": 0.2413,
      "step": 55560
    },
    {
      "epoch": 166.88,
      "grad_norm": 9.674182891845703,
      "learning_rate": 4.437437437437438e-06,
      "loss": 0.2152,
      "step": 55570
    },
    {
      "epoch": 166.91,
      "grad_norm": 9.12603759765625,
      "learning_rate": 4.436436436436437e-06,
      "loss": 0.2322,
      "step": 55580
    },
    {
      "epoch": 166.94,
      "grad_norm": 8.362848281860352,
      "learning_rate": 4.435435435435436e-06,
      "loss": 0.2782,
      "step": 55590
    },
    {
      "epoch": 166.97,
      "grad_norm": 9.425893783569336,
      "learning_rate": 4.434434434434435e-06,
      "loss": 0.2004,
      "step": 55600
    },
    {
      "epoch": 167.0,
      "grad_norm": 8.628999710083008,
      "learning_rate": 4.433433433433434e-06,
      "loss": 0.2151,
      "step": 55610
    },
    {
      "epoch": 167.0,
      "eval_accuracy": 0.9229,
      "eval_loss": 0.3355526328086853,
      "eval_runtime": 30.4697,
      "eval_samples_per_second": 328.195,
      "eval_steps_per_second": 1.313,
      "step": 55611
    },
    {
      "epoch": 167.03,
      "grad_norm": 5.478910446166992,
      "learning_rate": 4.432432432432433e-06,
      "loss": 0.2028,
      "step": 55620
    },
    {
      "epoch": 167.06,
      "grad_norm": 28.119827270507812,
      "learning_rate": 4.431431431431432e-06,
      "loss": 0.2501,
      "step": 55630
    },
    {
      "epoch": 167.09,
      "grad_norm": 12.776832580566406,
      "learning_rate": 4.43043043043043e-06,
      "loss": 0.2409,
      "step": 55640
    },
    {
      "epoch": 167.12,
      "grad_norm": 8.859868049621582,
      "learning_rate": 4.42942942942943e-06,
      "loss": 0.2619,
      "step": 55650
    },
    {
      "epoch": 167.15,
      "grad_norm": 38.61552810668945,
      "learning_rate": 4.428428428428429e-06,
      "loss": 0.1937,
      "step": 55660
    },
    {
      "epoch": 167.18,
      "grad_norm": 9.353055000305176,
      "learning_rate": 4.427427427427428e-06,
      "loss": 0.2695,
      "step": 55670
    },
    {
      "epoch": 167.21,
      "grad_norm": 8.449651718139648,
      "learning_rate": 4.4264264264264265e-06,
      "loss": 0.2235,
      "step": 55680
    },
    {
      "epoch": 167.24,
      "grad_norm": 5.7258219718933105,
      "learning_rate": 4.425425425425426e-06,
      "loss": 0.1903,
      "step": 55690
    },
    {
      "epoch": 167.27,
      "grad_norm": 10.39810848236084,
      "learning_rate": 4.424424424424425e-06,
      "loss": 0.2109,
      "step": 55700
    },
    {
      "epoch": 167.3,
      "grad_norm": 10.712234497070312,
      "learning_rate": 4.423423423423424e-06,
      "loss": 0.2182,
      "step": 55710
    },
    {
      "epoch": 167.33,
      "grad_norm": 11.314194679260254,
      "learning_rate": 4.4224224224224226e-06,
      "loss": 0.193,
      "step": 55720
    },
    {
      "epoch": 167.36,
      "grad_norm": 13.908848762512207,
      "learning_rate": 4.421421421421422e-06,
      "loss": 0.2536,
      "step": 55730
    },
    {
      "epoch": 167.39,
      "grad_norm": 8.19444751739502,
      "learning_rate": 4.420420420420421e-06,
      "loss": 0.2352,
      "step": 55740
    },
    {
      "epoch": 167.42,
      "grad_norm": 8.349532127380371,
      "learning_rate": 4.419419419419419e-06,
      "loss": 0.178,
      "step": 55750
    },
    {
      "epoch": 167.45,
      "grad_norm": 8.674473762512207,
      "learning_rate": 4.418418418418419e-06,
      "loss": 0.2057,
      "step": 55760
    },
    {
      "epoch": 167.48,
      "grad_norm": 12.32136058807373,
      "learning_rate": 4.417417417417418e-06,
      "loss": 0.206,
      "step": 55770
    },
    {
      "epoch": 167.51,
      "grad_norm": 20.167131423950195,
      "learning_rate": 4.416416416416417e-06,
      "loss": 0.2173,
      "step": 55780
    },
    {
      "epoch": 167.54,
      "grad_norm": 9.758892059326172,
      "learning_rate": 4.415415415415415e-06,
      "loss": 0.1841,
      "step": 55790
    },
    {
      "epoch": 167.57,
      "grad_norm": 9.53329086303711,
      "learning_rate": 4.414414414414415e-06,
      "loss": 0.2313,
      "step": 55800
    },
    {
      "epoch": 167.6,
      "grad_norm": 7.99965238571167,
      "learning_rate": 4.413413413413414e-06,
      "loss": 0.2008,
      "step": 55810
    },
    {
      "epoch": 167.63,
      "grad_norm": 9.25546932220459,
      "learning_rate": 4.412412412412413e-06,
      "loss": 0.2251,
      "step": 55820
    },
    {
      "epoch": 167.66,
      "grad_norm": 7.730353832244873,
      "learning_rate": 4.411411411411412e-06,
      "loss": 0.2635,
      "step": 55830
    },
    {
      "epoch": 167.69,
      "grad_norm": 9.383145332336426,
      "learning_rate": 4.410410410410411e-06,
      "loss": 0.202,
      "step": 55840
    },
    {
      "epoch": 167.72,
      "grad_norm": 7.822163105010986,
      "learning_rate": 4.409409409409409e-06,
      "loss": 0.1769,
      "step": 55850
    },
    {
      "epoch": 167.75,
      "grad_norm": 8.482521057128906,
      "learning_rate": 4.408408408408409e-06,
      "loss": 0.2257,
      "step": 55860
    },
    {
      "epoch": 167.78,
      "grad_norm": 7.568439483642578,
      "learning_rate": 4.407407407407408e-06,
      "loss": 0.2512,
      "step": 55870
    },
    {
      "epoch": 167.81,
      "grad_norm": 7.044400215148926,
      "learning_rate": 4.406406406406407e-06,
      "loss": 0.227,
      "step": 55880
    },
    {
      "epoch": 167.84,
      "grad_norm": 8.567204475402832,
      "learning_rate": 4.4054054054054054e-06,
      "loss": 0.2292,
      "step": 55890
    },
    {
      "epoch": 167.87,
      "grad_norm": 12.517962455749512,
      "learning_rate": 4.404404404404405e-06,
      "loss": 0.2332,
      "step": 55900
    },
    {
      "epoch": 167.9,
      "grad_norm": 9.252352714538574,
      "learning_rate": 4.403403403403404e-06,
      "loss": 0.2558,
      "step": 55910
    },
    {
      "epoch": 167.93,
      "grad_norm": 21.831830978393555,
      "learning_rate": 4.402402402402403e-06,
      "loss": 0.2412,
      "step": 55920
    },
    {
      "epoch": 167.96,
      "grad_norm": 12.067607879638672,
      "learning_rate": 4.4014014014014015e-06,
      "loss": 0.2491,
      "step": 55930
    },
    {
      "epoch": 167.99,
      "grad_norm": 8.172059059143066,
      "learning_rate": 4.400400400400401e-06,
      "loss": 0.2049,
      "step": 55940
    },
    {
      "epoch": 168.0,
      "eval_accuracy": 0.9214,
      "eval_loss": 0.3354950249195099,
      "eval_runtime": 30.0641,
      "eval_samples_per_second": 332.623,
      "eval_steps_per_second": 1.33,
      "step": 55944
    },
    {
      "epoch": 168.02,
      "grad_norm": 8.37796688079834,
      "learning_rate": 4.3993993993993996e-06,
      "loss": 0.2286,
      "step": 55950
    },
    {
      "epoch": 168.05,
      "grad_norm": 7.130244255065918,
      "learning_rate": 4.398398398398398e-06,
      "loss": 0.2214,
      "step": 55960
    },
    {
      "epoch": 168.08,
      "grad_norm": 7.086636066436768,
      "learning_rate": 4.397397397397398e-06,
      "loss": 0.2282,
      "step": 55970
    },
    {
      "epoch": 168.11,
      "grad_norm": 9.761493682861328,
      "learning_rate": 4.396396396396397e-06,
      "loss": 0.2257,
      "step": 55980
    },
    {
      "epoch": 168.14,
      "grad_norm": 28.345571517944336,
      "learning_rate": 4.395395395395396e-06,
      "loss": 0.234,
      "step": 55990
    },
    {
      "epoch": 168.17,
      "grad_norm": 9.198378562927246,
      "learning_rate": 4.394394394394394e-06,
      "loss": 0.2055,
      "step": 56000
    },
    {
      "epoch": 168.2,
      "grad_norm": 14.422659873962402,
      "learning_rate": 4.393393393393394e-06,
      "loss": 0.2305,
      "step": 56010
    },
    {
      "epoch": 168.23,
      "grad_norm": 4.256718635559082,
      "learning_rate": 4.392392392392393e-06,
      "loss": 0.2782,
      "step": 56020
    },
    {
      "epoch": 168.26,
      "grad_norm": 16.245046615600586,
      "learning_rate": 4.391391391391392e-06,
      "loss": 0.2139,
      "step": 56030
    },
    {
      "epoch": 168.29,
      "grad_norm": 12.839461326599121,
      "learning_rate": 4.39039039039039e-06,
      "loss": 0.2584,
      "step": 56040
    },
    {
      "epoch": 168.32,
      "grad_norm": 8.36087703704834,
      "learning_rate": 4.38938938938939e-06,
      "loss": 0.2165,
      "step": 56050
    },
    {
      "epoch": 168.35,
      "grad_norm": 9.190874099731445,
      "learning_rate": 4.388388388388389e-06,
      "loss": 0.2353,
      "step": 56060
    },
    {
      "epoch": 168.38,
      "grad_norm": 7.542410373687744,
      "learning_rate": 4.387387387387388e-06,
      "loss": 0.224,
      "step": 56070
    },
    {
      "epoch": 168.41,
      "grad_norm": 8.880455017089844,
      "learning_rate": 4.386386386386386e-06,
      "loss": 0.2265,
      "step": 56080
    },
    {
      "epoch": 168.44,
      "grad_norm": 9.60269546508789,
      "learning_rate": 4.385385385385386e-06,
      "loss": 0.2018,
      "step": 56090
    },
    {
      "epoch": 168.47,
      "grad_norm": 9.635449409484863,
      "learning_rate": 4.384384384384384e-06,
      "loss": 0.1966,
      "step": 56100
    },
    {
      "epoch": 168.5,
      "grad_norm": 5.2141218185424805,
      "learning_rate": 4.383383383383384e-06,
      "loss": 0.2113,
      "step": 56110
    },
    {
      "epoch": 168.53,
      "grad_norm": 5.809586048126221,
      "learning_rate": 4.382382382382383e-06,
      "loss": 0.1895,
      "step": 56120
    },
    {
      "epoch": 168.56,
      "grad_norm": 11.645040512084961,
      "learning_rate": 4.381381381381382e-06,
      "loss": 0.2411,
      "step": 56130
    },
    {
      "epoch": 168.59,
      "grad_norm": 8.509325981140137,
      "learning_rate": 4.3803803803803805e-06,
      "loss": 0.2313,
      "step": 56140
    },
    {
      "epoch": 168.62,
      "grad_norm": 9.731284141540527,
      "learning_rate": 4.37937937937938e-06,
      "loss": 0.2193,
      "step": 56150
    },
    {
      "epoch": 168.65,
      "grad_norm": 8.938517570495605,
      "learning_rate": 4.378378378378379e-06,
      "loss": 0.2042,
      "step": 56160
    },
    {
      "epoch": 168.68,
      "grad_norm": 8.153093338012695,
      "learning_rate": 4.377377377377378e-06,
      "loss": 0.269,
      "step": 56170
    },
    {
      "epoch": 168.71,
      "grad_norm": 10.72727108001709,
      "learning_rate": 4.3763763763763765e-06,
      "loss": 0.2081,
      "step": 56180
    },
    {
      "epoch": 168.74,
      "grad_norm": 5.190488338470459,
      "learning_rate": 4.375375375375376e-06,
      "loss": 0.2175,
      "step": 56190
    },
    {
      "epoch": 168.77,
      "grad_norm": 26.439273834228516,
      "learning_rate": 4.374374374374375e-06,
      "loss": 0.196,
      "step": 56200
    },
    {
      "epoch": 168.8,
      "grad_norm": 7.575514316558838,
      "learning_rate": 4.373373373373373e-06,
      "loss": 0.1995,
      "step": 56210
    },
    {
      "epoch": 168.83,
      "grad_norm": 6.689413070678711,
      "learning_rate": 4.372372372372373e-06,
      "loss": 0.208,
      "step": 56220
    },
    {
      "epoch": 168.86,
      "grad_norm": 10.437664031982422,
      "learning_rate": 4.371371371371372e-06,
      "loss": 0.2335,
      "step": 56230
    },
    {
      "epoch": 168.89,
      "grad_norm": 5.626933574676514,
      "learning_rate": 4.370370370370371e-06,
      "loss": 0.203,
      "step": 56240
    },
    {
      "epoch": 168.92,
      "grad_norm": 10.172431945800781,
      "learning_rate": 4.369369369369369e-06,
      "loss": 0.2543,
      "step": 56250
    },
    {
      "epoch": 168.95,
      "grad_norm": 18.517038345336914,
      "learning_rate": 4.368368368368369e-06,
      "loss": 0.2237,
      "step": 56260
    },
    {
      "epoch": 168.98,
      "grad_norm": 9.376913070678711,
      "learning_rate": 4.367367367367368e-06,
      "loss": 0.2356,
      "step": 56270
    },
    {
      "epoch": 169.0,
      "eval_accuracy": 0.9226,
      "eval_loss": 0.33692997694015503,
      "eval_runtime": 30.3151,
      "eval_samples_per_second": 329.868,
      "eval_steps_per_second": 1.319,
      "step": 56277
    },
    {
      "epoch": 169.01,
      "grad_norm": 30.043195724487305,
      "learning_rate": 4.366366366366367e-06,
      "loss": 0.3071,
      "step": 56280
    },
    {
      "epoch": 169.04,
      "grad_norm": 6.427437782287598,
      "learning_rate": 4.365365365365365e-06,
      "loss": 0.1969,
      "step": 56290
    },
    {
      "epoch": 169.07,
      "grad_norm": 9.531576156616211,
      "learning_rate": 4.364364364364365e-06,
      "loss": 0.2704,
      "step": 56300
    },
    {
      "epoch": 169.1,
      "grad_norm": 10.303248405456543,
      "learning_rate": 4.363363363363363e-06,
      "loss": 0.2536,
      "step": 56310
    },
    {
      "epoch": 169.13,
      "grad_norm": 4.812479019165039,
      "learning_rate": 4.362362362362363e-06,
      "loss": 0.2276,
      "step": 56320
    },
    {
      "epoch": 169.16,
      "grad_norm": 12.937299728393555,
      "learning_rate": 4.361361361361361e-06,
      "loss": 0.2174,
      "step": 56330
    },
    {
      "epoch": 169.19,
      "grad_norm": 14.455060958862305,
      "learning_rate": 4.360360360360361e-06,
      "loss": 0.1981,
      "step": 56340
    },
    {
      "epoch": 169.22,
      "grad_norm": 6.721241474151611,
      "learning_rate": 4.3593593593593594e-06,
      "loss": 0.1917,
      "step": 56350
    },
    {
      "epoch": 169.25,
      "grad_norm": 9.243278503417969,
      "learning_rate": 4.358358358358359e-06,
      "loss": 0.1973,
      "step": 56360
    },
    {
      "epoch": 169.28,
      "grad_norm": 7.041635036468506,
      "learning_rate": 4.357357357357358e-06,
      "loss": 0.2162,
      "step": 56370
    },
    {
      "epoch": 169.31,
      "grad_norm": 9.719215393066406,
      "learning_rate": 4.356356356356357e-06,
      "loss": 0.2138,
      "step": 56380
    },
    {
      "epoch": 169.34,
      "grad_norm": 9.53799057006836,
      "learning_rate": 4.3553553553553555e-06,
      "loss": 0.2198,
      "step": 56390
    },
    {
      "epoch": 169.37,
      "grad_norm": 9.82392406463623,
      "learning_rate": 4.354354354354355e-06,
      "loss": 0.2453,
      "step": 56400
    },
    {
      "epoch": 169.4,
      "grad_norm": 14.756361961364746,
      "learning_rate": 4.3533533533533535e-06,
      "loss": 0.2162,
      "step": 56410
    },
    {
      "epoch": 169.43,
      "grad_norm": 21.968862533569336,
      "learning_rate": 4.352352352352352e-06,
      "loss": 0.2698,
      "step": 56420
    },
    {
      "epoch": 169.46,
      "grad_norm": 8.488511085510254,
      "learning_rate": 4.351351351351352e-06,
      "loss": 0.2114,
      "step": 56430
    },
    {
      "epoch": 169.49,
      "grad_norm": 6.522435665130615,
      "learning_rate": 4.350350350350351e-06,
      "loss": 0.2533,
      "step": 56440
    },
    {
      "epoch": 169.52,
      "grad_norm": 11.069390296936035,
      "learning_rate": 4.34934934934935e-06,
      "loss": 0.2089,
      "step": 56450
    },
    {
      "epoch": 169.55,
      "grad_norm": 10.15515422821045,
      "learning_rate": 4.348348348348348e-06,
      "loss": 0.2619,
      "step": 56460
    },
    {
      "epoch": 169.58,
      "grad_norm": 12.12926197052002,
      "learning_rate": 4.347347347347348e-06,
      "loss": 0.2485,
      "step": 56470
    },
    {
      "epoch": 169.61,
      "grad_norm": 10.26136589050293,
      "learning_rate": 4.346346346346347e-06,
      "loss": 0.2822,
      "step": 56480
    },
    {
      "epoch": 169.64,
      "grad_norm": 8.706718444824219,
      "learning_rate": 4.345345345345346e-06,
      "loss": 0.2046,
      "step": 56490
    },
    {
      "epoch": 169.67,
      "grad_norm": 12.703659057617188,
      "learning_rate": 4.344344344344344e-06,
      "loss": 0.2202,
      "step": 56500
    },
    {
      "epoch": 169.7,
      "grad_norm": 11.624723434448242,
      "learning_rate": 4.343343343343344e-06,
      "loss": 0.1975,
      "step": 56510
    },
    {
      "epoch": 169.73,
      "grad_norm": 6.937861919403076,
      "learning_rate": 4.342342342342343e-06,
      "loss": 0.2282,
      "step": 56520
    },
    {
      "epoch": 169.76,
      "grad_norm": 10.365509986877441,
      "learning_rate": 4.341341341341342e-06,
      "loss": 0.2529,
      "step": 56530
    },
    {
      "epoch": 169.79,
      "grad_norm": 9.068633079528809,
      "learning_rate": 4.34034034034034e-06,
      "loss": 0.2235,
      "step": 56540
    },
    {
      "epoch": 169.82,
      "grad_norm": 11.11232852935791,
      "learning_rate": 4.33933933933934e-06,
      "loss": 0.237,
      "step": 56550
    },
    {
      "epoch": 169.85,
      "grad_norm": 12.168546676635742,
      "learning_rate": 4.338338338338338e-06,
      "loss": 0.2266,
      "step": 56560
    },
    {
      "epoch": 169.88,
      "grad_norm": 5.994349479675293,
      "learning_rate": 4.337337337337338e-06,
      "loss": 0.2603,
      "step": 56570
    },
    {
      "epoch": 169.91,
      "grad_norm": 6.548464298248291,
      "learning_rate": 4.3363363363363364e-06,
      "loss": 0.2087,
      "step": 56580
    },
    {
      "epoch": 169.94,
      "grad_norm": 10.570487976074219,
      "learning_rate": 4.335335335335336e-06,
      "loss": 0.2083,
      "step": 56590
    },
    {
      "epoch": 169.97,
      "grad_norm": 10.166498184204102,
      "learning_rate": 4.3343343343343345e-06,
      "loss": 0.2487,
      "step": 56600
    },
    {
      "epoch": 170.0,
      "grad_norm": 2.263578414916992,
      "learning_rate": 4.333333333333334e-06,
      "loss": 0.2092,
      "step": 56610
    },
    {
      "epoch": 170.0,
      "eval_accuracy": 0.923,
      "eval_loss": 0.3306306004524231,
      "eval_runtime": 30.4528,
      "eval_samples_per_second": 328.377,
      "eval_steps_per_second": 1.314,
      "step": 56610
    },
    {
      "epoch": 170.03,
      "grad_norm": 9.827398300170898,
      "learning_rate": 4.3323323323323325e-06,
      "loss": 0.2,
      "step": 56620
    },
    {
      "epoch": 170.06,
      "grad_norm": 7.300294399261475,
      "learning_rate": 4.331331331331332e-06,
      "loss": 0.2462,
      "step": 56630
    },
    {
      "epoch": 170.09,
      "grad_norm": 7.264718532562256,
      "learning_rate": 4.3303303303303305e-06,
      "loss": 0.2379,
      "step": 56640
    },
    {
      "epoch": 170.12,
      "grad_norm": 11.797207832336426,
      "learning_rate": 4.32932932932933e-06,
      "loss": 0.1928,
      "step": 56650
    },
    {
      "epoch": 170.15,
      "grad_norm": 8.073410034179688,
      "learning_rate": 4.328328328328329e-06,
      "loss": 0.2238,
      "step": 56660
    },
    {
      "epoch": 170.18,
      "grad_norm": 10.51210880279541,
      "learning_rate": 4.327327327327327e-06,
      "loss": 0.2109,
      "step": 56670
    },
    {
      "epoch": 170.21,
      "grad_norm": 9.315170288085938,
      "learning_rate": 4.326326326326327e-06,
      "loss": 0.2152,
      "step": 56680
    },
    {
      "epoch": 170.24,
      "grad_norm": 12.081932067871094,
      "learning_rate": 4.325325325325326e-06,
      "loss": 0.2313,
      "step": 56690
    },
    {
      "epoch": 170.27,
      "grad_norm": 6.861198902130127,
      "learning_rate": 4.324324324324325e-06,
      "loss": 0.2282,
      "step": 56700
    },
    {
      "epoch": 170.3,
      "grad_norm": 18.568439483642578,
      "learning_rate": 4.323323323323323e-06,
      "loss": 0.2155,
      "step": 56710
    },
    {
      "epoch": 170.33,
      "grad_norm": 7.353390216827393,
      "learning_rate": 4.322322322322323e-06,
      "loss": 0.2356,
      "step": 56720
    },
    {
      "epoch": 170.36,
      "grad_norm": 10.519542694091797,
      "learning_rate": 4.321321321321322e-06,
      "loss": 0.239,
      "step": 56730
    },
    {
      "epoch": 170.39,
      "grad_norm": 9.910476684570312,
      "learning_rate": 4.320320320320321e-06,
      "loss": 0.2525,
      "step": 56740
    },
    {
      "epoch": 170.42,
      "grad_norm": 9.434966087341309,
      "learning_rate": 4.319319319319319e-06,
      "loss": 0.2499,
      "step": 56750
    },
    {
      "epoch": 170.45,
      "grad_norm": 6.501190662384033,
      "learning_rate": 4.318318318318319e-06,
      "loss": 0.2298,
      "step": 56760
    },
    {
      "epoch": 170.48,
      "grad_norm": 8.682838439941406,
      "learning_rate": 4.317317317317317e-06,
      "loss": 0.2019,
      "step": 56770
    },
    {
      "epoch": 170.51,
      "grad_norm": 7.586459159851074,
      "learning_rate": 4.316316316316317e-06,
      "loss": 0.254,
      "step": 56780
    },
    {
      "epoch": 170.54,
      "grad_norm": 5.4187211990356445,
      "learning_rate": 4.315315315315315e-06,
      "loss": 0.226,
      "step": 56790
    },
    {
      "epoch": 170.57,
      "grad_norm": 7.594554424285889,
      "learning_rate": 4.314314314314315e-06,
      "loss": 0.2365,
      "step": 56800
    },
    {
      "epoch": 170.6,
      "grad_norm": 6.372311115264893,
      "learning_rate": 4.3133133133133134e-06,
      "loss": 0.2005,
      "step": 56810
    },
    {
      "epoch": 170.63,
      "grad_norm": 6.932861804962158,
      "learning_rate": 4.312312312312313e-06,
      "loss": 0.2745,
      "step": 56820
    },
    {
      "epoch": 170.66,
      "grad_norm": 17.212358474731445,
      "learning_rate": 4.3113113113113115e-06,
      "loss": 0.2326,
      "step": 56830
    },
    {
      "epoch": 170.69,
      "grad_norm": 6.479152202606201,
      "learning_rate": 4.310310310310311e-06,
      "loss": 0.2091,
      "step": 56840
    },
    {
      "epoch": 170.72,
      "grad_norm": 6.895174026489258,
      "learning_rate": 4.3093093093093095e-06,
      "loss": 0.2326,
      "step": 56850
    },
    {
      "epoch": 170.75,
      "grad_norm": 10.32742691040039,
      "learning_rate": 4.308308308308309e-06,
      "loss": 0.21,
      "step": 56860
    },
    {
      "epoch": 170.78,
      "grad_norm": 9.569794654846191,
      "learning_rate": 4.3073073073073075e-06,
      "loss": 0.2252,
      "step": 56870
    },
    {
      "epoch": 170.81,
      "grad_norm": 12.077244758605957,
      "learning_rate": 4.306306306306306e-06,
      "loss": 0.2148,
      "step": 56880
    },
    {
      "epoch": 170.84,
      "grad_norm": 5.180423736572266,
      "learning_rate": 4.305305305305306e-06,
      "loss": 0.26,
      "step": 56890
    },
    {
      "epoch": 170.87,
      "grad_norm": 7.150249004364014,
      "learning_rate": 4.304304304304305e-06,
      "loss": 0.2224,
      "step": 56900
    },
    {
      "epoch": 170.9,
      "grad_norm": 10.290822982788086,
      "learning_rate": 4.303303303303304e-06,
      "loss": 0.2415,
      "step": 56910
    },
    {
      "epoch": 170.93,
      "grad_norm": 8.094365119934082,
      "learning_rate": 4.302302302302302e-06,
      "loss": 0.2104,
      "step": 56920
    },
    {
      "epoch": 170.96,
      "grad_norm": 10.508177757263184,
      "learning_rate": 4.301301301301302e-06,
      "loss": 0.2378,
      "step": 56930
    },
    {
      "epoch": 170.99,
      "grad_norm": 11.909675598144531,
      "learning_rate": 4.300300300300301e-06,
      "loss": 0.2239,
      "step": 56940
    },
    {
      "epoch": 171.0,
      "eval_accuracy": 0.9237,
      "eval_loss": 0.3386991024017334,
      "eval_runtime": 29.9991,
      "eval_samples_per_second": 333.343,
      "eval_steps_per_second": 1.333,
      "step": 56943
    },
    {
      "epoch": 171.02,
      "grad_norm": 9.324881553649902,
      "learning_rate": 4.2992992992993e-06,
      "loss": 0.1871,
      "step": 56950
    },
    {
      "epoch": 171.05,
      "grad_norm": 6.504618167877197,
      "learning_rate": 4.298298298298298e-06,
      "loss": 0.221,
      "step": 56960
    },
    {
      "epoch": 171.08,
      "grad_norm": 8.553045272827148,
      "learning_rate": 4.297297297297298e-06,
      "loss": 0.2293,
      "step": 56970
    },
    {
      "epoch": 171.11,
      "grad_norm": 10.500686645507812,
      "learning_rate": 4.296296296296296e-06,
      "loss": 0.2069,
      "step": 56980
    },
    {
      "epoch": 171.14,
      "grad_norm": 8.642040252685547,
      "learning_rate": 4.295295295295296e-06,
      "loss": 0.2392,
      "step": 56990
    },
    {
      "epoch": 171.17,
      "grad_norm": 6.772470474243164,
      "learning_rate": 4.294294294294294e-06,
      "loss": 0.2127,
      "step": 57000
    },
    {
      "epoch": 171.2,
      "grad_norm": 7.563378810882568,
      "learning_rate": 4.293293293293294e-06,
      "loss": 0.2453,
      "step": 57010
    },
    {
      "epoch": 171.23,
      "grad_norm": 8.290887832641602,
      "learning_rate": 4.292292292292292e-06,
      "loss": 0.2036,
      "step": 57020
    },
    {
      "epoch": 171.26,
      "grad_norm": 8.112096786499023,
      "learning_rate": 4.291291291291292e-06,
      "loss": 0.2449,
      "step": 57030
    },
    {
      "epoch": 171.29,
      "grad_norm": 7.749941825866699,
      "learning_rate": 4.2902902902902904e-06,
      "loss": 0.2294,
      "step": 57040
    },
    {
      "epoch": 171.32,
      "grad_norm": 12.484456062316895,
      "learning_rate": 4.28928928928929e-06,
      "loss": 0.2428,
      "step": 57050
    },
    {
      "epoch": 171.35,
      "grad_norm": 10.670555114746094,
      "learning_rate": 4.2882882882882885e-06,
      "loss": 0.249,
      "step": 57060
    },
    {
      "epoch": 171.38,
      "grad_norm": 4.591343402862549,
      "learning_rate": 4.287287287287288e-06,
      "loss": 0.2291,
      "step": 57070
    },
    {
      "epoch": 171.41,
      "grad_norm": 9.391070365905762,
      "learning_rate": 4.2862862862862865e-06,
      "loss": 0.2021,
      "step": 57080
    },
    {
      "epoch": 171.44,
      "grad_norm": 9.157939910888672,
      "learning_rate": 4.285285285285286e-06,
      "loss": 0.2253,
      "step": 57090
    },
    {
      "epoch": 171.47,
      "grad_norm": 15.3589448928833,
      "learning_rate": 4.2842842842842845e-06,
      "loss": 0.1898,
      "step": 57100
    },
    {
      "epoch": 171.5,
      "grad_norm": 6.151900768280029,
      "learning_rate": 4.283283283283284e-06,
      "loss": 0.225,
      "step": 57110
    },
    {
      "epoch": 171.53,
      "grad_norm": 10.636837005615234,
      "learning_rate": 4.282282282282283e-06,
      "loss": 0.194,
      "step": 57120
    },
    {
      "epoch": 171.56,
      "grad_norm": 11.628396987915039,
      "learning_rate": 4.281281281281281e-06,
      "loss": 0.244,
      "step": 57130
    },
    {
      "epoch": 171.59,
      "grad_norm": 5.675658702850342,
      "learning_rate": 4.280280280280281e-06,
      "loss": 0.2503,
      "step": 57140
    },
    {
      "epoch": 171.62,
      "grad_norm": 6.410616397857666,
      "learning_rate": 4.27927927927928e-06,
      "loss": 0.2155,
      "step": 57150
    },
    {
      "epoch": 171.65,
      "grad_norm": 9.093680381774902,
      "learning_rate": 4.278278278278279e-06,
      "loss": 0.2127,
      "step": 57160
    },
    {
      "epoch": 171.68,
      "grad_norm": 5.299399375915527,
      "learning_rate": 4.277277277277277e-06,
      "loss": 0.1953,
      "step": 57170
    },
    {
      "epoch": 171.71,
      "grad_norm": 9.162246704101562,
      "learning_rate": 4.276276276276277e-06,
      "loss": 0.2353,
      "step": 57180
    },
    {
      "epoch": 171.74,
      "grad_norm": 12.715832710266113,
      "learning_rate": 4.275275275275276e-06,
      "loss": 0.2218,
      "step": 57190
    },
    {
      "epoch": 171.77,
      "grad_norm": 9.971089363098145,
      "learning_rate": 4.274274274274275e-06,
      "loss": 0.1996,
      "step": 57200
    },
    {
      "epoch": 171.8,
      "grad_norm": 14.476478576660156,
      "learning_rate": 4.273273273273273e-06,
      "loss": 0.2276,
      "step": 57210
    },
    {
      "epoch": 171.83,
      "grad_norm": 9.325133323669434,
      "learning_rate": 4.272272272272273e-06,
      "loss": 0.2103,
      "step": 57220
    },
    {
      "epoch": 171.86,
      "grad_norm": 6.979363918304443,
      "learning_rate": 4.271271271271271e-06,
      "loss": 0.262,
      "step": 57230
    },
    {
      "epoch": 171.89,
      "grad_norm": 9.328625679016113,
      "learning_rate": 4.270270270270271e-06,
      "loss": 0.2493,
      "step": 57240
    },
    {
      "epoch": 171.92,
      "grad_norm": 10.524125099182129,
      "learning_rate": 4.269269269269269e-06,
      "loss": 0.2001,
      "step": 57250
    },
    {
      "epoch": 171.95,
      "grad_norm": 6.549038887023926,
      "learning_rate": 4.268268268268269e-06,
      "loss": 0.2148,
      "step": 57260
    },
    {
      "epoch": 171.98,
      "grad_norm": 6.266604900360107,
      "learning_rate": 4.2672672672672674e-06,
      "loss": 0.2157,
      "step": 57270
    },
    {
      "epoch": 172.0,
      "eval_accuracy": 0.9222,
      "eval_loss": 0.3346765339374542,
      "eval_runtime": 29.7092,
      "eval_samples_per_second": 336.596,
      "eval_steps_per_second": 1.346,
      "step": 57276
    },
    {
      "epoch": 172.01,
      "grad_norm": 13.361581802368164,
      "learning_rate": 4.266266266266267e-06,
      "loss": 0.2802,
      "step": 57280
    },
    {
      "epoch": 172.04,
      "grad_norm": 12.950690269470215,
      "learning_rate": 4.2652652652652655e-06,
      "loss": 0.2425,
      "step": 57290
    },
    {
      "epoch": 172.07,
      "grad_norm": 6.890408992767334,
      "learning_rate": 4.264264264264265e-06,
      "loss": 0.2349,
      "step": 57300
    },
    {
      "epoch": 172.1,
      "grad_norm": 21.62329864501953,
      "learning_rate": 4.2632632632632635e-06,
      "loss": 0.234,
      "step": 57310
    },
    {
      "epoch": 172.13,
      "grad_norm": 9.510270118713379,
      "learning_rate": 4.262262262262263e-06,
      "loss": 0.2169,
      "step": 57320
    },
    {
      "epoch": 172.16,
      "grad_norm": 7.492715358734131,
      "learning_rate": 4.2612612612612615e-06,
      "loss": 0.2051,
      "step": 57330
    },
    {
      "epoch": 172.19,
      "grad_norm": 5.936410903930664,
      "learning_rate": 4.26026026026026e-06,
      "loss": 0.1688,
      "step": 57340
    },
    {
      "epoch": 172.22,
      "grad_norm": 5.468741416931152,
      "learning_rate": 4.2592592592592596e-06,
      "loss": 0.1923,
      "step": 57350
    },
    {
      "epoch": 172.25,
      "grad_norm": 6.186549186706543,
      "learning_rate": 4.258258258258259e-06,
      "loss": 0.1888,
      "step": 57360
    },
    {
      "epoch": 172.28,
      "grad_norm": 10.714591979980469,
      "learning_rate": 4.257257257257258e-06,
      "loss": 0.2019,
      "step": 57370
    },
    {
      "epoch": 172.31,
      "grad_norm": 8.641070365905762,
      "learning_rate": 4.256256256256256e-06,
      "loss": 0.2159,
      "step": 57380
    },
    {
      "epoch": 172.34,
      "grad_norm": 4.645728588104248,
      "learning_rate": 4.255255255255256e-06,
      "loss": 0.2039,
      "step": 57390
    },
    {
      "epoch": 172.37,
      "grad_norm": 17.750829696655273,
      "learning_rate": 4.254254254254255e-06,
      "loss": 0.2065,
      "step": 57400
    },
    {
      "epoch": 172.4,
      "grad_norm": 12.792646408081055,
      "learning_rate": 4.253253253253254e-06,
      "loss": 0.2274,
      "step": 57410
    },
    {
      "epoch": 172.43,
      "grad_norm": 18.138282775878906,
      "learning_rate": 4.252252252252252e-06,
      "loss": 0.2104,
      "step": 57420
    },
    {
      "epoch": 172.46,
      "grad_norm": 7.524521350860596,
      "learning_rate": 4.251251251251252e-06,
      "loss": 0.1633,
      "step": 57430
    },
    {
      "epoch": 172.49,
      "grad_norm": 7.067676067352295,
      "learning_rate": 4.25025025025025e-06,
      "loss": 0.2,
      "step": 57440
    },
    {
      "epoch": 172.52,
      "grad_norm": 7.228219985961914,
      "learning_rate": 4.24924924924925e-06,
      "loss": 0.2189,
      "step": 57450
    },
    {
      "epoch": 172.55,
      "grad_norm": 10.218392372131348,
      "learning_rate": 4.248248248248248e-06,
      "loss": 0.2188,
      "step": 57460
    },
    {
      "epoch": 172.58,
      "grad_norm": 9.370089530944824,
      "learning_rate": 4.247247247247248e-06,
      "loss": 0.2441,
      "step": 57470
    },
    {
      "epoch": 172.61,
      "grad_norm": 9.190145492553711,
      "learning_rate": 4.246246246246246e-06,
      "loss": 0.2295,
      "step": 57480
    },
    {
      "epoch": 172.64,
      "grad_norm": 6.537145137786865,
      "learning_rate": 4.245245245245246e-06,
      "loss": 0.2491,
      "step": 57490
    },
    {
      "epoch": 172.67,
      "grad_norm": 5.943038463592529,
      "learning_rate": 4.2442442442442444e-06,
      "loss": 0.2143,
      "step": 57500
    },
    {
      "epoch": 172.7,
      "grad_norm": 5.958277225494385,
      "learning_rate": 4.243243243243244e-06,
      "loss": 0.2481,
      "step": 57510
    },
    {
      "epoch": 172.73,
      "grad_norm": 12.843267440795898,
      "learning_rate": 4.2422422422422425e-06,
      "loss": 0.2201,
      "step": 57520
    },
    {
      "epoch": 172.76,
      "grad_norm": 8.235824584960938,
      "learning_rate": 4.241241241241242e-06,
      "loss": 0.2194,
      "step": 57530
    },
    {
      "epoch": 172.79,
      "grad_norm": 10.644575119018555,
      "learning_rate": 4.2402402402402405e-06,
      "loss": 0.2541,
      "step": 57540
    },
    {
      "epoch": 172.82,
      "grad_norm": 5.65022611618042,
      "learning_rate": 4.23923923923924e-06,
      "loss": 0.201,
      "step": 57550
    },
    {
      "epoch": 172.85,
      "grad_norm": 5.462368965148926,
      "learning_rate": 4.2382382382382385e-06,
      "loss": 0.2116,
      "step": 57560
    },
    {
      "epoch": 172.88,
      "grad_norm": 7.097109317779541,
      "learning_rate": 4.237237237237238e-06,
      "loss": 0.2198,
      "step": 57570
    },
    {
      "epoch": 172.91,
      "grad_norm": 8.001241683959961,
      "learning_rate": 4.2362362362362366e-06,
      "loss": 0.1833,
      "step": 57580
    },
    {
      "epoch": 172.94,
      "grad_norm": 4.821046352386475,
      "learning_rate": 4.235235235235235e-06,
      "loss": 0.2136,
      "step": 57590
    },
    {
      "epoch": 172.97,
      "grad_norm": 8.394944190979004,
      "learning_rate": 4.234234234234235e-06,
      "loss": 0.1877,
      "step": 57600
    },
    {
      "epoch": 173.0,
      "eval_accuracy": 0.9224,
      "eval_loss": 0.3423205018043518,
      "eval_runtime": 29.9671,
      "eval_samples_per_second": 333.699,
      "eval_steps_per_second": 1.335,
      "step": 57609
    },
    {
      "epoch": 173.0,
      "grad_norm": 6.315654754638672,
      "learning_rate": 4.233233233233234e-06,
      "loss": 0.2371,
      "step": 57610
    },
    {
      "epoch": 173.03,
      "grad_norm": 6.175839424133301,
      "learning_rate": 4.232232232232233e-06,
      "loss": 0.173,
      "step": 57620
    },
    {
      "epoch": 173.06,
      "grad_norm": 7.198038101196289,
      "learning_rate": 4.231231231231231e-06,
      "loss": 0.1829,
      "step": 57630
    },
    {
      "epoch": 173.09,
      "grad_norm": 10.908315658569336,
      "learning_rate": 4.230230230230231e-06,
      "loss": 0.2245,
      "step": 57640
    },
    {
      "epoch": 173.12,
      "grad_norm": 9.971311569213867,
      "learning_rate": 4.22922922922923e-06,
      "loss": 0.2247,
      "step": 57650
    },
    {
      "epoch": 173.15,
      "grad_norm": 8.214251518249512,
      "learning_rate": 4.228228228228229e-06,
      "loss": 0.1984,
      "step": 57660
    },
    {
      "epoch": 173.18,
      "grad_norm": 11.873804092407227,
      "learning_rate": 4.227227227227227e-06,
      "loss": 0.1957,
      "step": 57670
    },
    {
      "epoch": 173.21,
      "grad_norm": 9.74660587310791,
      "learning_rate": 4.226226226226227e-06,
      "loss": 0.277,
      "step": 57680
    },
    {
      "epoch": 173.24,
      "grad_norm": 9.93647575378418,
      "learning_rate": 4.225225225225225e-06,
      "loss": 0.1945,
      "step": 57690
    },
    {
      "epoch": 173.27,
      "grad_norm": 13.643671035766602,
      "learning_rate": 4.224224224224225e-06,
      "loss": 0.2077,
      "step": 57700
    },
    {
      "epoch": 173.3,
      "grad_norm": 6.899477958679199,
      "learning_rate": 4.223223223223223e-06,
      "loss": 0.268,
      "step": 57710
    },
    {
      "epoch": 173.33,
      "grad_norm": 6.821637153625488,
      "learning_rate": 4.222222222222223e-06,
      "loss": 0.2244,
      "step": 57720
    },
    {
      "epoch": 173.36,
      "grad_norm": 10.013042449951172,
      "learning_rate": 4.2212212212212214e-06,
      "loss": 0.2229,
      "step": 57730
    },
    {
      "epoch": 173.39,
      "grad_norm": 7.914147853851318,
      "learning_rate": 4.220220220220221e-06,
      "loss": 0.2126,
      "step": 57740
    },
    {
      "epoch": 173.42,
      "grad_norm": 13.919772148132324,
      "learning_rate": 4.2192192192192195e-06,
      "loss": 0.2316,
      "step": 57750
    },
    {
      "epoch": 173.45,
      "grad_norm": 10.300623893737793,
      "learning_rate": 4.218218218218219e-06,
      "loss": 0.2081,
      "step": 57760
    },
    {
      "epoch": 173.48,
      "grad_norm": 8.269445419311523,
      "learning_rate": 4.2172172172172175e-06,
      "loss": 0.2384,
      "step": 57770
    },
    {
      "epoch": 173.51,
      "grad_norm": 7.458897590637207,
      "learning_rate": 4.216216216216217e-06,
      "loss": 0.2038,
      "step": 57780
    },
    {
      "epoch": 173.54,
      "grad_norm": 9.803679466247559,
      "learning_rate": 4.2152152152152155e-06,
      "loss": 0.2062,
      "step": 57790
    },
    {
      "epoch": 173.57,
      "grad_norm": 5.631911754608154,
      "learning_rate": 4.214214214214214e-06,
      "loss": 0.2186,
      "step": 57800
    },
    {
      "epoch": 173.6,
      "grad_norm": 10.289484977722168,
      "learning_rate": 4.2132132132132136e-06,
      "loss": 0.1976,
      "step": 57810
    },
    {
      "epoch": 173.63,
      "grad_norm": 8.610011100769043,
      "learning_rate": 4.212212212212213e-06,
      "loss": 0.2371,
      "step": 57820
    },
    {
      "epoch": 173.66,
      "grad_norm": 6.874488830566406,
      "learning_rate": 4.211211211211212e-06,
      "loss": 0.2396,
      "step": 57830
    },
    {
      "epoch": 173.69,
      "grad_norm": 9.344332695007324,
      "learning_rate": 4.21021021021021e-06,
      "loss": 0.1762,
      "step": 57840
    },
    {
      "epoch": 173.72,
      "grad_norm": 6.085450172424316,
      "learning_rate": 4.20920920920921e-06,
      "loss": 0.2189,
      "step": 57850
    },
    {
      "epoch": 173.75,
      "grad_norm": 11.465705871582031,
      "learning_rate": 4.208208208208209e-06,
      "loss": 0.25,
      "step": 57860
    },
    {
      "epoch": 173.78,
      "grad_norm": 9.614089965820312,
      "learning_rate": 4.207207207207208e-06,
      "loss": 0.2452,
      "step": 57870
    },
    {
      "epoch": 173.81,
      "grad_norm": 7.592823505401611,
      "learning_rate": 4.206206206206206e-06,
      "loss": 0.1907,
      "step": 57880
    },
    {
      "epoch": 173.84,
      "grad_norm": 8.995274543762207,
      "learning_rate": 4.205205205205206e-06,
      "loss": 0.2524,
      "step": 57890
    },
    {
      "epoch": 173.87,
      "grad_norm": 10.288630485534668,
      "learning_rate": 4.204204204204204e-06,
      "loss": 0.2283,
      "step": 57900
    },
    {
      "epoch": 173.9,
      "grad_norm": 10.407158851623535,
      "learning_rate": 4.203203203203203e-06,
      "loss": 0.1872,
      "step": 57910
    },
    {
      "epoch": 173.93,
      "grad_norm": 9.764894485473633,
      "learning_rate": 4.202202202202202e-06,
      "loss": 0.268,
      "step": 57920
    },
    {
      "epoch": 173.96,
      "grad_norm": 12.616869926452637,
      "learning_rate": 4.201201201201202e-06,
      "loss": 0.2191,
      "step": 57930
    },
    {
      "epoch": 173.99,
      "grad_norm": 9.27846908569336,
      "learning_rate": 4.2002002002002e-06,
      "loss": 0.2532,
      "step": 57940
    },
    {
      "epoch": 174.0,
      "eval_accuracy": 0.9231,
      "eval_loss": 0.33423763513565063,
      "eval_runtime": 30.3648,
      "eval_samples_per_second": 329.329,
      "eval_steps_per_second": 1.317,
      "step": 57942
    },
    {
      "epoch": 174.02,
      "grad_norm": 9.811721801757812,
      "learning_rate": 4.199199199199199e-06,
      "loss": 0.1722,
      "step": 57950
    },
    {
      "epoch": 174.05,
      "grad_norm": 5.832336902618408,
      "learning_rate": 4.1981981981981984e-06,
      "loss": 0.2338,
      "step": 57960
    },
    {
      "epoch": 174.08,
      "grad_norm": 3.7718422412872314,
      "learning_rate": 4.197197197197198e-06,
      "loss": 0.2222,
      "step": 57970
    },
    {
      "epoch": 174.11,
      "grad_norm": 12.225495338439941,
      "learning_rate": 4.1961961961961965e-06,
      "loss": 0.2722,
      "step": 57980
    },
    {
      "epoch": 174.14,
      "grad_norm": 11.95918083190918,
      "learning_rate": 4.195195195195196e-06,
      "loss": 0.2062,
      "step": 57990
    },
    {
      "epoch": 174.17,
      "grad_norm": 7.998116493225098,
      "learning_rate": 4.1941941941941945e-06,
      "loss": 0.213,
      "step": 58000
    },
    {
      "epoch": 174.2,
      "grad_norm": 14.918134689331055,
      "learning_rate": 4.193193193193193e-06,
      "loss": 0.2815,
      "step": 58010
    },
    {
      "epoch": 174.23,
      "grad_norm": 10.239768981933594,
      "learning_rate": 4.1921921921921925e-06,
      "loss": 0.2343,
      "step": 58020
    },
    {
      "epoch": 174.26,
      "grad_norm": 10.131564140319824,
      "learning_rate": 4.191191191191192e-06,
      "loss": 0.2162,
      "step": 58030
    },
    {
      "epoch": 174.29,
      "grad_norm": 9.199658393859863,
      "learning_rate": 4.1901901901901906e-06,
      "loss": 0.2604,
      "step": 58040
    },
    {
      "epoch": 174.32,
      "grad_norm": 13.29453182220459,
      "learning_rate": 4.189189189189189e-06,
      "loss": 0.1742,
      "step": 58050
    },
    {
      "epoch": 174.35,
      "grad_norm": 4.838201522827148,
      "learning_rate": 4.188188188188189e-06,
      "loss": 0.1991,
      "step": 58060
    },
    {
      "epoch": 174.38,
      "grad_norm": 6.56643533706665,
      "learning_rate": 4.187187187187188e-06,
      "loss": 0.2359,
      "step": 58070
    },
    {
      "epoch": 174.41,
      "grad_norm": 8.983172416687012,
      "learning_rate": 4.186186186186187e-06,
      "loss": 0.1972,
      "step": 58080
    },
    {
      "epoch": 174.44,
      "grad_norm": 7.786691188812256,
      "learning_rate": 4.185185185185185e-06,
      "loss": 0.1925,
      "step": 58090
    },
    {
      "epoch": 174.47,
      "grad_norm": 11.083693504333496,
      "learning_rate": 4.184184184184185e-06,
      "loss": 0.1895,
      "step": 58100
    },
    {
      "epoch": 174.5,
      "grad_norm": 8.96174144744873,
      "learning_rate": 4.183183183183184e-06,
      "loss": 0.2101,
      "step": 58110
    },
    {
      "epoch": 174.53,
      "grad_norm": 8.416348457336426,
      "learning_rate": 4.182182182182183e-06,
      "loss": 0.2268,
      "step": 58120
    },
    {
      "epoch": 174.56,
      "grad_norm": 9.577929496765137,
      "learning_rate": 4.181181181181181e-06,
      "loss": 0.2191,
      "step": 58130
    },
    {
      "epoch": 174.59,
      "grad_norm": 7.846167087554932,
      "learning_rate": 4.180180180180181e-06,
      "loss": 0.2241,
      "step": 58140
    },
    {
      "epoch": 174.62,
      "grad_norm": 5.442991733551025,
      "learning_rate": 4.179179179179179e-06,
      "loss": 0.2303,
      "step": 58150
    },
    {
      "epoch": 174.65,
      "grad_norm": 11.855753898620605,
      "learning_rate": 4.178178178178178e-06,
      "loss": 0.2186,
      "step": 58160
    },
    {
      "epoch": 174.68,
      "grad_norm": 9.799480438232422,
      "learning_rate": 4.177177177177177e-06,
      "loss": 0.2295,
      "step": 58170
    },
    {
      "epoch": 174.71,
      "grad_norm": 7.193188667297363,
      "learning_rate": 4.176176176176177e-06,
      "loss": 0.2146,
      "step": 58180
    },
    {
      "epoch": 174.74,
      "grad_norm": 5.449995517730713,
      "learning_rate": 4.175175175175175e-06,
      "loss": 0.2074,
      "step": 58190
    },
    {
      "epoch": 174.77,
      "grad_norm": 7.046781539916992,
      "learning_rate": 4.174174174174174e-06,
      "loss": 0.2357,
      "step": 58200
    },
    {
      "epoch": 174.8,
      "grad_norm": 14.473947525024414,
      "learning_rate": 4.1731731731731735e-06,
      "loss": 0.2205,
      "step": 58210
    },
    {
      "epoch": 174.83,
      "grad_norm": 7.724979400634766,
      "learning_rate": 4.172172172172173e-06,
      "loss": 0.267,
      "step": 58220
    },
    {
      "epoch": 174.86,
      "grad_norm": 14.857132911682129,
      "learning_rate": 4.1711711711711715e-06,
      "loss": 0.2697,
      "step": 58230
    },
    {
      "epoch": 174.89,
      "grad_norm": 7.877825736999512,
      "learning_rate": 4.170170170170171e-06,
      "loss": 0.2357,
      "step": 58240
    },
    {
      "epoch": 174.92,
      "grad_norm": 5.219593524932861,
      "learning_rate": 4.1691691691691695e-06,
      "loss": 0.1978,
      "step": 58250
    },
    {
      "epoch": 174.95,
      "grad_norm": 24.46590805053711,
      "learning_rate": 4.168168168168168e-06,
      "loss": 0.2458,
      "step": 58260
    },
    {
      "epoch": 174.98,
      "grad_norm": 7.763651371002197,
      "learning_rate": 4.1671671671671676e-06,
      "loss": 0.2306,
      "step": 58270
    },
    {
      "epoch": 175.0,
      "eval_accuracy": 0.9223,
      "eval_loss": 0.3363642692565918,
      "eval_runtime": 30.4261,
      "eval_samples_per_second": 328.666,
      "eval_steps_per_second": 1.315,
      "step": 58275
    },
    {
      "epoch": 175.02,
      "grad_norm": 9.673617362976074,
      "learning_rate": 4.166166166166167e-06,
      "loss": 0.2224,
      "step": 58280
    },
    {
      "epoch": 175.05,
      "grad_norm": 10.603177070617676,
      "learning_rate": 4.165165165165166e-06,
      "loss": 0.239,
      "step": 58290
    },
    {
      "epoch": 175.08,
      "grad_norm": 8.857147216796875,
      "learning_rate": 4.164164164164164e-06,
      "loss": 0.2234,
      "step": 58300
    },
    {
      "epoch": 175.11,
      "grad_norm": 13.711175918579102,
      "learning_rate": 4.163163163163164e-06,
      "loss": 0.2196,
      "step": 58310
    },
    {
      "epoch": 175.14,
      "grad_norm": 7.084354877471924,
      "learning_rate": 4.162162162162163e-06,
      "loss": 0.188,
      "step": 58320
    },
    {
      "epoch": 175.17,
      "grad_norm": 9.78482437133789,
      "learning_rate": 4.161161161161162e-06,
      "loss": 0.19,
      "step": 58330
    },
    {
      "epoch": 175.2,
      "grad_norm": 10.502204895019531,
      "learning_rate": 4.16016016016016e-06,
      "loss": 0.2401,
      "step": 58340
    },
    {
      "epoch": 175.23,
      "grad_norm": 9.788403511047363,
      "learning_rate": 4.15915915915916e-06,
      "loss": 0.2143,
      "step": 58350
    },
    {
      "epoch": 175.26,
      "grad_norm": 5.398682117462158,
      "learning_rate": 4.158158158158158e-06,
      "loss": 0.2232,
      "step": 58360
    },
    {
      "epoch": 175.29,
      "grad_norm": 4.740347862243652,
      "learning_rate": 4.157157157157157e-06,
      "loss": 0.2198,
      "step": 58370
    },
    {
      "epoch": 175.32,
      "grad_norm": 33.813533782958984,
      "learning_rate": 4.156156156156156e-06,
      "loss": 0.2025,
      "step": 58380
    },
    {
      "epoch": 175.35,
      "grad_norm": 5.945501327514648,
      "learning_rate": 4.155155155155156e-06,
      "loss": 0.2081,
      "step": 58390
    },
    {
      "epoch": 175.38,
      "grad_norm": 8.19374942779541,
      "learning_rate": 4.154154154154154e-06,
      "loss": 0.2098,
      "step": 58400
    },
    {
      "epoch": 175.41,
      "grad_norm": 7.910248756408691,
      "learning_rate": 4.153153153153153e-06,
      "loss": 0.1969,
      "step": 58410
    },
    {
      "epoch": 175.44,
      "grad_norm": 9.54606819152832,
      "learning_rate": 4.152152152152152e-06,
      "loss": 0.2115,
      "step": 58420
    },
    {
      "epoch": 175.47,
      "grad_norm": 6.637490272521973,
      "learning_rate": 4.151151151151152e-06,
      "loss": 0.2166,
      "step": 58430
    },
    {
      "epoch": 175.5,
      "grad_norm": 6.465303897857666,
      "learning_rate": 4.1501501501501505e-06,
      "loss": 0.2156,
      "step": 58440
    },
    {
      "epoch": 175.53,
      "grad_norm": 8.124563217163086,
      "learning_rate": 4.149149149149149e-06,
      "loss": 0.2304,
      "step": 58450
    },
    {
      "epoch": 175.56,
      "grad_norm": 16.654239654541016,
      "learning_rate": 4.1481481481481485e-06,
      "loss": 0.2047,
      "step": 58460
    },
    {
      "epoch": 175.59,
      "grad_norm": 8.982221603393555,
      "learning_rate": 4.147147147147147e-06,
      "loss": 0.2042,
      "step": 58470
    },
    {
      "epoch": 175.62,
      "grad_norm": 19.698036193847656,
      "learning_rate": 4.1461461461461465e-06,
      "loss": 0.2568,
      "step": 58480
    },
    {
      "epoch": 175.65,
      "grad_norm": 8.450174331665039,
      "learning_rate": 4.145145145145145e-06,
      "loss": 0.2079,
      "step": 58490
    },
    {
      "epoch": 175.68,
      "grad_norm": 8.816652297973633,
      "learning_rate": 4.1441441441441446e-06,
      "loss": 0.2472,
      "step": 58500
    },
    {
      "epoch": 175.71,
      "grad_norm": 12.273414611816406,
      "learning_rate": 4.143143143143143e-06,
      "loss": 0.2116,
      "step": 58510
    },
    {
      "epoch": 175.74,
      "grad_norm": 22.421741485595703,
      "learning_rate": 4.142142142142143e-06,
      "loss": 0.1957,
      "step": 58520
    },
    {
      "epoch": 175.77,
      "grad_norm": 6.9170989990234375,
      "learning_rate": 4.141141141141142e-06,
      "loss": 0.2035,
      "step": 58530
    },
    {
      "epoch": 175.8,
      "grad_norm": 11.440621376037598,
      "learning_rate": 4.140140140140141e-06,
      "loss": 0.2578,
      "step": 58540
    },
    {
      "epoch": 175.83,
      "grad_norm": 4.67933988571167,
      "learning_rate": 4.139139139139139e-06,
      "loss": 0.2226,
      "step": 58550
    },
    {
      "epoch": 175.86,
      "grad_norm": 53.61032485961914,
      "learning_rate": 4.138138138138139e-06,
      "loss": 0.2334,
      "step": 58560
    },
    {
      "epoch": 175.89,
      "grad_norm": 9.788360595703125,
      "learning_rate": 4.137137137137138e-06,
      "loss": 0.2423,
      "step": 58570
    },
    {
      "epoch": 175.92,
      "grad_norm": 11.223185539245605,
      "learning_rate": 4.136136136136137e-06,
      "loss": 0.2393,
      "step": 58580
    },
    {
      "epoch": 175.95,
      "grad_norm": 8.687237739562988,
      "learning_rate": 4.135135135135135e-06,
      "loss": 0.244,
      "step": 58590
    },
    {
      "epoch": 175.98,
      "grad_norm": 6.76766300201416,
      "learning_rate": 4.134134134134135e-06,
      "loss": 0.2247,
      "step": 58600
    },
    {
      "epoch": 176.0,
      "eval_accuracy": 0.9219,
      "eval_loss": 0.3375799357891083,
      "eval_runtime": 30.4544,
      "eval_samples_per_second": 328.36,
      "eval_steps_per_second": 1.313,
      "step": 58608
    },
    {
      "epoch": 176.01,
      "grad_norm": 12.773330688476562,
      "learning_rate": 4.133133133133133e-06,
      "loss": 0.2111,
      "step": 58610
    },
    {
      "epoch": 176.04,
      "grad_norm": 6.381572246551514,
      "learning_rate": 4.132132132132132e-06,
      "loss": 0.2222,
      "step": 58620
    },
    {
      "epoch": 176.07,
      "grad_norm": 7.154778003692627,
      "learning_rate": 4.131131131131131e-06,
      "loss": 0.2195,
      "step": 58630
    },
    {
      "epoch": 176.1,
      "grad_norm": 11.225522994995117,
      "learning_rate": 4.130130130130131e-06,
      "loss": 0.1839,
      "step": 58640
    },
    {
      "epoch": 176.13,
      "grad_norm": 9.033132553100586,
      "learning_rate": 4.129129129129129e-06,
      "loss": 0.246,
      "step": 58650
    },
    {
      "epoch": 176.16,
      "grad_norm": 6.154885292053223,
      "learning_rate": 4.128128128128128e-06,
      "loss": 0.1962,
      "step": 58660
    },
    {
      "epoch": 176.19,
      "grad_norm": 10.361030578613281,
      "learning_rate": 4.1271271271271275e-06,
      "loss": 0.2187,
      "step": 58670
    },
    {
      "epoch": 176.22,
      "grad_norm": 6.707146644592285,
      "learning_rate": 4.126126126126127e-06,
      "loss": 0.2353,
      "step": 58680
    },
    {
      "epoch": 176.25,
      "grad_norm": 6.848292827606201,
      "learning_rate": 4.1251251251251255e-06,
      "loss": 0.2392,
      "step": 58690
    },
    {
      "epoch": 176.28,
      "grad_norm": 5.849206447601318,
      "learning_rate": 4.124124124124124e-06,
      "loss": 0.2223,
      "step": 58700
    },
    {
      "epoch": 176.31,
      "grad_norm": 10.987763404846191,
      "learning_rate": 4.1231231231231235e-06,
      "loss": 0.202,
      "step": 58710
    },
    {
      "epoch": 176.34,
      "grad_norm": 6.678238391876221,
      "learning_rate": 4.122122122122122e-06,
      "loss": 0.2398,
      "step": 58720
    },
    {
      "epoch": 176.37,
      "grad_norm": 28.669260025024414,
      "learning_rate": 4.1211211211211216e-06,
      "loss": 0.2433,
      "step": 58730
    },
    {
      "epoch": 176.4,
      "grad_norm": 7.05283784866333,
      "learning_rate": 4.12012012012012e-06,
      "loss": 0.1921,
      "step": 58740
    },
    {
      "epoch": 176.43,
      "grad_norm": 7.984471321105957,
      "learning_rate": 4.11911911911912e-06,
      "loss": 0.1726,
      "step": 58750
    },
    {
      "epoch": 176.46,
      "grad_norm": 11.834104537963867,
      "learning_rate": 4.118118118118118e-06,
      "loss": 0.2135,
      "step": 58760
    },
    {
      "epoch": 176.49,
      "grad_norm": 9.792325973510742,
      "learning_rate": 4.117117117117118e-06,
      "loss": 0.2271,
      "step": 58770
    },
    {
      "epoch": 176.52,
      "grad_norm": 7.517824649810791,
      "learning_rate": 4.116116116116117e-06,
      "loss": 0.1987,
      "step": 58780
    },
    {
      "epoch": 176.55,
      "grad_norm": 18.826980590820312,
      "learning_rate": 4.115115115115116e-06,
      "loss": 0.2455,
      "step": 58790
    },
    {
      "epoch": 176.58,
      "grad_norm": 4.447484970092773,
      "learning_rate": 4.114114114114114e-06,
      "loss": 0.1815,
      "step": 58800
    },
    {
      "epoch": 176.61,
      "grad_norm": 19.489717483520508,
      "learning_rate": 4.113113113113114e-06,
      "loss": 0.2062,
      "step": 58810
    },
    {
      "epoch": 176.64,
      "grad_norm": 4.74967622756958,
      "learning_rate": 4.112112112112112e-06,
      "loss": 0.1702,
      "step": 58820
    },
    {
      "epoch": 176.67,
      "grad_norm": 12.527389526367188,
      "learning_rate": 4.111111111111111e-06,
      "loss": 0.2352,
      "step": 58830
    },
    {
      "epoch": 176.7,
      "grad_norm": 18.37725067138672,
      "learning_rate": 4.11011011011011e-06,
      "loss": 0.2142,
      "step": 58840
    },
    {
      "epoch": 176.73,
      "grad_norm": 8.646803855895996,
      "learning_rate": 4.10910910910911e-06,
      "loss": 0.1979,
      "step": 58850
    },
    {
      "epoch": 176.76,
      "grad_norm": 11.514555931091309,
      "learning_rate": 4.108108108108108e-06,
      "loss": 0.1937,
      "step": 58860
    },
    {
      "epoch": 176.79,
      "grad_norm": 10.622339248657227,
      "learning_rate": 4.107107107107107e-06,
      "loss": 0.2573,
      "step": 58870
    },
    {
      "epoch": 176.82,
      "grad_norm": 7.706355094909668,
      "learning_rate": 4.106106106106106e-06,
      "loss": 0.193,
      "step": 58880
    },
    {
      "epoch": 176.85,
      "grad_norm": 8.182690620422363,
      "learning_rate": 4.105105105105106e-06,
      "loss": 0.2315,
      "step": 58890
    },
    {
      "epoch": 176.88,
      "grad_norm": 13.771315574645996,
      "learning_rate": 4.1041041041041045e-06,
      "loss": 0.2232,
      "step": 58900
    },
    {
      "epoch": 176.91,
      "grad_norm": 12.260208129882812,
      "learning_rate": 4.103103103103103e-06,
      "loss": 0.2475,
      "step": 58910
    },
    {
      "epoch": 176.94,
      "grad_norm": 10.296512603759766,
      "learning_rate": 4.1021021021021025e-06,
      "loss": 0.2067,
      "step": 58920
    },
    {
      "epoch": 176.97,
      "grad_norm": 8.099014282226562,
      "learning_rate": 4.101101101101101e-06,
      "loss": 0.2197,
      "step": 58930
    },
    {
      "epoch": 177.0,
      "grad_norm": 7.086528778076172,
      "learning_rate": 4.1001001001001005e-06,
      "loss": 0.2548,
      "step": 58940
    },
    {
      "epoch": 177.0,
      "eval_accuracy": 0.9217,
      "eval_loss": 0.3389529883861542,
      "eval_runtime": 29.9925,
      "eval_samples_per_second": 333.417,
      "eval_steps_per_second": 1.334,
      "step": 58941
    },
    {
      "epoch": 177.03,
      "grad_norm": 7.3199076652526855,
      "learning_rate": 4.099099099099099e-06,
      "loss": 0.1836,
      "step": 58950
    },
    {
      "epoch": 177.06,
      "grad_norm": 7.64718770980835,
      "learning_rate": 4.0980980980980986e-06,
      "loss": 0.221,
      "step": 58960
    },
    {
      "epoch": 177.09,
      "grad_norm": 20.628135681152344,
      "learning_rate": 4.097097097097097e-06,
      "loss": 0.2398,
      "step": 58970
    },
    {
      "epoch": 177.12,
      "grad_norm": 13.729960441589355,
      "learning_rate": 4.096096096096097e-06,
      "loss": 0.2174,
      "step": 58980
    },
    {
      "epoch": 177.15,
      "grad_norm": 6.7833075523376465,
      "learning_rate": 4.095095095095095e-06,
      "loss": 0.1935,
      "step": 58990
    },
    {
      "epoch": 177.18,
      "grad_norm": 10.577786445617676,
      "learning_rate": 4.094094094094095e-06,
      "loss": 0.2103,
      "step": 59000
    },
    {
      "epoch": 177.21,
      "grad_norm": 10.048870086669922,
      "learning_rate": 4.093093093093093e-06,
      "loss": 0.1846,
      "step": 59010
    },
    {
      "epoch": 177.24,
      "grad_norm": 8.68513011932373,
      "learning_rate": 4.092092092092093e-06,
      "loss": 0.1663,
      "step": 59020
    },
    {
      "epoch": 177.27,
      "grad_norm": 18.999998092651367,
      "learning_rate": 4.091091091091091e-06,
      "loss": 0.2548,
      "step": 59030
    },
    {
      "epoch": 177.3,
      "grad_norm": 15.028038024902344,
      "learning_rate": 4.09009009009009e-06,
      "loss": 0.2139,
      "step": 59040
    },
    {
      "epoch": 177.33,
      "grad_norm": 12.836033821105957,
      "learning_rate": 4.089089089089089e-06,
      "loss": 0.1932,
      "step": 59050
    },
    {
      "epoch": 177.36,
      "grad_norm": 8.610918045043945,
      "learning_rate": 4.088088088088089e-06,
      "loss": 0.2052,
      "step": 59060
    },
    {
      "epoch": 177.39,
      "grad_norm": 7.272721767425537,
      "learning_rate": 4.087087087087087e-06,
      "loss": 0.2018,
      "step": 59070
    },
    {
      "epoch": 177.42,
      "grad_norm": 18.922649383544922,
      "learning_rate": 4.086086086086086e-06,
      "loss": 0.2206,
      "step": 59080
    },
    {
      "epoch": 177.45,
      "grad_norm": 10.191604614257812,
      "learning_rate": 4.085085085085085e-06,
      "loss": 0.2078,
      "step": 59090
    },
    {
      "epoch": 177.48,
      "grad_norm": 10.828603744506836,
      "learning_rate": 4.084084084084085e-06,
      "loss": 0.1968,
      "step": 59100
    },
    {
      "epoch": 177.51,
      "grad_norm": 9.657793998718262,
      "learning_rate": 4.083083083083083e-06,
      "loss": 0.2059,
      "step": 59110
    },
    {
      "epoch": 177.54,
      "grad_norm": 9.794089317321777,
      "learning_rate": 4.082082082082082e-06,
      "loss": 0.2242,
      "step": 59120
    },
    {
      "epoch": 177.57,
      "grad_norm": 16.47228240966797,
      "learning_rate": 4.0810810810810815e-06,
      "loss": 0.2188,
      "step": 59130
    },
    {
      "epoch": 177.6,
      "grad_norm": 5.680071830749512,
      "learning_rate": 4.080080080080081e-06,
      "loss": 0.1738,
      "step": 59140
    },
    {
      "epoch": 177.63,
      "grad_norm": 7.945130825042725,
      "learning_rate": 4.0790790790790795e-06,
      "loss": 0.1941,
      "step": 59150
    },
    {
      "epoch": 177.66,
      "grad_norm": 8.214943885803223,
      "learning_rate": 4.078078078078078e-06,
      "loss": 0.1832,
      "step": 59160
    },
    {
      "epoch": 177.69,
      "grad_norm": 9.845799446105957,
      "learning_rate": 4.0770770770770775e-06,
      "loss": 0.2402,
      "step": 59170
    },
    {
      "epoch": 177.72,
      "grad_norm": 19.164531707763672,
      "learning_rate": 4.076076076076076e-06,
      "loss": 0.2384,
      "step": 59180
    },
    {
      "epoch": 177.75,
      "grad_norm": 14.700833320617676,
      "learning_rate": 4.0750750750750756e-06,
      "loss": 0.2234,
      "step": 59190
    },
    {
      "epoch": 177.78,
      "grad_norm": 11.827143669128418,
      "learning_rate": 4.074074074074074e-06,
      "loss": 0.2007,
      "step": 59200
    },
    {
      "epoch": 177.81,
      "grad_norm": 8.810311317443848,
      "learning_rate": 4.073073073073074e-06,
      "loss": 0.2055,
      "step": 59210
    },
    {
      "epoch": 177.84,
      "grad_norm": 13.449331283569336,
      "learning_rate": 4.072072072072072e-06,
      "loss": 0.19,
      "step": 59220
    },
    {
      "epoch": 177.87,
      "grad_norm": 10.002969741821289,
      "learning_rate": 4.071071071071072e-06,
      "loss": 0.1863,
      "step": 59230
    },
    {
      "epoch": 177.9,
      "grad_norm": 11.407214164733887,
      "learning_rate": 4.07007007007007e-06,
      "loss": 0.252,
      "step": 59240
    },
    {
      "epoch": 177.93,
      "grad_norm": 5.781986713409424,
      "learning_rate": 4.06906906906907e-06,
      "loss": 0.1644,
      "step": 59250
    },
    {
      "epoch": 177.96,
      "grad_norm": 7.590579032897949,
      "learning_rate": 4.068068068068068e-06,
      "loss": 0.2519,
      "step": 59260
    },
    {
      "epoch": 177.99,
      "grad_norm": 10.626959800720215,
      "learning_rate": 4.067067067067068e-06,
      "loss": 0.1797,
      "step": 59270
    },
    {
      "epoch": 178.0,
      "eval_accuracy": 0.9214,
      "eval_loss": 0.34331947565078735,
      "eval_runtime": 30.4922,
      "eval_samples_per_second": 327.952,
      "eval_steps_per_second": 1.312,
      "step": 59274
    },
    {
      "epoch": 178.02,
      "grad_norm": 8.529282569885254,
      "learning_rate": 4.066066066066066e-06,
      "loss": 0.3306,
      "step": 59280
    },
    {
      "epoch": 178.05,
      "grad_norm": 10.412162780761719,
      "learning_rate": 4.065065065065065e-06,
      "loss": 0.2473,
      "step": 59290
    },
    {
      "epoch": 178.08,
      "grad_norm": 9.110191345214844,
      "learning_rate": 4.064064064064064e-06,
      "loss": 0.2041,
      "step": 59300
    },
    {
      "epoch": 178.11,
      "grad_norm": 9.971551895141602,
      "learning_rate": 4.063063063063064e-06,
      "loss": 0.2335,
      "step": 59310
    },
    {
      "epoch": 178.14,
      "grad_norm": 9.253573417663574,
      "learning_rate": 4.062062062062062e-06,
      "loss": 0.1916,
      "step": 59320
    },
    {
      "epoch": 178.17,
      "grad_norm": 6.387475490570068,
      "learning_rate": 4.061061061061061e-06,
      "loss": 0.1849,
      "step": 59330
    },
    {
      "epoch": 178.2,
      "grad_norm": 10.173002243041992,
      "learning_rate": 4.06006006006006e-06,
      "loss": 0.2252,
      "step": 59340
    },
    {
      "epoch": 178.23,
      "grad_norm": 5.404430389404297,
      "learning_rate": 4.05905905905906e-06,
      "loss": 0.2096,
      "step": 59350
    },
    {
      "epoch": 178.26,
      "grad_norm": 9.724614143371582,
      "learning_rate": 4.0580580580580584e-06,
      "loss": 0.2264,
      "step": 59360
    },
    {
      "epoch": 178.29,
      "grad_norm": 11.656150817871094,
      "learning_rate": 4.057057057057057e-06,
      "loss": 0.2229,
      "step": 59370
    },
    {
      "epoch": 178.32,
      "grad_norm": 48.839263916015625,
      "learning_rate": 4.0560560560560565e-06,
      "loss": 0.2639,
      "step": 59380
    },
    {
      "epoch": 178.35,
      "grad_norm": 6.930693626403809,
      "learning_rate": 4.055055055055055e-06,
      "loss": 0.183,
      "step": 59390
    },
    {
      "epoch": 178.38,
      "grad_norm": 37.019775390625,
      "learning_rate": 4.0540540540540545e-06,
      "loss": 0.2237,
      "step": 59400
    },
    {
      "epoch": 178.41,
      "grad_norm": 6.963565349578857,
      "learning_rate": 4.053053053053053e-06,
      "loss": 0.1858,
      "step": 59410
    },
    {
      "epoch": 178.44,
      "grad_norm": 11.594789505004883,
      "learning_rate": 4.0520520520520526e-06,
      "loss": 0.2202,
      "step": 59420
    },
    {
      "epoch": 178.47,
      "grad_norm": 18.96141242980957,
      "learning_rate": 4.051051051051051e-06,
      "loss": 0.2285,
      "step": 59430
    },
    {
      "epoch": 178.5,
      "grad_norm": 9.03508472442627,
      "learning_rate": 4.050050050050051e-06,
      "loss": 0.2149,
      "step": 59440
    },
    {
      "epoch": 178.53,
      "grad_norm": 16.72518539428711,
      "learning_rate": 4.049049049049049e-06,
      "loss": 0.2524,
      "step": 59450
    },
    {
      "epoch": 178.56,
      "grad_norm": 7.688474655151367,
      "learning_rate": 4.048048048048049e-06,
      "loss": 0.2346,
      "step": 59460
    },
    {
      "epoch": 178.59,
      "grad_norm": 15.503273963928223,
      "learning_rate": 4.047047047047047e-06,
      "loss": 0.1965,
      "step": 59470
    },
    {
      "epoch": 178.62,
      "grad_norm": 7.474960803985596,
      "learning_rate": 4.046046046046047e-06,
      "loss": 0.2379,
      "step": 59480
    },
    {
      "epoch": 178.65,
      "grad_norm": 10.694663047790527,
      "learning_rate": 4.045045045045045e-06,
      "loss": 0.2162,
      "step": 59490
    },
    {
      "epoch": 178.68,
      "grad_norm": 7.5724005699157715,
      "learning_rate": 4.044044044044044e-06,
      "loss": 0.2043,
      "step": 59500
    },
    {
      "epoch": 178.71,
      "grad_norm": 7.494312286376953,
      "learning_rate": 4.043043043043043e-06,
      "loss": 0.2257,
      "step": 59510
    },
    {
      "epoch": 178.74,
      "grad_norm": 8.567968368530273,
      "learning_rate": 4.042042042042043e-06,
      "loss": 0.208,
      "step": 59520
    },
    {
      "epoch": 178.77,
      "grad_norm": 8.073162078857422,
      "learning_rate": 4.041041041041041e-06,
      "loss": 0.2191,
      "step": 59530
    },
    {
      "epoch": 178.8,
      "grad_norm": 6.795145511627197,
      "learning_rate": 4.04004004004004e-06,
      "loss": 0.2223,
      "step": 59540
    },
    {
      "epoch": 178.83,
      "grad_norm": 11.85008716583252,
      "learning_rate": 4.039039039039039e-06,
      "loss": 0.2338,
      "step": 59550
    },
    {
      "epoch": 178.86,
      "grad_norm": 6.526969909667969,
      "learning_rate": 4.038038038038039e-06,
      "loss": 0.2383,
      "step": 59560
    },
    {
      "epoch": 178.89,
      "grad_norm": 16.605377197265625,
      "learning_rate": 4.037037037037037e-06,
      "loss": 0.2272,
      "step": 59570
    },
    {
      "epoch": 178.92,
      "grad_norm": 5.397552490234375,
      "learning_rate": 4.036036036036036e-06,
      "loss": 0.209,
      "step": 59580
    },
    {
      "epoch": 178.95,
      "grad_norm": 8.512609481811523,
      "learning_rate": 4.0350350350350354e-06,
      "loss": 0.2173,
      "step": 59590
    },
    {
      "epoch": 178.98,
      "grad_norm": 8.097075462341309,
      "learning_rate": 4.034034034034035e-06,
      "loss": 0.235,
      "step": 59600
    },
    {
      "epoch": 179.0,
      "eval_accuracy": 0.9206,
      "eval_loss": 0.33879607915878296,
      "eval_runtime": 30.6555,
      "eval_samples_per_second": 326.206,
      "eval_steps_per_second": 1.305,
      "step": 59607
    },
    {
      "epoch": 179.01,
      "grad_norm": 10.087231636047363,
      "learning_rate": 4.0330330330330335e-06,
      "loss": 0.2245,
      "step": 59610
    },
    {
      "epoch": 179.04,
      "grad_norm": 19.084575653076172,
      "learning_rate": 4.032032032032032e-06,
      "loss": 0.2383,
      "step": 59620
    },
    {
      "epoch": 179.07,
      "grad_norm": 9.595450401306152,
      "learning_rate": 4.0310310310310315e-06,
      "loss": 0.1992,
      "step": 59630
    },
    {
      "epoch": 179.1,
      "grad_norm": 9.152299880981445,
      "learning_rate": 4.03003003003003e-06,
      "loss": 0.1861,
      "step": 59640
    },
    {
      "epoch": 179.13,
      "grad_norm": 7.202277183532715,
      "learning_rate": 4.0290290290290296e-06,
      "loss": 0.2321,
      "step": 59650
    },
    {
      "epoch": 179.16,
      "grad_norm": 10.016426086425781,
      "learning_rate": 4.028028028028028e-06,
      "loss": 0.2256,
      "step": 59660
    },
    {
      "epoch": 179.19,
      "grad_norm": 8.094841957092285,
      "learning_rate": 4.027027027027028e-06,
      "loss": 0.2394,
      "step": 59670
    },
    {
      "epoch": 179.22,
      "grad_norm": 10.248344421386719,
      "learning_rate": 4.026026026026026e-06,
      "loss": 0.213,
      "step": 59680
    },
    {
      "epoch": 179.25,
      "grad_norm": 15.668526649475098,
      "learning_rate": 4.025025025025026e-06,
      "loss": 0.2182,
      "step": 59690
    },
    {
      "epoch": 179.28,
      "grad_norm": 8.220621109008789,
      "learning_rate": 4.024024024024024e-06,
      "loss": 0.2292,
      "step": 59700
    },
    {
      "epoch": 179.31,
      "grad_norm": 8.333517074584961,
      "learning_rate": 4.023023023023024e-06,
      "loss": 0.2177,
      "step": 59710
    },
    {
      "epoch": 179.34,
      "grad_norm": 9.700437545776367,
      "learning_rate": 4.022022022022022e-06,
      "loss": 0.2525,
      "step": 59720
    },
    {
      "epoch": 179.37,
      "grad_norm": 32.850154876708984,
      "learning_rate": 4.021021021021022e-06,
      "loss": 0.2316,
      "step": 59730
    },
    {
      "epoch": 179.4,
      "grad_norm": 11.458312034606934,
      "learning_rate": 4.02002002002002e-06,
      "loss": 0.237,
      "step": 59740
    },
    {
      "epoch": 179.43,
      "grad_norm": 12.506278991699219,
      "learning_rate": 4.019019019019019e-06,
      "loss": 0.2289,
      "step": 59750
    },
    {
      "epoch": 179.46,
      "grad_norm": 7.115773677825928,
      "learning_rate": 4.018018018018018e-06,
      "loss": 0.246,
      "step": 59760
    },
    {
      "epoch": 179.49,
      "grad_norm": 8.133695602416992,
      "learning_rate": 4.017017017017018e-06,
      "loss": 0.1636,
      "step": 59770
    },
    {
      "epoch": 179.52,
      "grad_norm": 7.627388000488281,
      "learning_rate": 4.016016016016016e-06,
      "loss": 0.239,
      "step": 59780
    },
    {
      "epoch": 179.55,
      "grad_norm": 7.677376747131348,
      "learning_rate": 4.015015015015015e-06,
      "loss": 0.2105,
      "step": 59790
    },
    {
      "epoch": 179.58,
      "grad_norm": 8.592795372009277,
      "learning_rate": 4.014014014014014e-06,
      "loss": 0.2275,
      "step": 59800
    },
    {
      "epoch": 179.61,
      "grad_norm": 5.632089614868164,
      "learning_rate": 4.013013013013014e-06,
      "loss": 0.2263,
      "step": 59810
    },
    {
      "epoch": 179.64,
      "grad_norm": 8.967724800109863,
      "learning_rate": 4.0120120120120124e-06,
      "loss": 0.1567,
      "step": 59820
    },
    {
      "epoch": 179.67,
      "grad_norm": 7.6224775314331055,
      "learning_rate": 4.011011011011011e-06,
      "loss": 0.2283,
      "step": 59830
    },
    {
      "epoch": 179.7,
      "grad_norm": 8.368711471557617,
      "learning_rate": 4.0100100100100105e-06,
      "loss": 0.2375,
      "step": 59840
    },
    {
      "epoch": 179.73,
      "grad_norm": 11.613204002380371,
      "learning_rate": 4.009009009009009e-06,
      "loss": 0.2316,
      "step": 59850
    },
    {
      "epoch": 179.76,
      "grad_norm": 6.122308254241943,
      "learning_rate": 4.0080080080080085e-06,
      "loss": 0.2348,
      "step": 59860
    },
    {
      "epoch": 179.79,
      "grad_norm": 7.3094096183776855,
      "learning_rate": 4.007007007007007e-06,
      "loss": 0.2187,
      "step": 59870
    },
    {
      "epoch": 179.82,
      "grad_norm": 9.253562927246094,
      "learning_rate": 4.0060060060060066e-06,
      "loss": 0.2265,
      "step": 59880
    },
    {
      "epoch": 179.85,
      "grad_norm": 9.863082885742188,
      "learning_rate": 4.005005005005005e-06,
      "loss": 0.213,
      "step": 59890
    },
    {
      "epoch": 179.88,
      "grad_norm": 7.997868537902832,
      "learning_rate": 4.004004004004005e-06,
      "loss": 0.2211,
      "step": 59900
    },
    {
      "epoch": 179.91,
      "grad_norm": 15.715087890625,
      "learning_rate": 4.003003003003003e-06,
      "loss": 0.2267,
      "step": 59910
    },
    {
      "epoch": 179.94,
      "grad_norm": 9.478219985961914,
      "learning_rate": 4.002002002002003e-06,
      "loss": 0.2153,
      "step": 59920
    },
    {
      "epoch": 179.97,
      "grad_norm": 9.782516479492188,
      "learning_rate": 4.001001001001001e-06,
      "loss": 0.2603,
      "step": 59930
    },
    {
      "epoch": 180.0,
      "grad_norm": 58.13569259643555,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.1707,
      "step": 59940
    },
    {
      "epoch": 180.0,
      "eval_accuracy": 0.9209,
      "eval_loss": 0.3375724256038666,
      "eval_runtime": 29.9616,
      "eval_samples_per_second": 333.761,
      "eval_steps_per_second": 1.335,
      "step": 59940
    },
    {
      "epoch": 180.03,
      "grad_norm": 13.381095886230469,
      "learning_rate": 3.998998998998999e-06,
      "loss": 0.2218,
      "step": 59950
    },
    {
      "epoch": 180.06,
      "grad_norm": 6.267689228057861,
      "learning_rate": 3.997997997997998e-06,
      "loss": 0.2479,
      "step": 59960
    },
    {
      "epoch": 180.09,
      "grad_norm": 8.08096694946289,
      "learning_rate": 3.996996996996997e-06,
      "loss": 0.23,
      "step": 59970
    },
    {
      "epoch": 180.12,
      "grad_norm": 7.929141521453857,
      "learning_rate": 3.995995995995997e-06,
      "loss": 0.2089,
      "step": 59980
    },
    {
      "epoch": 180.15,
      "grad_norm": 6.5121941566467285,
      "learning_rate": 3.994994994994995e-06,
      "loss": 0.2086,
      "step": 59990
    },
    {
      "epoch": 180.18,
      "grad_norm": 20.89447021484375,
      "learning_rate": 3.993993993993994e-06,
      "loss": 0.2154,
      "step": 60000
    },
    {
      "epoch": 180.21,
      "grad_norm": 8.740914344787598,
      "learning_rate": 3.992992992992993e-06,
      "loss": 0.2032,
      "step": 60010
    },
    {
      "epoch": 180.24,
      "grad_norm": 6.458169937133789,
      "learning_rate": 3.991991991991993e-06,
      "loss": 0.2267,
      "step": 60020
    },
    {
      "epoch": 180.27,
      "grad_norm": 7.842488765716553,
      "learning_rate": 3.990990990990991e-06,
      "loss": 0.199,
      "step": 60030
    },
    {
      "epoch": 180.3,
      "grad_norm": 8.298648834228516,
      "learning_rate": 3.98998998998999e-06,
      "loss": 0.2411,
      "step": 60040
    },
    {
      "epoch": 180.33,
      "grad_norm": 6.832301139831543,
      "learning_rate": 3.9889889889889894e-06,
      "loss": 0.201,
      "step": 60050
    },
    {
      "epoch": 180.36,
      "grad_norm": 4.882878303527832,
      "learning_rate": 3.987987987987989e-06,
      "loss": 0.213,
      "step": 60060
    },
    {
      "epoch": 180.39,
      "grad_norm": 12.647751808166504,
      "learning_rate": 3.986986986986987e-06,
      "loss": 0.2333,
      "step": 60070
    },
    {
      "epoch": 180.42,
      "grad_norm": 6.937251567840576,
      "learning_rate": 3.985985985985986e-06,
      "loss": 0.2435,
      "step": 60080
    },
    {
      "epoch": 180.45,
      "grad_norm": 6.875417232513428,
      "learning_rate": 3.9849849849849855e-06,
      "loss": 0.2204,
      "step": 60090
    },
    {
      "epoch": 180.48,
      "grad_norm": 14.498249053955078,
      "learning_rate": 3.983983983983984e-06,
      "loss": 0.2004,
      "step": 60100
    },
    {
      "epoch": 180.51,
      "grad_norm": 6.21645975112915,
      "learning_rate": 3.9829829829829836e-06,
      "loss": 0.2084,
      "step": 60110
    },
    {
      "epoch": 180.54,
      "grad_norm": 9.458342552185059,
      "learning_rate": 3.981981981981982e-06,
      "loss": 0.2331,
      "step": 60120
    },
    {
      "epoch": 180.57,
      "grad_norm": 7.846961498260498,
      "learning_rate": 3.980980980980982e-06,
      "loss": 0.1953,
      "step": 60130
    },
    {
      "epoch": 180.6,
      "grad_norm": 8.287810325622559,
      "learning_rate": 3.97997997997998e-06,
      "loss": 0.2626,
      "step": 60140
    },
    {
      "epoch": 180.63,
      "grad_norm": 6.0171942710876465,
      "learning_rate": 3.97897897897898e-06,
      "loss": 0.2284,
      "step": 60150
    },
    {
      "epoch": 180.66,
      "grad_norm": 20.226173400878906,
      "learning_rate": 3.977977977977978e-06,
      "loss": 0.2193,
      "step": 60160
    },
    {
      "epoch": 180.69,
      "grad_norm": 7.305109977722168,
      "learning_rate": 3.976976976976978e-06,
      "loss": 0.2177,
      "step": 60170
    },
    {
      "epoch": 180.72,
      "grad_norm": 7.582116603851318,
      "learning_rate": 3.975975975975976e-06,
      "loss": 0.2002,
      "step": 60180
    },
    {
      "epoch": 180.75,
      "grad_norm": 7.93876314163208,
      "learning_rate": 3.974974974974976e-06,
      "loss": 0.2682,
      "step": 60190
    },
    {
      "epoch": 180.78,
      "grad_norm": 8.2844820022583,
      "learning_rate": 3.973973973973974e-06,
      "loss": 0.2387,
      "step": 60200
    },
    {
      "epoch": 180.81,
      "grad_norm": 8.916633605957031,
      "learning_rate": 3.972972972972973e-06,
      "loss": 0.2002,
      "step": 60210
    },
    {
      "epoch": 180.84,
      "grad_norm": 7.271007061004639,
      "learning_rate": 3.971971971971972e-06,
      "loss": 0.2357,
      "step": 60220
    },
    {
      "epoch": 180.87,
      "grad_norm": 9.915923118591309,
      "learning_rate": 3.970970970970972e-06,
      "loss": 0.2225,
      "step": 60230
    },
    {
      "epoch": 180.9,
      "grad_norm": 8.463702201843262,
      "learning_rate": 3.96996996996997e-06,
      "loss": 0.2449,
      "step": 60240
    },
    {
      "epoch": 180.93,
      "grad_norm": 14.428045272827148,
      "learning_rate": 3.968968968968969e-06,
      "loss": 0.2094,
      "step": 60250
    },
    {
      "epoch": 180.96,
      "grad_norm": 9.99558162689209,
      "learning_rate": 3.967967967967968e-06,
      "loss": 0.2288,
      "step": 60260
    },
    {
      "epoch": 180.99,
      "grad_norm": 6.686193943023682,
      "learning_rate": 3.966966966966968e-06,
      "loss": 0.195,
      "step": 60270
    },
    {
      "epoch": 181.0,
      "eval_accuracy": 0.9224,
      "eval_loss": 0.33835887908935547,
      "eval_runtime": 30.0866,
      "eval_samples_per_second": 332.374,
      "eval_steps_per_second": 1.329,
      "step": 60273
    },
    {
      "epoch": 181.02,
      "grad_norm": 16.327045440673828,
      "learning_rate": 3.9659659659659664e-06,
      "loss": 0.2651,
      "step": 60280
    },
    {
      "epoch": 181.05,
      "grad_norm": 6.203663349151611,
      "learning_rate": 3.964964964964965e-06,
      "loss": 0.1816,
      "step": 60290
    },
    {
      "epoch": 181.08,
      "grad_norm": 5.6692795753479,
      "learning_rate": 3.9639639639639645e-06,
      "loss": 0.1643,
      "step": 60300
    },
    {
      "epoch": 181.11,
      "grad_norm": 12.413482666015625,
      "learning_rate": 3.962962962962963e-06,
      "loss": 0.2208,
      "step": 60310
    },
    {
      "epoch": 181.14,
      "grad_norm": 3.3665096759796143,
      "learning_rate": 3.961961961961962e-06,
      "loss": 0.1986,
      "step": 60320
    },
    {
      "epoch": 181.17,
      "grad_norm": 10.564810752868652,
      "learning_rate": 3.960960960960961e-06,
      "loss": 0.222,
      "step": 60330
    },
    {
      "epoch": 181.2,
      "grad_norm": 5.598283767700195,
      "learning_rate": 3.9599599599599606e-06,
      "loss": 0.2457,
      "step": 60340
    },
    {
      "epoch": 181.23,
      "grad_norm": 11.108146667480469,
      "learning_rate": 3.958958958958959e-06,
      "loss": 0.2156,
      "step": 60350
    },
    {
      "epoch": 181.26,
      "grad_norm": 8.657574653625488,
      "learning_rate": 3.957957957957958e-06,
      "loss": 0.2415,
      "step": 60360
    },
    {
      "epoch": 181.29,
      "grad_norm": 7.5750956535339355,
      "learning_rate": 3.956956956956957e-06,
      "loss": 0.22,
      "step": 60370
    },
    {
      "epoch": 181.32,
      "grad_norm": 8.133354187011719,
      "learning_rate": 3.955955955955957e-06,
      "loss": 0.2408,
      "step": 60380
    },
    {
      "epoch": 181.35,
      "grad_norm": 11.7205810546875,
      "learning_rate": 3.954954954954955e-06,
      "loss": 0.228,
      "step": 60390
    },
    {
      "epoch": 181.38,
      "grad_norm": 7.208806991577148,
      "learning_rate": 3.953953953953955e-06,
      "loss": 0.2451,
      "step": 60400
    },
    {
      "epoch": 181.41,
      "grad_norm": 18.9415225982666,
      "learning_rate": 3.952952952952953e-06,
      "loss": 0.2163,
      "step": 60410
    },
    {
      "epoch": 181.44,
      "grad_norm": 11.653281211853027,
      "learning_rate": 3.951951951951952e-06,
      "loss": 0.2319,
      "step": 60420
    },
    {
      "epoch": 181.47,
      "grad_norm": 9.192414283752441,
      "learning_rate": 3.950950950950951e-06,
      "loss": 0.2308,
      "step": 60430
    },
    {
      "epoch": 181.5,
      "grad_norm": 7.707842826843262,
      "learning_rate": 3.949949949949951e-06,
      "loss": 0.2234,
      "step": 60440
    },
    {
      "epoch": 181.53,
      "grad_norm": 10.01969051361084,
      "learning_rate": 3.948948948948949e-06,
      "loss": 0.2081,
      "step": 60450
    },
    {
      "epoch": 181.56,
      "grad_norm": 8.208342552185059,
      "learning_rate": 3.947947947947948e-06,
      "loss": 0.2131,
      "step": 60460
    },
    {
      "epoch": 181.59,
      "grad_norm": 5.674446105957031,
      "learning_rate": 3.946946946946947e-06,
      "loss": 0.2149,
      "step": 60470
    },
    {
      "epoch": 181.62,
      "grad_norm": 12.250094413757324,
      "learning_rate": 3.945945945945947e-06,
      "loss": 0.1859,
      "step": 60480
    },
    {
      "epoch": 181.65,
      "grad_norm": 9.286977767944336,
      "learning_rate": 3.944944944944945e-06,
      "loss": 0.2095,
      "step": 60490
    },
    {
      "epoch": 181.68,
      "grad_norm": 6.696424961090088,
      "learning_rate": 3.943943943943944e-06,
      "loss": 0.1965,
      "step": 60500
    },
    {
      "epoch": 181.71,
      "grad_norm": 7.639163970947266,
      "learning_rate": 3.9429429429429434e-06,
      "loss": 0.2312,
      "step": 60510
    },
    {
      "epoch": 181.74,
      "grad_norm": 4.874849796295166,
      "learning_rate": 3.941941941941943e-06,
      "loss": 0.2196,
      "step": 60520
    },
    {
      "epoch": 181.77,
      "grad_norm": 9.009918212890625,
      "learning_rate": 3.940940940940941e-06,
      "loss": 0.2114,
      "step": 60530
    },
    {
      "epoch": 181.8,
      "grad_norm": 6.338069915771484,
      "learning_rate": 3.93993993993994e-06,
      "loss": 0.2395,
      "step": 60540
    },
    {
      "epoch": 181.83,
      "grad_norm": 8.148040771484375,
      "learning_rate": 3.9389389389389395e-06,
      "loss": 0.1621,
      "step": 60550
    },
    {
      "epoch": 181.86,
      "grad_norm": 6.775726795196533,
      "learning_rate": 3.937937937937938e-06,
      "loss": 0.2072,
      "step": 60560
    },
    {
      "epoch": 181.89,
      "grad_norm": 6.170883655548096,
      "learning_rate": 3.936936936936937e-06,
      "loss": 0.21,
      "step": 60570
    },
    {
      "epoch": 181.92,
      "grad_norm": 14.370410919189453,
      "learning_rate": 3.935935935935936e-06,
      "loss": 0.2274,
      "step": 60580
    },
    {
      "epoch": 181.95,
      "grad_norm": 15.545294761657715,
      "learning_rate": 3.934934934934936e-06,
      "loss": 0.2516,
      "step": 60590
    },
    {
      "epoch": 181.98,
      "grad_norm": 9.852826118469238,
      "learning_rate": 3.933933933933934e-06,
      "loss": 0.2526,
      "step": 60600
    },
    {
      "epoch": 182.0,
      "eval_accuracy": 0.9215,
      "eval_loss": 0.34183624386787415,
      "eval_runtime": 30.374,
      "eval_samples_per_second": 329.229,
      "eval_steps_per_second": 1.317,
      "step": 60606
    },
    {
      "epoch": 182.01,
      "grad_norm": 8.677170753479004,
      "learning_rate": 3.932932932932933e-06,
      "loss": 0.269,
      "step": 60610
    },
    {
      "epoch": 182.04,
      "grad_norm": 5.8057050704956055,
      "learning_rate": 3.931931931931932e-06,
      "loss": 0.2425,
      "step": 60620
    },
    {
      "epoch": 182.07,
      "grad_norm": 12.525891304016113,
      "learning_rate": 3.930930930930932e-06,
      "loss": 0.1887,
      "step": 60630
    },
    {
      "epoch": 182.1,
      "grad_norm": 11.621131896972656,
      "learning_rate": 3.92992992992993e-06,
      "loss": 0.2263,
      "step": 60640
    },
    {
      "epoch": 182.13,
      "grad_norm": 13.482061386108398,
      "learning_rate": 3.92892892892893e-06,
      "loss": 0.2342,
      "step": 60650
    },
    {
      "epoch": 182.16,
      "grad_norm": 10.494664192199707,
      "learning_rate": 3.927927927927928e-06,
      "loss": 0.23,
      "step": 60660
    },
    {
      "epoch": 182.19,
      "grad_norm": 6.603015899658203,
      "learning_rate": 3.926926926926927e-06,
      "loss": 0.2393,
      "step": 60670
    },
    {
      "epoch": 182.22,
      "grad_norm": 8.237621307373047,
      "learning_rate": 3.925925925925926e-06,
      "loss": 0.2588,
      "step": 60680
    },
    {
      "epoch": 182.25,
      "grad_norm": 4.335064888000488,
      "learning_rate": 3.924924924924926e-06,
      "loss": 0.1923,
      "step": 60690
    },
    {
      "epoch": 182.28,
      "grad_norm": 8.338323593139648,
      "learning_rate": 3.923923923923924e-06,
      "loss": 0.1901,
      "step": 60700
    },
    {
      "epoch": 182.31,
      "grad_norm": 6.76269006729126,
      "learning_rate": 3.922922922922923e-06,
      "loss": 0.2257,
      "step": 60710
    },
    {
      "epoch": 182.34,
      "grad_norm": 8.758835792541504,
      "learning_rate": 3.921921921921922e-06,
      "loss": 0.2149,
      "step": 60720
    },
    {
      "epoch": 182.37,
      "grad_norm": 9.589883804321289,
      "learning_rate": 3.920920920920922e-06,
      "loss": 0.1937,
      "step": 60730
    },
    {
      "epoch": 182.4,
      "grad_norm": 8.92021656036377,
      "learning_rate": 3.9199199199199204e-06,
      "loss": 0.1853,
      "step": 60740
    },
    {
      "epoch": 182.43,
      "grad_norm": 10.833581924438477,
      "learning_rate": 3.918918918918919e-06,
      "loss": 0.2415,
      "step": 60750
    },
    {
      "epoch": 182.46,
      "grad_norm": 15.926843643188477,
      "learning_rate": 3.9179179179179185e-06,
      "loss": 0.3053,
      "step": 60760
    },
    {
      "epoch": 182.49,
      "grad_norm": 7.022651672363281,
      "learning_rate": 3.916916916916917e-06,
      "loss": 0.1678,
      "step": 60770
    },
    {
      "epoch": 182.52,
      "grad_norm": 10.216519355773926,
      "learning_rate": 3.915915915915916e-06,
      "loss": 0.2012,
      "step": 60780
    },
    {
      "epoch": 182.55,
      "grad_norm": 9.211398124694824,
      "learning_rate": 3.914914914914915e-06,
      "loss": 0.2166,
      "step": 60790
    },
    {
      "epoch": 182.58,
      "grad_norm": 11.354323387145996,
      "learning_rate": 3.9139139139139145e-06,
      "loss": 0.2187,
      "step": 60800
    },
    {
      "epoch": 182.61,
      "grad_norm": 7.167208671569824,
      "learning_rate": 3.912912912912913e-06,
      "loss": 0.2245,
      "step": 60810
    },
    {
      "epoch": 182.64,
      "grad_norm": 5.70239782333374,
      "learning_rate": 3.911911911911912e-06,
      "loss": 0.195,
      "step": 60820
    },
    {
      "epoch": 182.67,
      "grad_norm": 14.310766220092773,
      "learning_rate": 3.910910910910911e-06,
      "loss": 0.2251,
      "step": 60830
    },
    {
      "epoch": 182.7,
      "grad_norm": 24.64097785949707,
      "learning_rate": 3.909909909909911e-06,
      "loss": 0.2476,
      "step": 60840
    },
    {
      "epoch": 182.73,
      "grad_norm": 9.189980506896973,
      "learning_rate": 3.908908908908909e-06,
      "loss": 0.1749,
      "step": 60850
    },
    {
      "epoch": 182.76,
      "grad_norm": 9.230351448059082,
      "learning_rate": 3.907907907907908e-06,
      "loss": 0.2658,
      "step": 60860
    },
    {
      "epoch": 182.79,
      "grad_norm": 6.974701881408691,
      "learning_rate": 3.906906906906907e-06,
      "loss": 0.2154,
      "step": 60870
    },
    {
      "epoch": 182.82,
      "grad_norm": 13.052568435668945,
      "learning_rate": 3.905905905905906e-06,
      "loss": 0.2669,
      "step": 60880
    },
    {
      "epoch": 182.85,
      "grad_norm": 7.0593791007995605,
      "learning_rate": 3.904904904904905e-06,
      "loss": 0.1731,
      "step": 60890
    },
    {
      "epoch": 182.88,
      "grad_norm": 8.249703407287598,
      "learning_rate": 3.903903903903904e-06,
      "loss": 0.219,
      "step": 60900
    },
    {
      "epoch": 182.91,
      "grad_norm": 10.794869422912598,
      "learning_rate": 3.902902902902903e-06,
      "loss": 0.243,
      "step": 60910
    },
    {
      "epoch": 182.94,
      "grad_norm": 14.009035110473633,
      "learning_rate": 3.901901901901902e-06,
      "loss": 0.234,
      "step": 60920
    },
    {
      "epoch": 182.97,
      "grad_norm": 10.57756519317627,
      "learning_rate": 3.900900900900901e-06,
      "loss": 0.2041,
      "step": 60930
    },
    {
      "epoch": 183.0,
      "eval_accuracy": 0.921,
      "eval_loss": 0.3372883200645447,
      "eval_runtime": 29.9017,
      "eval_samples_per_second": 334.429,
      "eval_steps_per_second": 1.338,
      "step": 60939
    },
    {
      "epoch": 183.0,
      "grad_norm": 10.084001541137695,
      "learning_rate": 3.899899899899901e-06,
      "loss": 0.2043,
      "step": 60940
    },
    {
      "epoch": 183.03,
      "grad_norm": 9.689692497253418,
      "learning_rate": 3.898898898898899e-06,
      "loss": 0.2021,
      "step": 60950
    },
    {
      "epoch": 183.06,
      "grad_norm": 7.617337226867676,
      "learning_rate": 3.897897897897898e-06,
      "loss": 0.1818,
      "step": 60960
    },
    {
      "epoch": 183.09,
      "grad_norm": 8.142428398132324,
      "learning_rate": 3.8968968968968974e-06,
      "loss": 0.2133,
      "step": 60970
    },
    {
      "epoch": 183.12,
      "grad_norm": 5.180631160736084,
      "learning_rate": 3.895895895895896e-06,
      "loss": 0.2004,
      "step": 60980
    },
    {
      "epoch": 183.15,
      "grad_norm": 31.209413528442383,
      "learning_rate": 3.894894894894895e-06,
      "loss": 0.2395,
      "step": 60990
    },
    {
      "epoch": 183.18,
      "grad_norm": 8.140734672546387,
      "learning_rate": 3.893893893893894e-06,
      "loss": 0.1892,
      "step": 61000
    },
    {
      "epoch": 183.21,
      "grad_norm": 31.683191299438477,
      "learning_rate": 3.8928928928928935e-06,
      "loss": 0.2185,
      "step": 61010
    },
    {
      "epoch": 183.24,
      "grad_norm": 6.376895904541016,
      "learning_rate": 3.891891891891892e-06,
      "loss": 0.2222,
      "step": 61020
    },
    {
      "epoch": 183.27,
      "grad_norm": 10.881690979003906,
      "learning_rate": 3.890890890890891e-06,
      "loss": 0.2277,
      "step": 61030
    },
    {
      "epoch": 183.3,
      "grad_norm": 7.870832920074463,
      "learning_rate": 3.88988988988989e-06,
      "loss": 0.2286,
      "step": 61040
    },
    {
      "epoch": 183.33,
      "grad_norm": 11.93307113647461,
      "learning_rate": 3.88888888888889e-06,
      "loss": 0.189,
      "step": 61050
    },
    {
      "epoch": 183.36,
      "grad_norm": 3.252891778945923,
      "learning_rate": 3.887887887887888e-06,
      "loss": 0.1985,
      "step": 61060
    },
    {
      "epoch": 183.39,
      "grad_norm": 6.379495620727539,
      "learning_rate": 3.886886886886887e-06,
      "loss": 0.2022,
      "step": 61070
    },
    {
      "epoch": 183.42,
      "grad_norm": 5.122598648071289,
      "learning_rate": 3.885885885885886e-06,
      "loss": 0.2311,
      "step": 61080
    },
    {
      "epoch": 183.45,
      "grad_norm": 7.109132289886475,
      "learning_rate": 3.884884884884886e-06,
      "loss": 0.2444,
      "step": 61090
    },
    {
      "epoch": 183.48,
      "grad_norm": 7.585558891296387,
      "learning_rate": 3.883883883883884e-06,
      "loss": 0.215,
      "step": 61100
    },
    {
      "epoch": 183.51,
      "grad_norm": 7.487634658813477,
      "learning_rate": 3.882882882882883e-06,
      "loss": 0.1725,
      "step": 61110
    },
    {
      "epoch": 183.54,
      "grad_norm": 7.3074822425842285,
      "learning_rate": 3.881881881881882e-06,
      "loss": 0.2059,
      "step": 61120
    },
    {
      "epoch": 183.57,
      "grad_norm": 10.5478515625,
      "learning_rate": 3.880880880880881e-06,
      "loss": 0.2396,
      "step": 61130
    },
    {
      "epoch": 183.6,
      "grad_norm": 9.743653297424316,
      "learning_rate": 3.87987987987988e-06,
      "loss": 0.2197,
      "step": 61140
    },
    {
      "epoch": 183.63,
      "grad_norm": 6.376415252685547,
      "learning_rate": 3.878878878878879e-06,
      "loss": 0.1713,
      "step": 61150
    },
    {
      "epoch": 183.66,
      "grad_norm": 11.268939971923828,
      "learning_rate": 3.877877877877878e-06,
      "loss": 0.2067,
      "step": 61160
    },
    {
      "epoch": 183.69,
      "grad_norm": 9.316916465759277,
      "learning_rate": 3.876876876876877e-06,
      "loss": 0.2036,
      "step": 61170
    },
    {
      "epoch": 183.72,
      "grad_norm": 10.382431983947754,
      "learning_rate": 3.875875875875876e-06,
      "loss": 0.2622,
      "step": 61180
    },
    {
      "epoch": 183.75,
      "grad_norm": 7.239348411560059,
      "learning_rate": 3.874874874874876e-06,
      "loss": 0.1888,
      "step": 61190
    },
    {
      "epoch": 183.78,
      "grad_norm": 12.036039352416992,
      "learning_rate": 3.8738738738738744e-06,
      "loss": 0.2406,
      "step": 61200
    },
    {
      "epoch": 183.81,
      "grad_norm": 7.5970845222473145,
      "learning_rate": 3.872872872872873e-06,
      "loss": 0.2159,
      "step": 61210
    },
    {
      "epoch": 183.84,
      "grad_norm": 7.887533664703369,
      "learning_rate": 3.8718718718718725e-06,
      "loss": 0.2514,
      "step": 61220
    },
    {
      "epoch": 183.87,
      "grad_norm": 6.6489973068237305,
      "learning_rate": 3.870870870870871e-06,
      "loss": 0.2234,
      "step": 61230
    },
    {
      "epoch": 183.9,
      "grad_norm": 6.0929484367370605,
      "learning_rate": 3.86986986986987e-06,
      "loss": 0.1992,
      "step": 61240
    },
    {
      "epoch": 183.93,
      "grad_norm": 7.969050884246826,
      "learning_rate": 3.868868868868869e-06,
      "loss": 0.2484,
      "step": 61250
    },
    {
      "epoch": 183.96,
      "grad_norm": 14.100822448730469,
      "learning_rate": 3.8678678678678685e-06,
      "loss": 0.297,
      "step": 61260
    },
    {
      "epoch": 183.99,
      "grad_norm": 7.186295986175537,
      "learning_rate": 3.866866866866867e-06,
      "loss": 0.2251,
      "step": 61270
    },
    {
      "epoch": 184.0,
      "eval_accuracy": 0.922,
      "eval_loss": 0.34086284041404724,
      "eval_runtime": 29.9515,
      "eval_samples_per_second": 333.873,
      "eval_steps_per_second": 1.335,
      "step": 61272
    },
    {
      "epoch": 184.02,
      "grad_norm": 5.183709144592285,
      "learning_rate": 3.865865865865866e-06,
      "loss": 0.1693,
      "step": 61280
    },
    {
      "epoch": 184.05,
      "grad_norm": 8.353272438049316,
      "learning_rate": 3.864864864864865e-06,
      "loss": 0.2099,
      "step": 61290
    },
    {
      "epoch": 184.08,
      "grad_norm": 6.761512756347656,
      "learning_rate": 3.863863863863865e-06,
      "loss": 0.1574,
      "step": 61300
    },
    {
      "epoch": 184.11,
      "grad_norm": 7.3511457443237305,
      "learning_rate": 3.862862862862863e-06,
      "loss": 0.2117,
      "step": 61310
    },
    {
      "epoch": 184.14,
      "grad_norm": 21.552804946899414,
      "learning_rate": 3.861861861861862e-06,
      "loss": 0.2309,
      "step": 61320
    },
    {
      "epoch": 184.17,
      "grad_norm": 7.601670265197754,
      "learning_rate": 3.860860860860861e-06,
      "loss": 0.2298,
      "step": 61330
    },
    {
      "epoch": 184.2,
      "grad_norm": 8.122052192687988,
      "learning_rate": 3.85985985985986e-06,
      "loss": 0.1958,
      "step": 61340
    },
    {
      "epoch": 184.23,
      "grad_norm": 13.269152641296387,
      "learning_rate": 3.858858858858859e-06,
      "loss": 0.2139,
      "step": 61350
    },
    {
      "epoch": 184.26,
      "grad_norm": 8.260092735290527,
      "learning_rate": 3.857857857857858e-06,
      "loss": 0.1958,
      "step": 61360
    },
    {
      "epoch": 184.29,
      "grad_norm": 5.210981845855713,
      "learning_rate": 3.856856856856857e-06,
      "loss": 0.2125,
      "step": 61370
    },
    {
      "epoch": 184.32,
      "grad_norm": 21.452550888061523,
      "learning_rate": 3.855855855855856e-06,
      "loss": 0.2383,
      "step": 61380
    },
    {
      "epoch": 184.35,
      "grad_norm": 7.594136714935303,
      "learning_rate": 3.854854854854855e-06,
      "loss": 0.2203,
      "step": 61390
    },
    {
      "epoch": 184.38,
      "grad_norm": 9.006404876708984,
      "learning_rate": 3.853853853853854e-06,
      "loss": 0.2201,
      "step": 61400
    },
    {
      "epoch": 184.41,
      "grad_norm": 12.7974853515625,
      "learning_rate": 3.852852852852853e-06,
      "loss": 0.2805,
      "step": 61410
    },
    {
      "epoch": 184.44,
      "grad_norm": 15.652777671813965,
      "learning_rate": 3.851851851851852e-06,
      "loss": 0.21,
      "step": 61420
    },
    {
      "epoch": 184.47,
      "grad_norm": 5.554023265838623,
      "learning_rate": 3.8508508508508514e-06,
      "loss": 0.2084,
      "step": 61430
    },
    {
      "epoch": 184.5,
      "grad_norm": 10.802529335021973,
      "learning_rate": 3.84984984984985e-06,
      "loss": 0.2103,
      "step": 61440
    },
    {
      "epoch": 184.53,
      "grad_norm": 9.247808456420898,
      "learning_rate": 3.848848848848849e-06,
      "loss": 0.2489,
      "step": 61450
    },
    {
      "epoch": 184.56,
      "grad_norm": 10.940279006958008,
      "learning_rate": 3.847847847847848e-06,
      "loss": 0.2312,
      "step": 61460
    },
    {
      "epoch": 184.59,
      "grad_norm": 7.1483988761901855,
      "learning_rate": 3.8468468468468475e-06,
      "loss": 0.2164,
      "step": 61470
    },
    {
      "epoch": 184.62,
      "grad_norm": 5.086723327636719,
      "learning_rate": 3.845845845845846e-06,
      "loss": 0.1965,
      "step": 61480
    },
    {
      "epoch": 184.65,
      "grad_norm": 12.099916458129883,
      "learning_rate": 3.844844844844845e-06,
      "loss": 0.1823,
      "step": 61490
    },
    {
      "epoch": 184.68,
      "grad_norm": 5.10136079788208,
      "learning_rate": 3.843843843843844e-06,
      "loss": 0.1759,
      "step": 61500
    },
    {
      "epoch": 184.71,
      "grad_norm": 6.916656970977783,
      "learning_rate": 3.842842842842844e-06,
      "loss": 0.1926,
      "step": 61510
    },
    {
      "epoch": 184.74,
      "grad_norm": 11.073264122009277,
      "learning_rate": 3.841841841841842e-06,
      "loss": 0.2428,
      "step": 61520
    },
    {
      "epoch": 184.77,
      "grad_norm": 10.919217109680176,
      "learning_rate": 3.840840840840841e-06,
      "loss": 0.2087,
      "step": 61530
    },
    {
      "epoch": 184.8,
      "grad_norm": 7.475068092346191,
      "learning_rate": 3.83983983983984e-06,
      "loss": 0.1885,
      "step": 61540
    },
    {
      "epoch": 184.83,
      "grad_norm": 7.736336708068848,
      "learning_rate": 3.83883883883884e-06,
      "loss": 0.2048,
      "step": 61550
    },
    {
      "epoch": 184.86,
      "grad_norm": 53.232181549072266,
      "learning_rate": 3.837837837837838e-06,
      "loss": 0.2245,
      "step": 61560
    },
    {
      "epoch": 184.89,
      "grad_norm": 9.875178337097168,
      "learning_rate": 3.836836836836837e-06,
      "loss": 0.2152,
      "step": 61570
    },
    {
      "epoch": 184.92,
      "grad_norm": 9.994438171386719,
      "learning_rate": 3.835835835835836e-06,
      "loss": 0.2329,
      "step": 61580
    },
    {
      "epoch": 184.95,
      "grad_norm": 4.951380252838135,
      "learning_rate": 3.834834834834835e-06,
      "loss": 0.1924,
      "step": 61590
    },
    {
      "epoch": 184.98,
      "grad_norm": 8.396260261535645,
      "learning_rate": 3.833833833833834e-06,
      "loss": 0.2562,
      "step": 61600
    },
    {
      "epoch": 185.0,
      "eval_accuracy": 0.9239,
      "eval_loss": 0.3356321156024933,
      "eval_runtime": 30.0912,
      "eval_samples_per_second": 332.323,
      "eval_steps_per_second": 1.329,
      "step": 61605
    },
    {
      "epoch": 185.02,
      "grad_norm": 11.998372077941895,
      "learning_rate": 3.832832832832833e-06,
      "loss": 0.1886,
      "step": 61610
    },
    {
      "epoch": 185.05,
      "grad_norm": 5.986937046051025,
      "learning_rate": 3.831831831831832e-06,
      "loss": 0.2022,
      "step": 61620
    },
    {
      "epoch": 185.08,
      "grad_norm": 6.512185096740723,
      "learning_rate": 3.830830830830831e-06,
      "loss": 0.215,
      "step": 61630
    },
    {
      "epoch": 185.11,
      "grad_norm": 5.097291946411133,
      "learning_rate": 3.82982982982983e-06,
      "loss": 0.1977,
      "step": 61640
    },
    {
      "epoch": 185.14,
      "grad_norm": 9.12633228302002,
      "learning_rate": 3.828828828828829e-06,
      "loss": 0.2317,
      "step": 61650
    },
    {
      "epoch": 185.17,
      "grad_norm": 8.104333877563477,
      "learning_rate": 3.8278278278278284e-06,
      "loss": 0.1957,
      "step": 61660
    },
    {
      "epoch": 185.2,
      "grad_norm": 8.334063529968262,
      "learning_rate": 3.826826826826827e-06,
      "loss": 0.1948,
      "step": 61670
    },
    {
      "epoch": 185.23,
      "grad_norm": 10.69513988494873,
      "learning_rate": 3.8258258258258265e-06,
      "loss": 0.221,
      "step": 61680
    },
    {
      "epoch": 185.26,
      "grad_norm": 8.382991790771484,
      "learning_rate": 3.824824824824825e-06,
      "loss": 0.2281,
      "step": 61690
    },
    {
      "epoch": 185.29,
      "grad_norm": 12.366327285766602,
      "learning_rate": 3.823823823823824e-06,
      "loss": 0.2301,
      "step": 61700
    },
    {
      "epoch": 185.32,
      "grad_norm": 6.969593524932861,
      "learning_rate": 3.822822822822823e-06,
      "loss": 0.2166,
      "step": 61710
    },
    {
      "epoch": 185.35,
      "grad_norm": 21.135414123535156,
      "learning_rate": 3.8218218218218225e-06,
      "loss": 0.2403,
      "step": 61720
    },
    {
      "epoch": 185.38,
      "grad_norm": 11.940510749816895,
      "learning_rate": 3.820820820820821e-06,
      "loss": 0.2567,
      "step": 61730
    },
    {
      "epoch": 185.41,
      "grad_norm": 9.045462608337402,
      "learning_rate": 3.81981981981982e-06,
      "loss": 0.2144,
      "step": 61740
    },
    {
      "epoch": 185.44,
      "grad_norm": 8.785630226135254,
      "learning_rate": 3.818818818818819e-06,
      "loss": 0.2111,
      "step": 61750
    },
    {
      "epoch": 185.47,
      "grad_norm": 8.678489685058594,
      "learning_rate": 3.817817817817819e-06,
      "loss": 0.2353,
      "step": 61760
    },
    {
      "epoch": 185.5,
      "grad_norm": 7.3672776222229,
      "learning_rate": 3.816816816816817e-06,
      "loss": 0.233,
      "step": 61770
    },
    {
      "epoch": 185.53,
      "grad_norm": 12.436849594116211,
      "learning_rate": 3.815815815815816e-06,
      "loss": 0.2045,
      "step": 61780
    },
    {
      "epoch": 185.56,
      "grad_norm": 6.782108783721924,
      "learning_rate": 3.814814814814815e-06,
      "loss": 0.2262,
      "step": 61790
    },
    {
      "epoch": 185.59,
      "grad_norm": 13.741801261901855,
      "learning_rate": 3.8138138138138143e-06,
      "loss": 0.2078,
      "step": 61800
    },
    {
      "epoch": 185.62,
      "grad_norm": 13.535161972045898,
      "learning_rate": 3.8128128128128133e-06,
      "loss": 0.2123,
      "step": 61810
    },
    {
      "epoch": 185.65,
      "grad_norm": 12.732610702514648,
      "learning_rate": 3.811811811811812e-06,
      "loss": 0.22,
      "step": 61820
    },
    {
      "epoch": 185.68,
      "grad_norm": 26.220535278320312,
      "learning_rate": 3.810810810810811e-06,
      "loss": 0.1959,
      "step": 61830
    },
    {
      "epoch": 185.71,
      "grad_norm": 9.455082893371582,
      "learning_rate": 3.8098098098098103e-06,
      "loss": 0.2329,
      "step": 61840
    },
    {
      "epoch": 185.74,
      "grad_norm": 4.216155052185059,
      "learning_rate": 3.8088088088088094e-06,
      "loss": 0.1867,
      "step": 61850
    },
    {
      "epoch": 185.77,
      "grad_norm": 13.221193313598633,
      "learning_rate": 3.807807807807808e-06,
      "loss": 0.1949,
      "step": 61860
    },
    {
      "epoch": 185.8,
      "grad_norm": 6.680513381958008,
      "learning_rate": 3.806806806806807e-06,
      "loss": 0.2071,
      "step": 61870
    },
    {
      "epoch": 185.83,
      "grad_norm": 9.744094848632812,
      "learning_rate": 3.805805805805806e-06,
      "loss": 0.2107,
      "step": 61880
    },
    {
      "epoch": 185.86,
      "grad_norm": 8.364263534545898,
      "learning_rate": 3.8048048048048054e-06,
      "loss": 0.2074,
      "step": 61890
    },
    {
      "epoch": 185.89,
      "grad_norm": 5.897045135498047,
      "learning_rate": 3.803803803803804e-06,
      "loss": 0.2178,
      "step": 61900
    },
    {
      "epoch": 185.92,
      "grad_norm": 6.86648416519165,
      "learning_rate": 3.802802802802803e-06,
      "loss": 0.204,
      "step": 61910
    },
    {
      "epoch": 185.95,
      "grad_norm": 6.065003395080566,
      "learning_rate": 3.801801801801802e-06,
      "loss": 0.203,
      "step": 61920
    },
    {
      "epoch": 185.98,
      "grad_norm": 7.174644470214844,
      "learning_rate": 3.800800800800801e-06,
      "loss": 0.2225,
      "step": 61930
    },
    {
      "epoch": 186.0,
      "eval_accuracy": 0.9228,
      "eval_loss": 0.33896103501319885,
      "eval_runtime": 30.034,
      "eval_samples_per_second": 332.956,
      "eval_steps_per_second": 1.332,
      "step": 61938
    },
    {
      "epoch": 186.01,
      "grad_norm": 6.992156982421875,
      "learning_rate": 3.7997997997997997e-06,
      "loss": 0.2118,
      "step": 61940
    },
    {
      "epoch": 186.04,
      "grad_norm": 5.253401756286621,
      "learning_rate": 3.798798798798799e-06,
      "loss": 0.2129,
      "step": 61950
    },
    {
      "epoch": 186.07,
      "grad_norm": 9.844226837158203,
      "learning_rate": 3.797797797797798e-06,
      "loss": 0.2156,
      "step": 61960
    },
    {
      "epoch": 186.1,
      "grad_norm": 5.815432071685791,
      "learning_rate": 3.796796796796797e-06,
      "loss": 0.198,
      "step": 61970
    },
    {
      "epoch": 186.13,
      "grad_norm": 15.371514320373535,
      "learning_rate": 3.795795795795796e-06,
      "loss": 0.1972,
      "step": 61980
    },
    {
      "epoch": 186.16,
      "grad_norm": 18.65663719177246,
      "learning_rate": 3.7947947947947948e-06,
      "loss": 0.2076,
      "step": 61990
    },
    {
      "epoch": 186.19,
      "grad_norm": 12.439769744873047,
      "learning_rate": 3.793793793793794e-06,
      "loss": 0.2218,
      "step": 62000
    },
    {
      "epoch": 186.22,
      "grad_norm": 11.399972915649414,
      "learning_rate": 3.7927927927927932e-06,
      "loss": 0.2348,
      "step": 62010
    },
    {
      "epoch": 186.25,
      "grad_norm": 7.956319332122803,
      "learning_rate": 3.7917917917917922e-06,
      "loss": 0.2146,
      "step": 62020
    },
    {
      "epoch": 186.28,
      "grad_norm": 5.330765724182129,
      "learning_rate": 3.790790790790791e-06,
      "loss": 0.2304,
      "step": 62030
    },
    {
      "epoch": 186.31,
      "grad_norm": 8.88321304321289,
      "learning_rate": 3.78978978978979e-06,
      "loss": 0.2272,
      "step": 62040
    },
    {
      "epoch": 186.34,
      "grad_norm": 7.152944087982178,
      "learning_rate": 3.7887887887887893e-06,
      "loss": 0.229,
      "step": 62050
    },
    {
      "epoch": 186.37,
      "grad_norm": 8.217324256896973,
      "learning_rate": 3.7877877877877883e-06,
      "loss": 0.1832,
      "step": 62060
    },
    {
      "epoch": 186.4,
      "grad_norm": 9.154486656188965,
      "learning_rate": 3.786786786786787e-06,
      "loss": 0.209,
      "step": 62070
    },
    {
      "epoch": 186.43,
      "grad_norm": 5.546656131744385,
      "learning_rate": 3.785785785785786e-06,
      "loss": 0.2056,
      "step": 62080
    },
    {
      "epoch": 186.46,
      "grad_norm": 10.328149795532227,
      "learning_rate": 3.7847847847847854e-06,
      "loss": 0.2401,
      "step": 62090
    },
    {
      "epoch": 186.49,
      "grad_norm": 4.219549179077148,
      "learning_rate": 3.7837837837837844e-06,
      "loss": 0.2287,
      "step": 62100
    },
    {
      "epoch": 186.52,
      "grad_norm": 10.346301078796387,
      "learning_rate": 3.782782782782783e-06,
      "loss": 0.2391,
      "step": 62110
    },
    {
      "epoch": 186.55,
      "grad_norm": 7.331993103027344,
      "learning_rate": 3.781781781781782e-06,
      "loss": 0.1737,
      "step": 62120
    },
    {
      "epoch": 186.58,
      "grad_norm": 10.479643821716309,
      "learning_rate": 3.780780780780781e-06,
      "loss": 0.18,
      "step": 62130
    },
    {
      "epoch": 186.61,
      "grad_norm": 10.275055885314941,
      "learning_rate": 3.7797797797797805e-06,
      "loss": 0.2471,
      "step": 62140
    },
    {
      "epoch": 186.64,
      "grad_norm": 13.710062026977539,
      "learning_rate": 3.778778778778779e-06,
      "loss": 0.2363,
      "step": 62150
    },
    {
      "epoch": 186.67,
      "grad_norm": 8.776274681091309,
      "learning_rate": 3.777777777777778e-06,
      "loss": 0.202,
      "step": 62160
    },
    {
      "epoch": 186.7,
      "grad_norm": 6.370782852172852,
      "learning_rate": 3.776776776776777e-06,
      "loss": 0.1508,
      "step": 62170
    },
    {
      "epoch": 186.73,
      "grad_norm": 7.128628730773926,
      "learning_rate": 3.775775775775776e-06,
      "loss": 0.2067,
      "step": 62180
    },
    {
      "epoch": 186.76,
      "grad_norm": 12.391769409179688,
      "learning_rate": 3.7747747747747747e-06,
      "loss": 0.234,
      "step": 62190
    },
    {
      "epoch": 186.79,
      "grad_norm": 14.370572090148926,
      "learning_rate": 3.773773773773774e-06,
      "loss": 0.2169,
      "step": 62200
    },
    {
      "epoch": 186.82,
      "grad_norm": 23.70587730407715,
      "learning_rate": 3.772772772772773e-06,
      "loss": 0.2046,
      "step": 62210
    },
    {
      "epoch": 186.85,
      "grad_norm": 10.201399803161621,
      "learning_rate": 3.771771771771772e-06,
      "loss": 0.2273,
      "step": 62220
    },
    {
      "epoch": 186.88,
      "grad_norm": 7.190038681030273,
      "learning_rate": 3.7707707707707708e-06,
      "loss": 0.2128,
      "step": 62230
    },
    {
      "epoch": 186.91,
      "grad_norm": 35.50861740112305,
      "learning_rate": 3.76976976976977e-06,
      "loss": 0.2027,
      "step": 62240
    },
    {
      "epoch": 186.94,
      "grad_norm": 6.857219219207764,
      "learning_rate": 3.7687687687687692e-06,
      "loss": 0.2346,
      "step": 62250
    },
    {
      "epoch": 186.97,
      "grad_norm": 8.03045654296875,
      "learning_rate": 3.7677677677677683e-06,
      "loss": 0.2434,
      "step": 62260
    },
    {
      "epoch": 187.0,
      "grad_norm": 4.647548675537109,
      "learning_rate": 3.7667667667667673e-06,
      "loss": 0.1772,
      "step": 62270
    },
    {
      "epoch": 187.0,
      "eval_accuracy": 0.9219,
      "eval_loss": 0.3390255272388458,
      "eval_runtime": 30.2639,
      "eval_samples_per_second": 330.427,
      "eval_steps_per_second": 1.322,
      "step": 62271
    },
    {
      "epoch": 187.03,
      "grad_norm": 12.290539741516113,
      "learning_rate": 3.765765765765766e-06,
      "loss": 0.2149,
      "step": 62280
    },
    {
      "epoch": 187.06,
      "grad_norm": 8.37639045715332,
      "learning_rate": 3.764764764764765e-06,
      "loss": 0.1973,
      "step": 62290
    },
    {
      "epoch": 187.09,
      "grad_norm": 12.13231372833252,
      "learning_rate": 3.7637637637637643e-06,
      "loss": 0.2068,
      "step": 62300
    },
    {
      "epoch": 187.12,
      "grad_norm": 7.914624214172363,
      "learning_rate": 3.7627627627627634e-06,
      "loss": 0.2738,
      "step": 62310
    },
    {
      "epoch": 187.15,
      "grad_norm": 13.070964813232422,
      "learning_rate": 3.761761761761762e-06,
      "loss": 0.2005,
      "step": 62320
    },
    {
      "epoch": 187.18,
      "grad_norm": 11.182637214660645,
      "learning_rate": 3.760760760760761e-06,
      "loss": 0.2001,
      "step": 62330
    },
    {
      "epoch": 187.21,
      "grad_norm": 7.87782096862793,
      "learning_rate": 3.75975975975976e-06,
      "loss": 0.178,
      "step": 62340
    },
    {
      "epoch": 187.24,
      "grad_norm": 33.062435150146484,
      "learning_rate": 3.7587587587587594e-06,
      "loss": 0.2246,
      "step": 62350
    },
    {
      "epoch": 187.27,
      "grad_norm": 11.016501426696777,
      "learning_rate": 3.757757757757758e-06,
      "loss": 0.18,
      "step": 62360
    },
    {
      "epoch": 187.3,
      "grad_norm": 21.772947311401367,
      "learning_rate": 3.756756756756757e-06,
      "loss": 0.2151,
      "step": 62370
    },
    {
      "epoch": 187.33,
      "grad_norm": 7.32255220413208,
      "learning_rate": 3.755755755755756e-06,
      "loss": 0.2056,
      "step": 62380
    },
    {
      "epoch": 187.36,
      "grad_norm": 21.55680274963379,
      "learning_rate": 3.754754754754755e-06,
      "loss": 0.1755,
      "step": 62390
    },
    {
      "epoch": 187.39,
      "grad_norm": 6.988755226135254,
      "learning_rate": 3.7537537537537537e-06,
      "loss": 0.214,
      "step": 62400
    },
    {
      "epoch": 187.42,
      "grad_norm": 9.494267463684082,
      "learning_rate": 3.752752752752753e-06,
      "loss": 0.2191,
      "step": 62410
    },
    {
      "epoch": 187.45,
      "grad_norm": 6.782876491546631,
      "learning_rate": 3.751751751751752e-06,
      "loss": 0.1958,
      "step": 62420
    },
    {
      "epoch": 187.48,
      "grad_norm": 7.545032024383545,
      "learning_rate": 3.750750750750751e-06,
      "loss": 0.2007,
      "step": 62430
    },
    {
      "epoch": 187.51,
      "grad_norm": 12.051483154296875,
      "learning_rate": 3.7497497497497497e-06,
      "loss": 0.2022,
      "step": 62440
    },
    {
      "epoch": 187.54,
      "grad_norm": 9.032574653625488,
      "learning_rate": 3.7487487487487488e-06,
      "loss": 0.2071,
      "step": 62450
    },
    {
      "epoch": 187.57,
      "grad_norm": 6.183919906616211,
      "learning_rate": 3.747747747747748e-06,
      "loss": 0.2333,
      "step": 62460
    },
    {
      "epoch": 187.6,
      "grad_norm": 20.40022850036621,
      "learning_rate": 3.7467467467467472e-06,
      "loss": 0.2071,
      "step": 62470
    },
    {
      "epoch": 187.63,
      "grad_norm": 10.715044975280762,
      "learning_rate": 3.745745745745746e-06,
      "loss": 0.2503,
      "step": 62480
    },
    {
      "epoch": 187.66,
      "grad_norm": 17.40686798095703,
      "learning_rate": 3.744744744744745e-06,
      "loss": 0.212,
      "step": 62490
    },
    {
      "epoch": 187.69,
      "grad_norm": 9.159751892089844,
      "learning_rate": 3.743743743743744e-06,
      "loss": 0.1864,
      "step": 62500
    },
    {
      "epoch": 187.72,
      "grad_norm": 20.893836975097656,
      "learning_rate": 3.7427427427427433e-06,
      "loss": 0.2486,
      "step": 62510
    },
    {
      "epoch": 187.75,
      "grad_norm": 8.359509468078613,
      "learning_rate": 3.7417417417417423e-06,
      "loss": 0.2204,
      "step": 62520
    },
    {
      "epoch": 187.78,
      "grad_norm": 8.823046684265137,
      "learning_rate": 3.740740740740741e-06,
      "loss": 0.2062,
      "step": 62530
    },
    {
      "epoch": 187.81,
      "grad_norm": 7.956148624420166,
      "learning_rate": 3.73973973973974e-06,
      "loss": 0.1914,
      "step": 62540
    },
    {
      "epoch": 187.84,
      "grad_norm": 5.6153388023376465,
      "learning_rate": 3.7387387387387394e-06,
      "loss": 0.2356,
      "step": 62550
    },
    {
      "epoch": 187.87,
      "grad_norm": 7.970269203186035,
      "learning_rate": 3.7377377377377384e-06,
      "loss": 0.2268,
      "step": 62560
    },
    {
      "epoch": 187.9,
      "grad_norm": 9.045745849609375,
      "learning_rate": 3.736736736736737e-06,
      "loss": 0.2266,
      "step": 62570
    },
    {
      "epoch": 187.93,
      "grad_norm": 6.7748918533325195,
      "learning_rate": 3.735735735735736e-06,
      "loss": 0.2178,
      "step": 62580
    },
    {
      "epoch": 187.96,
      "grad_norm": 24.076160430908203,
      "learning_rate": 3.734734734734735e-06,
      "loss": 0.2263,
      "step": 62590
    },
    {
      "epoch": 187.99,
      "grad_norm": 8.884088516235352,
      "learning_rate": 3.7337337337337345e-06,
      "loss": 0.2343,
      "step": 62600
    },
    {
      "epoch": 188.0,
      "eval_accuracy": 0.9219,
      "eval_loss": 0.34187188744544983,
      "eval_runtime": 30.3816,
      "eval_samples_per_second": 329.146,
      "eval_steps_per_second": 1.317,
      "step": 62604
    },
    {
      "epoch": 188.02,
      "grad_norm": 7.67901611328125,
      "learning_rate": 3.732732732732733e-06,
      "loss": 0.2,
      "step": 62610
    },
    {
      "epoch": 188.05,
      "grad_norm": 14.602952003479004,
      "learning_rate": 3.731731731731732e-06,
      "loss": 0.213,
      "step": 62620
    },
    {
      "epoch": 188.08,
      "grad_norm": 10.667397499084473,
      "learning_rate": 3.730730730730731e-06,
      "loss": 0.2326,
      "step": 62630
    },
    {
      "epoch": 188.11,
      "grad_norm": 6.943629264831543,
      "learning_rate": 3.72972972972973e-06,
      "loss": 0.2291,
      "step": 62640
    },
    {
      "epoch": 188.14,
      "grad_norm": 10.898261070251465,
      "learning_rate": 3.7287287287287287e-06,
      "loss": 0.1901,
      "step": 62650
    },
    {
      "epoch": 188.17,
      "grad_norm": 15.912822723388672,
      "learning_rate": 3.727727727727728e-06,
      "loss": 0.2035,
      "step": 62660
    },
    {
      "epoch": 188.2,
      "grad_norm": 7.217033863067627,
      "learning_rate": 3.726726726726727e-06,
      "loss": 0.217,
      "step": 62670
    },
    {
      "epoch": 188.23,
      "grad_norm": 6.071277141571045,
      "learning_rate": 3.725725725725726e-06,
      "loss": 0.1947,
      "step": 62680
    },
    {
      "epoch": 188.26,
      "grad_norm": 15.210527420043945,
      "learning_rate": 3.7247247247247248e-06,
      "loss": 0.2189,
      "step": 62690
    },
    {
      "epoch": 188.29,
      "grad_norm": 7.534692287445068,
      "learning_rate": 3.723723723723724e-06,
      "loss": 0.1979,
      "step": 62700
    },
    {
      "epoch": 188.32,
      "grad_norm": 8.80034065246582,
      "learning_rate": 3.7227227227227232e-06,
      "loss": 0.1845,
      "step": 62710
    },
    {
      "epoch": 188.35,
      "grad_norm": 9.514727592468262,
      "learning_rate": 3.7217217217217223e-06,
      "loss": 0.226,
      "step": 62720
    },
    {
      "epoch": 188.38,
      "grad_norm": 26.643857955932617,
      "learning_rate": 3.720720720720721e-06,
      "loss": 0.2168,
      "step": 62730
    },
    {
      "epoch": 188.41,
      "grad_norm": 5.602067947387695,
      "learning_rate": 3.71971971971972e-06,
      "loss": 0.2572,
      "step": 62740
    },
    {
      "epoch": 188.44,
      "grad_norm": 7.500553131103516,
      "learning_rate": 3.718718718718719e-06,
      "loss": 0.2204,
      "step": 62750
    },
    {
      "epoch": 188.47,
      "grad_norm": 26.567663192749023,
      "learning_rate": 3.7177177177177183e-06,
      "loss": 0.2167,
      "step": 62760
    },
    {
      "epoch": 188.5,
      "grad_norm": 7.621906280517578,
      "learning_rate": 3.716716716716717e-06,
      "loss": 0.2233,
      "step": 62770
    },
    {
      "epoch": 188.53,
      "grad_norm": 11.329455375671387,
      "learning_rate": 3.715715715715716e-06,
      "loss": 0.1975,
      "step": 62780
    },
    {
      "epoch": 188.56,
      "grad_norm": 6.2605743408203125,
      "learning_rate": 3.714714714714715e-06,
      "loss": 0.2508,
      "step": 62790
    },
    {
      "epoch": 188.59,
      "grad_norm": 8.118038177490234,
      "learning_rate": 3.713713713713714e-06,
      "loss": 0.1758,
      "step": 62800
    },
    {
      "epoch": 188.62,
      "grad_norm": 111.63927459716797,
      "learning_rate": 3.7127127127127134e-06,
      "loss": 0.2122,
      "step": 62810
    },
    {
      "epoch": 188.65,
      "grad_norm": 10.88803482055664,
      "learning_rate": 3.711711711711712e-06,
      "loss": 0.2301,
      "step": 62820
    },
    {
      "epoch": 188.68,
      "grad_norm": 6.06074333190918,
      "learning_rate": 3.710710710710711e-06,
      "loss": 0.2422,
      "step": 62830
    },
    {
      "epoch": 188.71,
      "grad_norm": 11.767058372497559,
      "learning_rate": 3.70970970970971e-06,
      "loss": 0.2428,
      "step": 62840
    },
    {
      "epoch": 188.74,
      "grad_norm": 12.005047798156738,
      "learning_rate": 3.708708708708709e-06,
      "loss": 0.268,
      "step": 62850
    },
    {
      "epoch": 188.77,
      "grad_norm": 13.168096542358398,
      "learning_rate": 3.7077077077077077e-06,
      "loss": 0.2027,
      "step": 62860
    },
    {
      "epoch": 188.8,
      "grad_norm": 6.164278984069824,
      "learning_rate": 3.706706706706707e-06,
      "loss": 0.1908,
      "step": 62870
    },
    {
      "epoch": 188.83,
      "grad_norm": 9.500144958496094,
      "learning_rate": 3.705705705705706e-06,
      "loss": 0.2335,
      "step": 62880
    },
    {
      "epoch": 188.86,
      "grad_norm": 4.983665466308594,
      "learning_rate": 3.704704704704705e-06,
      "loss": 0.207,
      "step": 62890
    },
    {
      "epoch": 188.89,
      "grad_norm": 9.625537872314453,
      "learning_rate": 3.7037037037037037e-06,
      "loss": 0.1983,
      "step": 62900
    },
    {
      "epoch": 188.92,
      "grad_norm": 11.18360710144043,
      "learning_rate": 3.7027027027027028e-06,
      "loss": 0.2679,
      "step": 62910
    },
    {
      "epoch": 188.95,
      "grad_norm": 10.102242469787598,
      "learning_rate": 3.701701701701702e-06,
      "loss": 0.1927,
      "step": 62920
    },
    {
      "epoch": 188.98,
      "grad_norm": 10.160902976989746,
      "learning_rate": 3.7007007007007012e-06,
      "loss": 0.2086,
      "step": 62930
    },
    {
      "epoch": 189.0,
      "eval_accuracy": 0.9222,
      "eval_loss": 0.33996573090553284,
      "eval_runtime": 30.2499,
      "eval_samples_per_second": 330.579,
      "eval_steps_per_second": 1.322,
      "step": 62937
    },
    {
      "epoch": 189.01,
      "grad_norm": 7.922623157501221,
      "learning_rate": 3.6996996996997e-06,
      "loss": 0.2254,
      "step": 62940
    },
    {
      "epoch": 189.04,
      "grad_norm": 9.309272766113281,
      "learning_rate": 3.698698698698699e-06,
      "loss": 0.2155,
      "step": 62950
    },
    {
      "epoch": 189.07,
      "grad_norm": 8.404796600341797,
      "learning_rate": 3.697697697697698e-06,
      "loss": 0.1847,
      "step": 62960
    },
    {
      "epoch": 189.1,
      "grad_norm": 7.991379261016846,
      "learning_rate": 3.6966966966966973e-06,
      "loss": 0.2191,
      "step": 62970
    },
    {
      "epoch": 189.13,
      "grad_norm": 8.856217384338379,
      "learning_rate": 3.695695695695696e-06,
      "loss": 0.2075,
      "step": 62980
    },
    {
      "epoch": 189.16,
      "grad_norm": 7.831416606903076,
      "learning_rate": 3.694694694694695e-06,
      "loss": 0.2048,
      "step": 62990
    },
    {
      "epoch": 189.19,
      "grad_norm": 7.626868724822998,
      "learning_rate": 3.693693693693694e-06,
      "loss": 0.1933,
      "step": 63000
    },
    {
      "epoch": 189.22,
      "grad_norm": 9.023187637329102,
      "learning_rate": 3.692692692692693e-06,
      "loss": 0.2251,
      "step": 63010
    },
    {
      "epoch": 189.25,
      "grad_norm": 8.904298782348633,
      "learning_rate": 3.6916916916916915e-06,
      "loss": 0.2262,
      "step": 63020
    },
    {
      "epoch": 189.28,
      "grad_norm": 11.566461563110352,
      "learning_rate": 3.690690690690691e-06,
      "loss": 0.2275,
      "step": 63030
    },
    {
      "epoch": 189.31,
      "grad_norm": 5.427104473114014,
      "learning_rate": 3.68968968968969e-06,
      "loss": 0.1504,
      "step": 63040
    },
    {
      "epoch": 189.34,
      "grad_norm": 10.506272315979004,
      "learning_rate": 3.688688688688689e-06,
      "loss": 0.2445,
      "step": 63050
    },
    {
      "epoch": 189.37,
      "grad_norm": 17.408329010009766,
      "learning_rate": 3.6876876876876885e-06,
      "loss": 0.2349,
      "step": 63060
    },
    {
      "epoch": 189.4,
      "grad_norm": 43.42242431640625,
      "learning_rate": 3.6866866866866866e-06,
      "loss": 0.1673,
      "step": 63070
    },
    {
      "epoch": 189.43,
      "grad_norm": 6.325368404388428,
      "learning_rate": 3.685685685685686e-06,
      "loss": 0.2572,
      "step": 63080
    },
    {
      "epoch": 189.46,
      "grad_norm": 6.338435173034668,
      "learning_rate": 3.684684684684685e-06,
      "loss": 0.22,
      "step": 63090
    },
    {
      "epoch": 189.49,
      "grad_norm": 13.216156959533691,
      "learning_rate": 3.683683683683684e-06,
      "loss": 0.2059,
      "step": 63100
    },
    {
      "epoch": 189.52,
      "grad_norm": 10.376838684082031,
      "learning_rate": 3.6826826826826827e-06,
      "loss": 0.1909,
      "step": 63110
    },
    {
      "epoch": 189.55,
      "grad_norm": 6.601317405700684,
      "learning_rate": 3.681681681681682e-06,
      "loss": 0.2067,
      "step": 63120
    },
    {
      "epoch": 189.58,
      "grad_norm": 7.264206886291504,
      "learning_rate": 3.680680680680681e-06,
      "loss": 0.2024,
      "step": 63130
    },
    {
      "epoch": 189.61,
      "grad_norm": 7.738499164581299,
      "learning_rate": 3.67967967967968e-06,
      "loss": 0.2099,
      "step": 63140
    },
    {
      "epoch": 189.64,
      "grad_norm": 9.30086898803711,
      "learning_rate": 3.6786786786786788e-06,
      "loss": 0.2113,
      "step": 63150
    },
    {
      "epoch": 189.67,
      "grad_norm": 7.840721130371094,
      "learning_rate": 3.677677677677678e-06,
      "loss": 0.2106,
      "step": 63160
    },
    {
      "epoch": 189.7,
      "grad_norm": 8.515464782714844,
      "learning_rate": 3.6766766766766772e-06,
      "loss": 0.1626,
      "step": 63170
    },
    {
      "epoch": 189.73,
      "grad_norm": 6.250887870788574,
      "learning_rate": 3.6756756756756763e-06,
      "loss": 0.2442,
      "step": 63180
    },
    {
      "epoch": 189.76,
      "grad_norm": 11.522494316101074,
      "learning_rate": 3.674674674674675e-06,
      "loss": 0.1761,
      "step": 63190
    },
    {
      "epoch": 189.79,
      "grad_norm": 8.378762245178223,
      "learning_rate": 3.673673673673674e-06,
      "loss": 0.1858,
      "step": 63200
    },
    {
      "epoch": 189.82,
      "grad_norm": 9.08778190612793,
      "learning_rate": 3.672672672672673e-06,
      "loss": 0.2312,
      "step": 63210
    },
    {
      "epoch": 189.85,
      "grad_norm": 7.369046688079834,
      "learning_rate": 3.6716716716716723e-06,
      "loss": 0.1977,
      "step": 63220
    },
    {
      "epoch": 189.88,
      "grad_norm": 11.291584014892578,
      "learning_rate": 3.670670670670671e-06,
      "loss": 0.2689,
      "step": 63230
    },
    {
      "epoch": 189.91,
      "grad_norm": 5.783355712890625,
      "learning_rate": 3.66966966966967e-06,
      "loss": 0.2068,
      "step": 63240
    },
    {
      "epoch": 189.94,
      "grad_norm": 9.560302734375,
      "learning_rate": 3.668668668668669e-06,
      "loss": 0.2258,
      "step": 63250
    },
    {
      "epoch": 189.97,
      "grad_norm": 7.925276279449463,
      "learning_rate": 3.667667667667668e-06,
      "loss": 0.2006,
      "step": 63260
    },
    {
      "epoch": 190.0,
      "grad_norm": 189.5849151611328,
      "learning_rate": 3.6666666666666666e-06,
      "loss": 0.3153,
      "step": 63270
    },
    {
      "epoch": 190.0,
      "eval_accuracy": 0.9203,
      "eval_loss": 0.343563437461853,
      "eval_runtime": 30.1726,
      "eval_samples_per_second": 331.426,
      "eval_steps_per_second": 1.326,
      "step": 63270
    },
    {
      "epoch": 190.03,
      "grad_norm": 9.544801712036133,
      "learning_rate": 3.665665665665666e-06,
      "loss": 0.1856,
      "step": 63280
    },
    {
      "epoch": 190.06,
      "grad_norm": 7.330122947692871,
      "learning_rate": 3.664664664664665e-06,
      "loss": 0.2183,
      "step": 63290
    },
    {
      "epoch": 190.09,
      "grad_norm": 6.477001667022705,
      "learning_rate": 3.663663663663664e-06,
      "loss": 0.2127,
      "step": 63300
    },
    {
      "epoch": 190.12,
      "grad_norm": 12.754526138305664,
      "learning_rate": 3.662662662662663e-06,
      "loss": 0.2084,
      "step": 63310
    },
    {
      "epoch": 190.15,
      "grad_norm": 8.851436614990234,
      "learning_rate": 3.6616616616616617e-06,
      "loss": 0.2014,
      "step": 63320
    },
    {
      "epoch": 190.18,
      "grad_norm": 6.957366466522217,
      "learning_rate": 3.660660660660661e-06,
      "loss": 0.2083,
      "step": 63330
    },
    {
      "epoch": 190.21,
      "grad_norm": 9.48468017578125,
      "learning_rate": 3.65965965965966e-06,
      "loss": 0.2373,
      "step": 63340
    },
    {
      "epoch": 190.24,
      "grad_norm": 9.972972869873047,
      "learning_rate": 3.658658658658659e-06,
      "loss": 0.2608,
      "step": 63350
    },
    {
      "epoch": 190.27,
      "grad_norm": 12.705851554870605,
      "learning_rate": 3.6576576576576577e-06,
      "loss": 0.2292,
      "step": 63360
    },
    {
      "epoch": 190.3,
      "grad_norm": 6.854614734649658,
      "learning_rate": 3.6566566566566568e-06,
      "loss": 0.2446,
      "step": 63370
    },
    {
      "epoch": 190.33,
      "grad_norm": 4.6282782554626465,
      "learning_rate": 3.655655655655656e-06,
      "loss": 0.1999,
      "step": 63380
    },
    {
      "epoch": 190.36,
      "grad_norm": 5.969880104064941,
      "learning_rate": 3.6546546546546552e-06,
      "loss": 0.1787,
      "step": 63390
    },
    {
      "epoch": 190.39,
      "grad_norm": 40.31931686401367,
      "learning_rate": 3.653653653653654e-06,
      "loss": 0.2626,
      "step": 63400
    },
    {
      "epoch": 190.42,
      "grad_norm": 11.468286514282227,
      "learning_rate": 3.652652652652653e-06,
      "loss": 0.2376,
      "step": 63410
    },
    {
      "epoch": 190.45,
      "grad_norm": 9.764470100402832,
      "learning_rate": 3.651651651651652e-06,
      "loss": 0.1808,
      "step": 63420
    },
    {
      "epoch": 190.48,
      "grad_norm": 15.194150924682617,
      "learning_rate": 3.6506506506506513e-06,
      "loss": 0.2063,
      "step": 63430
    },
    {
      "epoch": 190.51,
      "grad_norm": 6.187920093536377,
      "learning_rate": 3.64964964964965e-06,
      "loss": 0.1658,
      "step": 63440
    },
    {
      "epoch": 190.54,
      "grad_norm": 8.10157299041748,
      "learning_rate": 3.648648648648649e-06,
      "loss": 0.1643,
      "step": 63450
    },
    {
      "epoch": 190.57,
      "grad_norm": 11.623575210571289,
      "learning_rate": 3.647647647647648e-06,
      "loss": 0.1904,
      "step": 63460
    },
    {
      "epoch": 190.6,
      "grad_norm": 7.671225070953369,
      "learning_rate": 3.646646646646647e-06,
      "loss": 0.2075,
      "step": 63470
    },
    {
      "epoch": 190.63,
      "grad_norm": 9.210921287536621,
      "learning_rate": 3.6456456456456455e-06,
      "loss": 0.2432,
      "step": 63480
    },
    {
      "epoch": 190.66,
      "grad_norm": 8.379435539245605,
      "learning_rate": 3.644644644644645e-06,
      "loss": 0.2132,
      "step": 63490
    },
    {
      "epoch": 190.69,
      "grad_norm": 14.312080383300781,
      "learning_rate": 3.643643643643644e-06,
      "loss": 0.1919,
      "step": 63500
    },
    {
      "epoch": 190.72,
      "grad_norm": 6.039542198181152,
      "learning_rate": 3.642642642642643e-06,
      "loss": 0.1967,
      "step": 63510
    },
    {
      "epoch": 190.75,
      "grad_norm": 7.803469181060791,
      "learning_rate": 3.6416416416416416e-06,
      "loss": 0.2083,
      "step": 63520
    },
    {
      "epoch": 190.78,
      "grad_norm": 7.093661785125732,
      "learning_rate": 3.6406406406406406e-06,
      "loss": 0.2212,
      "step": 63530
    },
    {
      "epoch": 190.81,
      "grad_norm": 11.81009292602539,
      "learning_rate": 3.63963963963964e-06,
      "loss": 0.1902,
      "step": 63540
    },
    {
      "epoch": 190.84,
      "grad_norm": 10.518001556396484,
      "learning_rate": 3.638638638638639e-06,
      "loss": 0.227,
      "step": 63550
    },
    {
      "epoch": 190.87,
      "grad_norm": 9.337998390197754,
      "learning_rate": 3.6376376376376377e-06,
      "loss": 0.2363,
      "step": 63560
    },
    {
      "epoch": 190.9,
      "grad_norm": 11.252382278442383,
      "learning_rate": 3.6366366366366367e-06,
      "loss": 0.1871,
      "step": 63570
    },
    {
      "epoch": 190.93,
      "grad_norm": 5.511325836181641,
      "learning_rate": 3.635635635635636e-06,
      "loss": 0.1673,
      "step": 63580
    },
    {
      "epoch": 190.96,
      "grad_norm": 12.677375793457031,
      "learning_rate": 3.634634634634635e-06,
      "loss": 0.2429,
      "step": 63590
    },
    {
      "epoch": 190.99,
      "grad_norm": 7.890142440795898,
      "learning_rate": 3.633633633633634e-06,
      "loss": 0.2632,
      "step": 63600
    },
    {
      "epoch": 191.0,
      "eval_accuracy": 0.9226,
      "eval_loss": 0.34358957409858704,
      "eval_runtime": 30.3964,
      "eval_samples_per_second": 328.986,
      "eval_steps_per_second": 1.316,
      "step": 63603
    },
    {
      "epoch": 191.02,
      "grad_norm": 8.232089042663574,
      "learning_rate": 3.6326326326326328e-06,
      "loss": 0.2949,
      "step": 63610
    },
    {
      "epoch": 191.05,
      "grad_norm": 12.434263229370117,
      "learning_rate": 3.631631631631632e-06,
      "loss": 0.1966,
      "step": 63620
    },
    {
      "epoch": 191.08,
      "grad_norm": 7.860274791717529,
      "learning_rate": 3.6306306306306312e-06,
      "loss": 0.2214,
      "step": 63630
    },
    {
      "epoch": 191.11,
      "grad_norm": 6.346672534942627,
      "learning_rate": 3.6296296296296302e-06,
      "loss": 0.18,
      "step": 63640
    },
    {
      "epoch": 191.14,
      "grad_norm": 12.36599349975586,
      "learning_rate": 3.628628628628629e-06,
      "loss": 0.2175,
      "step": 63650
    },
    {
      "epoch": 191.17,
      "grad_norm": 13.884400367736816,
      "learning_rate": 3.627627627627628e-06,
      "loss": 0.1381,
      "step": 63660
    },
    {
      "epoch": 191.2,
      "grad_norm": 9.507542610168457,
      "learning_rate": 3.626626626626627e-06,
      "loss": 0.2188,
      "step": 63670
    },
    {
      "epoch": 191.23,
      "grad_norm": 12.99462604522705,
      "learning_rate": 3.6256256256256263e-06,
      "loss": 0.2393,
      "step": 63680
    },
    {
      "epoch": 191.26,
      "grad_norm": 10.403035163879395,
      "learning_rate": 3.624624624624625e-06,
      "loss": 0.2158,
      "step": 63690
    },
    {
      "epoch": 191.29,
      "grad_norm": 11.922537803649902,
      "learning_rate": 3.623623623623624e-06,
      "loss": 0.1999,
      "step": 63700
    },
    {
      "epoch": 191.32,
      "grad_norm": 8.211942672729492,
      "learning_rate": 3.622622622622623e-06,
      "loss": 0.2123,
      "step": 63710
    },
    {
      "epoch": 191.35,
      "grad_norm": 7.9579033851623535,
      "learning_rate": 3.621621621621622e-06,
      "loss": 0.2134,
      "step": 63720
    },
    {
      "epoch": 191.38,
      "grad_norm": 3.639238119125366,
      "learning_rate": 3.6206206206206206e-06,
      "loss": 0.2394,
      "step": 63730
    },
    {
      "epoch": 191.41,
      "grad_norm": 7.4707417488098145,
      "learning_rate": 3.61961961961962e-06,
      "loss": 0.2164,
      "step": 63740
    },
    {
      "epoch": 191.44,
      "grad_norm": 9.173367500305176,
      "learning_rate": 3.618618618618619e-06,
      "loss": 0.1947,
      "step": 63750
    },
    {
      "epoch": 191.47,
      "grad_norm": 6.641833305358887,
      "learning_rate": 3.617617617617618e-06,
      "loss": 0.2112,
      "step": 63760
    },
    {
      "epoch": 191.5,
      "grad_norm": 5.151862621307373,
      "learning_rate": 3.6166166166166166e-06,
      "loss": 0.2307,
      "step": 63770
    },
    {
      "epoch": 191.53,
      "grad_norm": 7.61786413192749,
      "learning_rate": 3.6156156156156157e-06,
      "loss": 0.2005,
      "step": 63780
    },
    {
      "epoch": 191.56,
      "grad_norm": 11.13754940032959,
      "learning_rate": 3.614614614614615e-06,
      "loss": 0.2397,
      "step": 63790
    },
    {
      "epoch": 191.59,
      "grad_norm": 11.164887428283691,
      "learning_rate": 3.613613613613614e-06,
      "loss": 0.2575,
      "step": 63800
    },
    {
      "epoch": 191.62,
      "grad_norm": 21.30828857421875,
      "learning_rate": 3.6126126126126127e-06,
      "loss": 0.2209,
      "step": 63810
    },
    {
      "epoch": 191.65,
      "grad_norm": 6.582218170166016,
      "learning_rate": 3.6116116116116117e-06,
      "loss": 0.2038,
      "step": 63820
    },
    {
      "epoch": 191.68,
      "grad_norm": 6.22662353515625,
      "learning_rate": 3.6106106106106108e-06,
      "loss": 0.1917,
      "step": 63830
    },
    {
      "epoch": 191.71,
      "grad_norm": 8.839052200317383,
      "learning_rate": 3.60960960960961e-06,
      "loss": 0.2107,
      "step": 63840
    },
    {
      "epoch": 191.74,
      "grad_norm": 9.213109016418457,
      "learning_rate": 3.608608608608609e-06,
      "loss": 0.2148,
      "step": 63850
    },
    {
      "epoch": 191.77,
      "grad_norm": 12.662994384765625,
      "learning_rate": 3.607607607607608e-06,
      "loss": 0.204,
      "step": 63860
    },
    {
      "epoch": 191.8,
      "grad_norm": 7.013178825378418,
      "learning_rate": 3.606606606606607e-06,
      "loss": 0.1513,
      "step": 63870
    },
    {
      "epoch": 191.83,
      "grad_norm": 8.045754432678223,
      "learning_rate": 3.605605605605606e-06,
      "loss": 0.2239,
      "step": 63880
    },
    {
      "epoch": 191.86,
      "grad_norm": 7.391717433929443,
      "learning_rate": 3.6046046046046053e-06,
      "loss": 0.2033,
      "step": 63890
    },
    {
      "epoch": 191.89,
      "grad_norm": 9.80645751953125,
      "learning_rate": 3.603603603603604e-06,
      "loss": 0.1852,
      "step": 63900
    },
    {
      "epoch": 191.92,
      "grad_norm": 5.854821681976318,
      "learning_rate": 3.602602602602603e-06,
      "loss": 0.2038,
      "step": 63910
    },
    {
      "epoch": 191.95,
      "grad_norm": 6.540364742279053,
      "learning_rate": 3.601601601601602e-06,
      "loss": 0.2298,
      "step": 63920
    },
    {
      "epoch": 191.98,
      "grad_norm": 6.970776557922363,
      "learning_rate": 3.600600600600601e-06,
      "loss": 0.2191,
      "step": 63930
    },
    {
      "epoch": 192.0,
      "eval_accuracy": 0.9218,
      "eval_loss": 0.3462984263896942,
      "eval_runtime": 30.294,
      "eval_samples_per_second": 330.099,
      "eval_steps_per_second": 1.32,
      "step": 63936
    },
    {
      "epoch": 192.01,
      "grad_norm": 16.20624351501465,
      "learning_rate": 3.5995995995995995e-06,
      "loss": 0.2173,
      "step": 63940
    },
    {
      "epoch": 192.04,
      "grad_norm": 7.9748640060424805,
      "learning_rate": 3.598598598598599e-06,
      "loss": 0.2345,
      "step": 63950
    },
    {
      "epoch": 192.07,
      "grad_norm": 9.669445037841797,
      "learning_rate": 3.597597597597598e-06,
      "loss": 0.2199,
      "step": 63960
    },
    {
      "epoch": 192.1,
      "grad_norm": 7.621309280395508,
      "learning_rate": 3.596596596596597e-06,
      "loss": 0.2292,
      "step": 63970
    },
    {
      "epoch": 192.13,
      "grad_norm": 12.471324920654297,
      "learning_rate": 3.5955955955955956e-06,
      "loss": 0.2261,
      "step": 63980
    },
    {
      "epoch": 192.16,
      "grad_norm": 13.742660522460938,
      "learning_rate": 3.5945945945945946e-06,
      "loss": 0.234,
      "step": 63990
    },
    {
      "epoch": 192.19,
      "grad_norm": 6.191371440887451,
      "learning_rate": 3.593593593593594e-06,
      "loss": 0.1617,
      "step": 64000
    },
    {
      "epoch": 192.22,
      "grad_norm": 9.751594543457031,
      "learning_rate": 3.592592592592593e-06,
      "loss": 0.1863,
      "step": 64010
    },
    {
      "epoch": 192.25,
      "grad_norm": 4.067356109619141,
      "learning_rate": 3.5915915915915917e-06,
      "loss": 0.1743,
      "step": 64020
    },
    {
      "epoch": 192.28,
      "grad_norm": 15.540249824523926,
      "learning_rate": 3.5905905905905907e-06,
      "loss": 0.2086,
      "step": 64030
    },
    {
      "epoch": 192.31,
      "grad_norm": 11.648731231689453,
      "learning_rate": 3.5895895895895897e-06,
      "loss": 0.2004,
      "step": 64040
    },
    {
      "epoch": 192.34,
      "grad_norm": 9.706010818481445,
      "learning_rate": 3.588588588588589e-06,
      "loss": 0.2349,
      "step": 64050
    },
    {
      "epoch": 192.37,
      "grad_norm": 6.475012302398682,
      "learning_rate": 3.5875875875875877e-06,
      "loss": 0.2041,
      "step": 64060
    },
    {
      "epoch": 192.4,
      "grad_norm": 7.752506732940674,
      "learning_rate": 3.5865865865865868e-06,
      "loss": 0.2071,
      "step": 64070
    },
    {
      "epoch": 192.43,
      "grad_norm": 7.306506633758545,
      "learning_rate": 3.5855855855855858e-06,
      "loss": 0.2266,
      "step": 64080
    },
    {
      "epoch": 192.46,
      "grad_norm": 8.533101081848145,
      "learning_rate": 3.5845845845845852e-06,
      "loss": 0.206,
      "step": 64090
    },
    {
      "epoch": 192.49,
      "grad_norm": 8.922447204589844,
      "learning_rate": 3.5835835835835834e-06,
      "loss": 0.2196,
      "step": 64100
    },
    {
      "epoch": 192.52,
      "grad_norm": 8.566412925720215,
      "learning_rate": 3.582582582582583e-06,
      "loss": 0.2103,
      "step": 64110
    },
    {
      "epoch": 192.55,
      "grad_norm": 7.984829425811768,
      "learning_rate": 3.581581581581582e-06,
      "loss": 0.1823,
      "step": 64120
    },
    {
      "epoch": 192.58,
      "grad_norm": 9.51751708984375,
      "learning_rate": 3.580580580580581e-06,
      "loss": 0.2204,
      "step": 64130
    },
    {
      "epoch": 192.61,
      "grad_norm": 6.554741859436035,
      "learning_rate": 3.5795795795795803e-06,
      "loss": 0.1868,
      "step": 64140
    },
    {
      "epoch": 192.64,
      "grad_norm": 17.278854370117188,
      "learning_rate": 3.578578578578579e-06,
      "loss": 0.1928,
      "step": 64150
    },
    {
      "epoch": 192.67,
      "grad_norm": 8.369423866271973,
      "learning_rate": 3.577577577577578e-06,
      "loss": 0.2611,
      "step": 64160
    },
    {
      "epoch": 192.7,
      "grad_norm": 10.969748497009277,
      "learning_rate": 3.576576576576577e-06,
      "loss": 0.1859,
      "step": 64170
    },
    {
      "epoch": 192.73,
      "grad_norm": 11.517537117004395,
      "learning_rate": 3.575575575575576e-06,
      "loss": 0.2225,
      "step": 64180
    },
    {
      "epoch": 192.76,
      "grad_norm": 6.211329460144043,
      "learning_rate": 3.5745745745745746e-06,
      "loss": 0.2017,
      "step": 64190
    },
    {
      "epoch": 192.79,
      "grad_norm": 15.454781532287598,
      "learning_rate": 3.573573573573574e-06,
      "loss": 0.2359,
      "step": 64200
    },
    {
      "epoch": 192.82,
      "grad_norm": 11.842299461364746,
      "learning_rate": 3.572572572572573e-06,
      "loss": 0.217,
      "step": 64210
    },
    {
      "epoch": 192.85,
      "grad_norm": 10.153335571289062,
      "learning_rate": 3.571571571571572e-06,
      "loss": 0.228,
      "step": 64220
    },
    {
      "epoch": 192.88,
      "grad_norm": 7.8100152015686035,
      "learning_rate": 3.5705705705705706e-06,
      "loss": 0.2139,
      "step": 64230
    },
    {
      "epoch": 192.91,
      "grad_norm": 13.460724830627441,
      "learning_rate": 3.5695695695695697e-06,
      "loss": 0.2071,
      "step": 64240
    },
    {
      "epoch": 192.94,
      "grad_norm": 18.04127311706543,
      "learning_rate": 3.568568568568569e-06,
      "loss": 0.2752,
      "step": 64250
    },
    {
      "epoch": 192.97,
      "grad_norm": 10.105291366577148,
      "learning_rate": 3.567567567567568e-06,
      "loss": 0.1892,
      "step": 64260
    },
    {
      "epoch": 193.0,
      "eval_accuracy": 0.9226,
      "eval_loss": 0.34552663564682007,
      "eval_runtime": 30.1084,
      "eval_samples_per_second": 332.133,
      "eval_steps_per_second": 1.329,
      "step": 64269
    },
    {
      "epoch": 193.0,
      "grad_norm": 7.507857322692871,
      "learning_rate": 3.5665665665665667e-06,
      "loss": 0.1907,
      "step": 64270
    },
    {
      "epoch": 193.03,
      "grad_norm": 11.050125122070312,
      "learning_rate": 3.5655655655655657e-06,
      "loss": 0.2463,
      "step": 64280
    },
    {
      "epoch": 193.06,
      "grad_norm": 5.831038475036621,
      "learning_rate": 3.5645645645645647e-06,
      "loss": 0.1762,
      "step": 64290
    },
    {
      "epoch": 193.09,
      "grad_norm": 8.647604942321777,
      "learning_rate": 3.563563563563564e-06,
      "loss": 0.2238,
      "step": 64300
    },
    {
      "epoch": 193.12,
      "grad_norm": 8.930716514587402,
      "learning_rate": 3.5625625625625628e-06,
      "loss": 0.2447,
      "step": 64310
    },
    {
      "epoch": 193.15,
      "grad_norm": 5.292147159576416,
      "learning_rate": 3.561561561561562e-06,
      "loss": 0.1952,
      "step": 64320
    },
    {
      "epoch": 193.18,
      "grad_norm": 8.670984268188477,
      "learning_rate": 3.560560560560561e-06,
      "loss": 0.1627,
      "step": 64330
    },
    {
      "epoch": 193.21,
      "grad_norm": 8.1242094039917,
      "learning_rate": 3.55955955955956e-06,
      "loss": 0.1866,
      "step": 64340
    },
    {
      "epoch": 193.24,
      "grad_norm": 7.98291015625,
      "learning_rate": 3.5585585585585584e-06,
      "loss": 0.2279,
      "step": 64350
    },
    {
      "epoch": 193.27,
      "grad_norm": 7.788426876068115,
      "learning_rate": 3.557557557557558e-06,
      "loss": 0.2163,
      "step": 64360
    },
    {
      "epoch": 193.3,
      "grad_norm": 5.309067249298096,
      "learning_rate": 3.556556556556557e-06,
      "loss": 0.2004,
      "step": 64370
    },
    {
      "epoch": 193.33,
      "grad_norm": 13.060711860656738,
      "learning_rate": 3.555555555555556e-06,
      "loss": 0.2122,
      "step": 64380
    },
    {
      "epoch": 193.36,
      "grad_norm": 10.357504844665527,
      "learning_rate": 3.554554554554555e-06,
      "loss": 0.1913,
      "step": 64390
    },
    {
      "epoch": 193.39,
      "grad_norm": 8.825194358825684,
      "learning_rate": 3.5535535535535535e-06,
      "loss": 0.1977,
      "step": 64400
    },
    {
      "epoch": 193.42,
      "grad_norm": 10.195891380310059,
      "learning_rate": 3.552552552552553e-06,
      "loss": 0.2232,
      "step": 64410
    },
    {
      "epoch": 193.45,
      "grad_norm": 5.876519680023193,
      "learning_rate": 3.551551551551552e-06,
      "loss": 0.2322,
      "step": 64420
    },
    {
      "epoch": 193.48,
      "grad_norm": 8.331722259521484,
      "learning_rate": 3.550550550550551e-06,
      "loss": 0.2061,
      "step": 64430
    },
    {
      "epoch": 193.51,
      "grad_norm": 6.499451160430908,
      "learning_rate": 3.5495495495495496e-06,
      "loss": 0.2034,
      "step": 64440
    },
    {
      "epoch": 193.54,
      "grad_norm": 21.573863983154297,
      "learning_rate": 3.5485485485485486e-06,
      "loss": 0.1775,
      "step": 64450
    },
    {
      "epoch": 193.57,
      "grad_norm": 7.849809169769287,
      "learning_rate": 3.547547547547548e-06,
      "loss": 0.2157,
      "step": 64460
    },
    {
      "epoch": 193.6,
      "grad_norm": 11.287116050720215,
      "learning_rate": 3.546546546546547e-06,
      "loss": 0.1922,
      "step": 64470
    },
    {
      "epoch": 193.63,
      "grad_norm": 10.136216163635254,
      "learning_rate": 3.5455455455455457e-06,
      "loss": 0.2275,
      "step": 64480
    },
    {
      "epoch": 193.66,
      "grad_norm": 13.088991165161133,
      "learning_rate": 3.5445445445445447e-06,
      "loss": 0.2334,
      "step": 64490
    },
    {
      "epoch": 193.69,
      "grad_norm": 8.46648120880127,
      "learning_rate": 3.5435435435435437e-06,
      "loss": 0.2218,
      "step": 64500
    },
    {
      "epoch": 193.72,
      "grad_norm": 6.965726375579834,
      "learning_rate": 3.542542542542543e-06,
      "loss": 0.2072,
      "step": 64510
    },
    {
      "epoch": 193.75,
      "grad_norm": 5.463528633117676,
      "learning_rate": 3.5415415415415417e-06,
      "loss": 0.202,
      "step": 64520
    },
    {
      "epoch": 193.78,
      "grad_norm": 8.852060317993164,
      "learning_rate": 3.5405405405405408e-06,
      "loss": 0.1897,
      "step": 64530
    },
    {
      "epoch": 193.81,
      "grad_norm": 17.842853546142578,
      "learning_rate": 3.5395395395395398e-06,
      "loss": 0.2065,
      "step": 64540
    },
    {
      "epoch": 193.84,
      "grad_norm": 8.491738319396973,
      "learning_rate": 3.5385385385385392e-06,
      "loss": 0.199,
      "step": 64550
    },
    {
      "epoch": 193.87,
      "grad_norm": 9.790806770324707,
      "learning_rate": 3.5375375375375374e-06,
      "loss": 0.2323,
      "step": 64560
    },
    {
      "epoch": 193.9,
      "grad_norm": 11.341293334960938,
      "learning_rate": 3.536536536536537e-06,
      "loss": 0.2141,
      "step": 64570
    },
    {
      "epoch": 193.93,
      "grad_norm": 8.132089614868164,
      "learning_rate": 3.535535535535536e-06,
      "loss": 0.2203,
      "step": 64580
    },
    {
      "epoch": 193.96,
      "grad_norm": 7.173344135284424,
      "learning_rate": 3.534534534534535e-06,
      "loss": 0.2244,
      "step": 64590
    },
    {
      "epoch": 193.99,
      "grad_norm": 12.309045791625977,
      "learning_rate": 3.5335335335335335e-06,
      "loss": 0.2246,
      "step": 64600
    },
    {
      "epoch": 194.0,
      "eval_accuracy": 0.9215,
      "eval_loss": 0.34541839361190796,
      "eval_runtime": 30.1447,
      "eval_samples_per_second": 331.733,
      "eval_steps_per_second": 1.327,
      "step": 64602
    },
    {
      "epoch": 194.02,
      "grad_norm": 5.834340572357178,
      "learning_rate": 3.532532532532533e-06,
      "loss": 0.2343,
      "step": 64610
    },
    {
      "epoch": 194.05,
      "grad_norm": 5.449021339416504,
      "learning_rate": 3.531531531531532e-06,
      "loss": 0.1904,
      "step": 64620
    },
    {
      "epoch": 194.08,
      "grad_norm": 8.815800666809082,
      "learning_rate": 3.530530530530531e-06,
      "loss": 0.2318,
      "step": 64630
    },
    {
      "epoch": 194.11,
      "grad_norm": 5.805193901062012,
      "learning_rate": 3.5295295295295295e-06,
      "loss": 0.1974,
      "step": 64640
    },
    {
      "epoch": 194.14,
      "grad_norm": 12.561065673828125,
      "learning_rate": 3.5285285285285286e-06,
      "loss": 0.2231,
      "step": 64650
    },
    {
      "epoch": 194.17,
      "grad_norm": 10.698709487915039,
      "learning_rate": 3.527527527527528e-06,
      "loss": 0.2442,
      "step": 64660
    },
    {
      "epoch": 194.2,
      "grad_norm": 7.310090065002441,
      "learning_rate": 3.526526526526527e-06,
      "loss": 0.1998,
      "step": 64670
    },
    {
      "epoch": 194.23,
      "grad_norm": 6.183576583862305,
      "learning_rate": 3.525525525525526e-06,
      "loss": 0.1714,
      "step": 64680
    },
    {
      "epoch": 194.26,
      "grad_norm": 9.807520866394043,
      "learning_rate": 3.5245245245245246e-06,
      "loss": 0.2009,
      "step": 64690
    },
    {
      "epoch": 194.29,
      "grad_norm": 6.8242621421813965,
      "learning_rate": 3.5235235235235237e-06,
      "loss": 0.2187,
      "step": 64700
    },
    {
      "epoch": 194.32,
      "grad_norm": 10.214282035827637,
      "learning_rate": 3.522522522522523e-06,
      "loss": 0.2248,
      "step": 64710
    },
    {
      "epoch": 194.35,
      "grad_norm": 10.87374496459961,
      "learning_rate": 3.521521521521522e-06,
      "loss": 0.2112,
      "step": 64720
    },
    {
      "epoch": 194.38,
      "grad_norm": 11.435918807983398,
      "learning_rate": 3.5205205205205207e-06,
      "loss": 0.1914,
      "step": 64730
    },
    {
      "epoch": 194.41,
      "grad_norm": 7.485071659088135,
      "learning_rate": 3.5195195195195197e-06,
      "loss": 0.1835,
      "step": 64740
    },
    {
      "epoch": 194.44,
      "grad_norm": 9.694918632507324,
      "learning_rate": 3.5185185185185187e-06,
      "loss": 0.2612,
      "step": 64750
    },
    {
      "epoch": 194.47,
      "grad_norm": 11.93000316619873,
      "learning_rate": 3.517517517517518e-06,
      "loss": 0.2069,
      "step": 64760
    },
    {
      "epoch": 194.5,
      "grad_norm": 7.05364990234375,
      "learning_rate": 3.5165165165165168e-06,
      "loss": 0.1847,
      "step": 64770
    },
    {
      "epoch": 194.53,
      "grad_norm": 6.488698482513428,
      "learning_rate": 3.515515515515516e-06,
      "loss": 0.2071,
      "step": 64780
    },
    {
      "epoch": 194.56,
      "grad_norm": 11.087560653686523,
      "learning_rate": 3.514514514514515e-06,
      "loss": 0.2,
      "step": 64790
    },
    {
      "epoch": 194.59,
      "grad_norm": 9.773676872253418,
      "learning_rate": 3.513513513513514e-06,
      "loss": 0.2382,
      "step": 64800
    },
    {
      "epoch": 194.62,
      "grad_norm": 7.835973262786865,
      "learning_rate": 3.5125125125125124e-06,
      "loss": 0.1801,
      "step": 64810
    },
    {
      "epoch": 194.65,
      "grad_norm": 12.473721504211426,
      "learning_rate": 3.511511511511512e-06,
      "loss": 0.2143,
      "step": 64820
    },
    {
      "epoch": 194.68,
      "grad_norm": 7.515393257141113,
      "learning_rate": 3.510510510510511e-06,
      "loss": 0.1767,
      "step": 64830
    },
    {
      "epoch": 194.71,
      "grad_norm": 9.571365356445312,
      "learning_rate": 3.50950950950951e-06,
      "loss": 0.2,
      "step": 64840
    },
    {
      "epoch": 194.74,
      "grad_norm": 9.731210708618164,
      "learning_rate": 3.5085085085085085e-06,
      "loss": 0.2034,
      "step": 64850
    },
    {
      "epoch": 194.77,
      "grad_norm": 8.72946834564209,
      "learning_rate": 3.5075075075075075e-06,
      "loss": 0.1779,
      "step": 64860
    },
    {
      "epoch": 194.8,
      "grad_norm": 10.727068901062012,
      "learning_rate": 3.506506506506507e-06,
      "loss": 0.1838,
      "step": 64870
    },
    {
      "epoch": 194.83,
      "grad_norm": 11.529535293579102,
      "learning_rate": 3.505505505505506e-06,
      "loss": 0.2452,
      "step": 64880
    },
    {
      "epoch": 194.86,
      "grad_norm": 7.009379863739014,
      "learning_rate": 3.5045045045045046e-06,
      "loss": 0.2367,
      "step": 64890
    },
    {
      "epoch": 194.89,
      "grad_norm": 9.535465240478516,
      "learning_rate": 3.5035035035035036e-06,
      "loss": 0.1969,
      "step": 64900
    },
    {
      "epoch": 194.92,
      "grad_norm": 6.760786056518555,
      "learning_rate": 3.5025025025025026e-06,
      "loss": 0.2191,
      "step": 64910
    },
    {
      "epoch": 194.95,
      "grad_norm": 18.083711624145508,
      "learning_rate": 3.501501501501502e-06,
      "loss": 0.1786,
      "step": 64920
    },
    {
      "epoch": 194.98,
      "grad_norm": 11.143521308898926,
      "learning_rate": 3.500500500500501e-06,
      "loss": 0.2485,
      "step": 64930
    },
    {
      "epoch": 195.0,
      "eval_accuracy": 0.9224,
      "eval_loss": 0.34122422337532043,
      "eval_runtime": 29.7476,
      "eval_samples_per_second": 336.161,
      "eval_steps_per_second": 1.345,
      "step": 64935
    },
    {
      "epoch": 195.02,
      "grad_norm": 10.117525100708008,
      "learning_rate": 3.4994994994994997e-06,
      "loss": 0.1826,
      "step": 64940
    },
    {
      "epoch": 195.05,
      "grad_norm": 12.360014915466309,
      "learning_rate": 3.4984984984984987e-06,
      "loss": 0.2166,
      "step": 64950
    },
    {
      "epoch": 195.08,
      "grad_norm": 8.09012508392334,
      "learning_rate": 3.4974974974974977e-06,
      "loss": 0.2438,
      "step": 64960
    },
    {
      "epoch": 195.11,
      "grad_norm": 6.431150913238525,
      "learning_rate": 3.496496496496497e-06,
      "loss": 0.1895,
      "step": 64970
    },
    {
      "epoch": 195.14,
      "grad_norm": 7.597049713134766,
      "learning_rate": 3.4954954954954957e-06,
      "loss": 0.191,
      "step": 64980
    },
    {
      "epoch": 195.17,
      "grad_norm": 14.608752250671387,
      "learning_rate": 3.4944944944944948e-06,
      "loss": 0.1999,
      "step": 64990
    },
    {
      "epoch": 195.2,
      "grad_norm": 6.014109134674072,
      "learning_rate": 3.4934934934934938e-06,
      "loss": 0.1974,
      "step": 65000
    },
    {
      "epoch": 195.23,
      "grad_norm": 9.761489868164062,
      "learning_rate": 3.492492492492493e-06,
      "loss": 0.2331,
      "step": 65010
    },
    {
      "epoch": 195.26,
      "grad_norm": 71.3414535522461,
      "learning_rate": 3.4914914914914914e-06,
      "loss": 0.2342,
      "step": 65020
    },
    {
      "epoch": 195.29,
      "grad_norm": 12.985770225524902,
      "learning_rate": 3.490490490490491e-06,
      "loss": 0.2114,
      "step": 65030
    },
    {
      "epoch": 195.32,
      "grad_norm": 5.221217632293701,
      "learning_rate": 3.48948948948949e-06,
      "loss": 0.2075,
      "step": 65040
    },
    {
      "epoch": 195.35,
      "grad_norm": 7.139613151550293,
      "learning_rate": 3.488488488488489e-06,
      "loss": 0.2261,
      "step": 65050
    },
    {
      "epoch": 195.38,
      "grad_norm": 8.551673889160156,
      "learning_rate": 3.4874874874874875e-06,
      "loss": 0.1604,
      "step": 65060
    },
    {
      "epoch": 195.41,
      "grad_norm": 8.033476829528809,
      "learning_rate": 3.4864864864864865e-06,
      "loss": 0.2098,
      "step": 65070
    },
    {
      "epoch": 195.44,
      "grad_norm": 10.748931884765625,
      "learning_rate": 3.485485485485486e-06,
      "loss": 0.2145,
      "step": 65080
    },
    {
      "epoch": 195.47,
      "grad_norm": 9.663415908813477,
      "learning_rate": 3.484484484484485e-06,
      "loss": 0.1956,
      "step": 65090
    },
    {
      "epoch": 195.5,
      "grad_norm": 12.189712524414062,
      "learning_rate": 3.4834834834834835e-06,
      "loss": 0.221,
      "step": 65100
    },
    {
      "epoch": 195.53,
      "grad_norm": 7.0851030349731445,
      "learning_rate": 3.4824824824824826e-06,
      "loss": 0.242,
      "step": 65110
    },
    {
      "epoch": 195.56,
      "grad_norm": 9.688054084777832,
      "learning_rate": 3.481481481481482e-06,
      "loss": 0.1949,
      "step": 65120
    },
    {
      "epoch": 195.59,
      "grad_norm": 11.332962989807129,
      "learning_rate": 3.480480480480481e-06,
      "loss": 0.2004,
      "step": 65130
    },
    {
      "epoch": 195.62,
      "grad_norm": 9.34135913848877,
      "learning_rate": 3.4794794794794796e-06,
      "loss": 0.2234,
      "step": 65140
    },
    {
      "epoch": 195.65,
      "grad_norm": 8.15683650970459,
      "learning_rate": 3.4784784784784786e-06,
      "loss": 0.1918,
      "step": 65150
    },
    {
      "epoch": 195.68,
      "grad_norm": 11.069890975952148,
      "learning_rate": 3.4774774774774776e-06,
      "loss": 0.1837,
      "step": 65160
    },
    {
      "epoch": 195.71,
      "grad_norm": 8.037181854248047,
      "learning_rate": 3.476476476476477e-06,
      "loss": 0.2392,
      "step": 65170
    },
    {
      "epoch": 195.74,
      "grad_norm": 8.973856925964355,
      "learning_rate": 3.475475475475476e-06,
      "loss": 0.2256,
      "step": 65180
    },
    {
      "epoch": 195.77,
      "grad_norm": 7.398205757141113,
      "learning_rate": 3.4744744744744747e-06,
      "loss": 0.2097,
      "step": 65190
    },
    {
      "epoch": 195.8,
      "grad_norm": 6.761328220367432,
      "learning_rate": 3.4734734734734737e-06,
      "loss": 0.2366,
      "step": 65200
    },
    {
      "epoch": 195.83,
      "grad_norm": 7.453217029571533,
      "learning_rate": 3.4724724724724727e-06,
      "loss": 0.2071,
      "step": 65210
    },
    {
      "epoch": 195.86,
      "grad_norm": 5.910012722015381,
      "learning_rate": 3.471471471471472e-06,
      "loss": 0.2004,
      "step": 65220
    },
    {
      "epoch": 195.89,
      "grad_norm": 8.72472858428955,
      "learning_rate": 3.4704704704704708e-06,
      "loss": 0.1951,
      "step": 65230
    },
    {
      "epoch": 195.92,
      "grad_norm": 8.996418952941895,
      "learning_rate": 3.46946946946947e-06,
      "loss": 0.2188,
      "step": 65240
    },
    {
      "epoch": 195.95,
      "grad_norm": 15.691521644592285,
      "learning_rate": 3.468468468468469e-06,
      "loss": 0.1855,
      "step": 65250
    },
    {
      "epoch": 195.98,
      "grad_norm": 5.117032051086426,
      "learning_rate": 3.467467467467468e-06,
      "loss": 0.2055,
      "step": 65260
    },
    {
      "epoch": 196.0,
      "eval_accuracy": 0.9209,
      "eval_loss": 0.3426472544670105,
      "eval_runtime": 30.0105,
      "eval_samples_per_second": 333.217,
      "eval_steps_per_second": 1.333,
      "step": 65268
    },
    {
      "epoch": 196.01,
      "grad_norm": 15.594942092895508,
      "learning_rate": 3.4664664664664664e-06,
      "loss": 0.2122,
      "step": 65270
    },
    {
      "epoch": 196.04,
      "grad_norm": 8.108100891113281,
      "learning_rate": 3.465465465465466e-06,
      "loss": 0.2573,
      "step": 65280
    },
    {
      "epoch": 196.07,
      "grad_norm": 24.62065887451172,
      "learning_rate": 3.464464464464465e-06,
      "loss": 0.1875,
      "step": 65290
    },
    {
      "epoch": 196.1,
      "grad_norm": 7.312531471252441,
      "learning_rate": 3.463463463463464e-06,
      "loss": 0.2344,
      "step": 65300
    },
    {
      "epoch": 196.13,
      "grad_norm": 5.945933818817139,
      "learning_rate": 3.4624624624624625e-06,
      "loss": 0.203,
      "step": 65310
    },
    {
      "epoch": 196.16,
      "grad_norm": 7.343894004821777,
      "learning_rate": 3.4614614614614615e-06,
      "loss": 0.2484,
      "step": 65320
    },
    {
      "epoch": 196.19,
      "grad_norm": 8.099315643310547,
      "learning_rate": 3.460460460460461e-06,
      "loss": 0.2122,
      "step": 65330
    },
    {
      "epoch": 196.22,
      "grad_norm": 7.715710639953613,
      "learning_rate": 3.45945945945946e-06,
      "loss": 0.2495,
      "step": 65340
    },
    {
      "epoch": 196.25,
      "grad_norm": 11.764954566955566,
      "learning_rate": 3.4584584584584586e-06,
      "loss": 0.2116,
      "step": 65350
    },
    {
      "epoch": 196.28,
      "grad_norm": 8.756845474243164,
      "learning_rate": 3.4574574574574576e-06,
      "loss": 0.1842,
      "step": 65360
    },
    {
      "epoch": 196.31,
      "grad_norm": 24.654842376708984,
      "learning_rate": 3.4564564564564566e-06,
      "loss": 0.2021,
      "step": 65370
    },
    {
      "epoch": 196.34,
      "grad_norm": 6.7150139808654785,
      "learning_rate": 3.455455455455456e-06,
      "loss": 0.139,
      "step": 65380
    },
    {
      "epoch": 196.37,
      "grad_norm": 9.088079452514648,
      "learning_rate": 3.4544544544544546e-06,
      "loss": 0.216,
      "step": 65390
    },
    {
      "epoch": 196.4,
      "grad_norm": 7.394824028015137,
      "learning_rate": 3.4534534534534537e-06,
      "loss": 0.1903,
      "step": 65400
    },
    {
      "epoch": 196.43,
      "grad_norm": 8.45055866241455,
      "learning_rate": 3.4524524524524527e-06,
      "loss": 0.2102,
      "step": 65410
    },
    {
      "epoch": 196.46,
      "grad_norm": 10.756978034973145,
      "learning_rate": 3.4514514514514517e-06,
      "loss": 0.1972,
      "step": 65420
    },
    {
      "epoch": 196.49,
      "grad_norm": 7.751649856567383,
      "learning_rate": 3.4504504504504503e-06,
      "loss": 0.2197,
      "step": 65430
    },
    {
      "epoch": 196.52,
      "grad_norm": 33.353912353515625,
      "learning_rate": 3.4494494494494497e-06,
      "loss": 0.2175,
      "step": 65440
    },
    {
      "epoch": 196.55,
      "grad_norm": 12.569924354553223,
      "learning_rate": 3.4484484484484488e-06,
      "loss": 0.2183,
      "step": 65450
    },
    {
      "epoch": 196.58,
      "grad_norm": 8.212811470031738,
      "learning_rate": 3.4474474474474478e-06,
      "loss": 0.232,
      "step": 65460
    },
    {
      "epoch": 196.61,
      "grad_norm": 5.135364532470703,
      "learning_rate": 3.446446446446447e-06,
      "loss": 0.1816,
      "step": 65470
    },
    {
      "epoch": 196.64,
      "grad_norm": 10.538021087646484,
      "learning_rate": 3.4454454454454454e-06,
      "loss": 0.1837,
      "step": 65480
    },
    {
      "epoch": 196.67,
      "grad_norm": 14.32591724395752,
      "learning_rate": 3.444444444444445e-06,
      "loss": 0.1974,
      "step": 65490
    },
    {
      "epoch": 196.7,
      "grad_norm": 10.894963264465332,
      "learning_rate": 3.443443443443444e-06,
      "loss": 0.2013,
      "step": 65500
    },
    {
      "epoch": 196.73,
      "grad_norm": 10.601717948913574,
      "learning_rate": 3.442442442442443e-06,
      "loss": 0.2478,
      "step": 65510
    },
    {
      "epoch": 196.76,
      "grad_norm": 16.711280822753906,
      "learning_rate": 3.4414414414414415e-06,
      "loss": 0.2533,
      "step": 65520
    },
    {
      "epoch": 196.79,
      "grad_norm": 8.361686706542969,
      "learning_rate": 3.4404404404404405e-06,
      "loss": 0.2007,
      "step": 65530
    },
    {
      "epoch": 196.82,
      "grad_norm": 7.885601997375488,
      "learning_rate": 3.43943943943944e-06,
      "loss": 0.2017,
      "step": 65540
    },
    {
      "epoch": 196.85,
      "grad_norm": 9.95803451538086,
      "learning_rate": 3.438438438438439e-06,
      "loss": 0.1828,
      "step": 65550
    },
    {
      "epoch": 196.88,
      "grad_norm": 7.6867451667785645,
      "learning_rate": 3.4374374374374375e-06,
      "loss": 0.2365,
      "step": 65560
    },
    {
      "epoch": 196.91,
      "grad_norm": 7.916510581970215,
      "learning_rate": 3.4364364364364366e-06,
      "loss": 0.194,
      "step": 65570
    },
    {
      "epoch": 196.94,
      "grad_norm": 10.3589448928833,
      "learning_rate": 3.435435435435436e-06,
      "loss": 0.1975,
      "step": 65580
    },
    {
      "epoch": 196.97,
      "grad_norm": 11.903532981872559,
      "learning_rate": 3.434434434434435e-06,
      "loss": 0.1776,
      "step": 65590
    },
    {
      "epoch": 197.0,
      "grad_norm": 4.4184184074401855,
      "learning_rate": 3.4334334334334336e-06,
      "loss": 0.2087,
      "step": 65600
    },
    {
      "epoch": 197.0,
      "eval_accuracy": 0.92,
      "eval_loss": 0.34563955664634705,
      "eval_runtime": 30.1929,
      "eval_samples_per_second": 331.204,
      "eval_steps_per_second": 1.325,
      "step": 65601
    },
    {
      "epoch": 197.03,
      "grad_norm": 8.926599502563477,
      "learning_rate": 3.4324324324324326e-06,
      "loss": 0.2232,
      "step": 65610
    },
    {
      "epoch": 197.06,
      "grad_norm": 7.414474010467529,
      "learning_rate": 3.4314314314314316e-06,
      "loss": 0.1935,
      "step": 65620
    },
    {
      "epoch": 197.09,
      "grad_norm": 7.8957085609436035,
      "learning_rate": 3.430430430430431e-06,
      "loss": 0.2233,
      "step": 65630
    },
    {
      "epoch": 197.12,
      "grad_norm": 11.80379581451416,
      "learning_rate": 3.4294294294294293e-06,
      "loss": 0.1728,
      "step": 65640
    },
    {
      "epoch": 197.15,
      "grad_norm": 7.742220878601074,
      "learning_rate": 3.4284284284284287e-06,
      "loss": 0.2321,
      "step": 65650
    },
    {
      "epoch": 197.18,
      "grad_norm": 7.248086452484131,
      "learning_rate": 3.4274274274274277e-06,
      "loss": 0.237,
      "step": 65660
    },
    {
      "epoch": 197.21,
      "grad_norm": 9.90302848815918,
      "learning_rate": 3.4264264264264267e-06,
      "loss": 0.2204,
      "step": 65670
    },
    {
      "epoch": 197.24,
      "grad_norm": 8.414325714111328,
      "learning_rate": 3.4254254254254253e-06,
      "loss": 0.1675,
      "step": 65680
    },
    {
      "epoch": 197.27,
      "grad_norm": 8.74123764038086,
      "learning_rate": 3.4244244244244248e-06,
      "loss": 0.2496,
      "step": 65690
    },
    {
      "epoch": 197.3,
      "grad_norm": 7.870460510253906,
      "learning_rate": 3.423423423423424e-06,
      "loss": 0.1756,
      "step": 65700
    },
    {
      "epoch": 197.33,
      "grad_norm": 10.371129035949707,
      "learning_rate": 3.422422422422423e-06,
      "loss": 0.2296,
      "step": 65710
    },
    {
      "epoch": 197.36,
      "grad_norm": 6.789724826812744,
      "learning_rate": 3.421421421421422e-06,
      "loss": 0.2549,
      "step": 65720
    },
    {
      "epoch": 197.39,
      "grad_norm": 11.534571647644043,
      "learning_rate": 3.4204204204204204e-06,
      "loss": 0.2295,
      "step": 65730
    },
    {
      "epoch": 197.42,
      "grad_norm": 14.271658897399902,
      "learning_rate": 3.41941941941942e-06,
      "loss": 0.1921,
      "step": 65740
    },
    {
      "epoch": 197.45,
      "grad_norm": 6.887492656707764,
      "learning_rate": 3.418418418418419e-06,
      "loss": 0.1801,
      "step": 65750
    },
    {
      "epoch": 197.48,
      "grad_norm": 6.9644646644592285,
      "learning_rate": 3.417417417417418e-06,
      "loss": 0.2062,
      "step": 65760
    },
    {
      "epoch": 197.51,
      "grad_norm": 18.947124481201172,
      "learning_rate": 3.4164164164164165e-06,
      "loss": 0.2132,
      "step": 65770
    },
    {
      "epoch": 197.54,
      "grad_norm": 7.977654457092285,
      "learning_rate": 3.4154154154154155e-06,
      "loss": 0.2048,
      "step": 65780
    },
    {
      "epoch": 197.57,
      "grad_norm": 7.148995399475098,
      "learning_rate": 3.414414414414415e-06,
      "loss": 0.24,
      "step": 65790
    },
    {
      "epoch": 197.6,
      "grad_norm": 6.497589588165283,
      "learning_rate": 3.413413413413414e-06,
      "loss": 0.2117,
      "step": 65800
    },
    {
      "epoch": 197.63,
      "grad_norm": 5.678431987762451,
      "learning_rate": 3.4124124124124126e-06,
      "loss": 0.1949,
      "step": 65810
    },
    {
      "epoch": 197.66,
      "grad_norm": 9.441134452819824,
      "learning_rate": 3.4114114114114116e-06,
      "loss": 0.2406,
      "step": 65820
    },
    {
      "epoch": 197.69,
      "grad_norm": 14.110630989074707,
      "learning_rate": 3.4104104104104106e-06,
      "loss": 0.2011,
      "step": 65830
    },
    {
      "epoch": 197.72,
      "grad_norm": 7.9216108322143555,
      "learning_rate": 3.40940940940941e-06,
      "loss": 0.2066,
      "step": 65840
    },
    {
      "epoch": 197.75,
      "grad_norm": 10.073457717895508,
      "learning_rate": 3.4084084084084086e-06,
      "loss": 0.2425,
      "step": 65850
    },
    {
      "epoch": 197.78,
      "grad_norm": 10.882193565368652,
      "learning_rate": 3.4074074074074077e-06,
      "loss": 0.2215,
      "step": 65860
    },
    {
      "epoch": 197.81,
      "grad_norm": 9.150156021118164,
      "learning_rate": 3.4064064064064067e-06,
      "loss": 0.1795,
      "step": 65870
    },
    {
      "epoch": 197.84,
      "grad_norm": 9.575854301452637,
      "learning_rate": 3.4054054054054057e-06,
      "loss": 0.1886,
      "step": 65880
    },
    {
      "epoch": 197.87,
      "grad_norm": 7.394041061401367,
      "learning_rate": 3.4044044044044043e-06,
      "loss": 0.1859,
      "step": 65890
    },
    {
      "epoch": 197.9,
      "grad_norm": 6.381305694580078,
      "learning_rate": 3.4034034034034037e-06,
      "loss": 0.2512,
      "step": 65900
    },
    {
      "epoch": 197.93,
      "grad_norm": 5.0840678215026855,
      "learning_rate": 3.4024024024024028e-06,
      "loss": 0.2049,
      "step": 65910
    },
    {
      "epoch": 197.96,
      "grad_norm": 8.702017784118652,
      "learning_rate": 3.4014014014014018e-06,
      "loss": 0.2109,
      "step": 65920
    },
    {
      "epoch": 197.99,
      "grad_norm": 10.111449241638184,
      "learning_rate": 3.4004004004004004e-06,
      "loss": 0.235,
      "step": 65930
    },
    {
      "epoch": 198.0,
      "eval_accuracy": 0.9218,
      "eval_loss": 0.3437337577342987,
      "eval_runtime": 30.1104,
      "eval_samples_per_second": 332.112,
      "eval_steps_per_second": 1.328,
      "step": 65934
    },
    {
      "epoch": 198.02,
      "grad_norm": 7.970890045166016,
      "learning_rate": 3.3993993993993994e-06,
      "loss": 0.2124,
      "step": 65940
    },
    {
      "epoch": 198.05,
      "grad_norm": 13.795842170715332,
      "learning_rate": 3.398398398398399e-06,
      "loss": 0.1691,
      "step": 65950
    },
    {
      "epoch": 198.08,
      "grad_norm": 12.441973686218262,
      "learning_rate": 3.397397397397398e-06,
      "loss": 0.2182,
      "step": 65960
    },
    {
      "epoch": 198.11,
      "grad_norm": 7.903041839599609,
      "learning_rate": 3.3963963963963964e-06,
      "loss": 0.1925,
      "step": 65970
    },
    {
      "epoch": 198.14,
      "grad_norm": 10.109343528747559,
      "learning_rate": 3.3953953953953955e-06,
      "loss": 0.2136,
      "step": 65980
    },
    {
      "epoch": 198.17,
      "grad_norm": 21.193334579467773,
      "learning_rate": 3.3943943943943945e-06,
      "loss": 0.1976,
      "step": 65990
    },
    {
      "epoch": 198.2,
      "grad_norm": 7.796104431152344,
      "learning_rate": 3.393393393393394e-06,
      "loss": 0.1706,
      "step": 66000
    },
    {
      "epoch": 198.23,
      "grad_norm": 8.311440467834473,
      "learning_rate": 3.392392392392393e-06,
      "loss": 0.2018,
      "step": 66010
    },
    {
      "epoch": 198.26,
      "grad_norm": 16.57115364074707,
      "learning_rate": 3.3913913913913915e-06,
      "loss": 0.1899,
      "step": 66020
    },
    {
      "epoch": 198.29,
      "grad_norm": 13.53213882446289,
      "learning_rate": 3.3903903903903905e-06,
      "loss": 0.1885,
      "step": 66030
    },
    {
      "epoch": 198.32,
      "grad_norm": 9.399784088134766,
      "learning_rate": 3.3893893893893896e-06,
      "loss": 0.1929,
      "step": 66040
    },
    {
      "epoch": 198.35,
      "grad_norm": 7.098597049713135,
      "learning_rate": 3.388388388388389e-06,
      "loss": 0.2142,
      "step": 66050
    },
    {
      "epoch": 198.38,
      "grad_norm": 9.350313186645508,
      "learning_rate": 3.3873873873873876e-06,
      "loss": 0.2357,
      "step": 66060
    },
    {
      "epoch": 198.41,
      "grad_norm": 11.781498908996582,
      "learning_rate": 3.3863863863863866e-06,
      "loss": 0.2092,
      "step": 66070
    },
    {
      "epoch": 198.44,
      "grad_norm": 8.887947082519531,
      "learning_rate": 3.3853853853853856e-06,
      "loss": 0.2548,
      "step": 66080
    },
    {
      "epoch": 198.47,
      "grad_norm": 9.210660934448242,
      "learning_rate": 3.384384384384385e-06,
      "loss": 0.1864,
      "step": 66090
    },
    {
      "epoch": 198.5,
      "grad_norm": 10.353461265563965,
      "learning_rate": 3.3833833833833833e-06,
      "loss": 0.2313,
      "step": 66100
    },
    {
      "epoch": 198.53,
      "grad_norm": 10.597054481506348,
      "learning_rate": 3.3823823823823827e-06,
      "loss": 0.2179,
      "step": 66110
    },
    {
      "epoch": 198.56,
      "grad_norm": 6.533252716064453,
      "learning_rate": 3.3813813813813817e-06,
      "loss": 0.2236,
      "step": 66120
    },
    {
      "epoch": 198.59,
      "grad_norm": 9.13060474395752,
      "learning_rate": 3.3803803803803807e-06,
      "loss": 0.1964,
      "step": 66130
    },
    {
      "epoch": 198.62,
      "grad_norm": 13.791937828063965,
      "learning_rate": 3.3793793793793793e-06,
      "loss": 0.1995,
      "step": 66140
    },
    {
      "epoch": 198.65,
      "grad_norm": 13.477198600769043,
      "learning_rate": 3.3783783783783788e-06,
      "loss": 0.179,
      "step": 66150
    },
    {
      "epoch": 198.68,
      "grad_norm": 9.87004280090332,
      "learning_rate": 3.3773773773773778e-06,
      "loss": 0.2084,
      "step": 66160
    },
    {
      "epoch": 198.71,
      "grad_norm": 13.279138565063477,
      "learning_rate": 3.376376376376377e-06,
      "loss": 0.1797,
      "step": 66170
    },
    {
      "epoch": 198.74,
      "grad_norm": 4.959941387176514,
      "learning_rate": 3.3753753753753754e-06,
      "loss": 0.2095,
      "step": 66180
    },
    {
      "epoch": 198.77,
      "grad_norm": 8.283129692077637,
      "learning_rate": 3.3743743743743744e-06,
      "loss": 0.2011,
      "step": 66190
    },
    {
      "epoch": 198.8,
      "grad_norm": 7.824742794036865,
      "learning_rate": 3.373373373373374e-06,
      "loss": 0.1662,
      "step": 66200
    },
    {
      "epoch": 198.83,
      "grad_norm": 11.038658142089844,
      "learning_rate": 3.372372372372373e-06,
      "loss": 0.2414,
      "step": 66210
    },
    {
      "epoch": 198.86,
      "grad_norm": 7.395485877990723,
      "learning_rate": 3.3713713713713715e-06,
      "loss": 0.178,
      "step": 66220
    },
    {
      "epoch": 198.89,
      "grad_norm": 8.222171783447266,
      "learning_rate": 3.3703703703703705e-06,
      "loss": 0.211,
      "step": 66230
    },
    {
      "epoch": 198.92,
      "grad_norm": 8.179375648498535,
      "learning_rate": 3.3693693693693695e-06,
      "loss": 0.1802,
      "step": 66240
    },
    {
      "epoch": 198.95,
      "grad_norm": 6.363645553588867,
      "learning_rate": 3.368368368368369e-06,
      "loss": 0.1985,
      "step": 66250
    },
    {
      "epoch": 198.98,
      "grad_norm": 6.243249416351318,
      "learning_rate": 3.367367367367368e-06,
      "loss": 0.2093,
      "step": 66260
    },
    {
      "epoch": 199.0,
      "eval_accuracy": 0.9231,
      "eval_loss": 0.3425036668777466,
      "eval_runtime": 30.0565,
      "eval_samples_per_second": 332.707,
      "eval_steps_per_second": 1.331,
      "step": 66267
    },
    {
      "epoch": 199.01,
      "grad_norm": 8.165533065795898,
      "learning_rate": 3.3663663663663666e-06,
      "loss": 0.2122,
      "step": 66270
    },
    {
      "epoch": 199.04,
      "grad_norm": 9.406436920166016,
      "learning_rate": 3.3653653653653656e-06,
      "loss": 0.1936,
      "step": 66280
    },
    {
      "epoch": 199.07,
      "grad_norm": 13.659327507019043,
      "learning_rate": 3.3643643643643646e-06,
      "loss": 0.2051,
      "step": 66290
    },
    {
      "epoch": 199.1,
      "grad_norm": 10.759275436401367,
      "learning_rate": 3.363363363363364e-06,
      "loss": 0.2249,
      "step": 66300
    },
    {
      "epoch": 199.13,
      "grad_norm": 12.736884117126465,
      "learning_rate": 3.3623623623623626e-06,
      "loss": 0.1968,
      "step": 66310
    },
    {
      "epoch": 199.16,
      "grad_norm": 8.873373031616211,
      "learning_rate": 3.3613613613613617e-06,
      "loss": 0.1739,
      "step": 66320
    },
    {
      "epoch": 199.19,
      "grad_norm": 13.299247741699219,
      "learning_rate": 3.3603603603603607e-06,
      "loss": 0.2204,
      "step": 66330
    },
    {
      "epoch": 199.22,
      "grad_norm": 32.7912712097168,
      "learning_rate": 3.3593593593593597e-06,
      "loss": 0.2628,
      "step": 66340
    },
    {
      "epoch": 199.25,
      "grad_norm": 12.069893836975098,
      "learning_rate": 3.3583583583583583e-06,
      "loss": 0.2263,
      "step": 66350
    },
    {
      "epoch": 199.28,
      "grad_norm": 7.298553466796875,
      "learning_rate": 3.3573573573573577e-06,
      "loss": 0.2024,
      "step": 66360
    },
    {
      "epoch": 199.31,
      "grad_norm": 12.512067794799805,
      "learning_rate": 3.3563563563563567e-06,
      "loss": 0.19,
      "step": 66370
    },
    {
      "epoch": 199.34,
      "grad_norm": 21.682897567749023,
      "learning_rate": 3.3553553553553558e-06,
      "loss": 0.2182,
      "step": 66380
    },
    {
      "epoch": 199.37,
      "grad_norm": 4.5556182861328125,
      "learning_rate": 3.3543543543543544e-06,
      "loss": 0.2274,
      "step": 66390
    },
    {
      "epoch": 199.4,
      "grad_norm": 6.911951065063477,
      "learning_rate": 3.3533533533533534e-06,
      "loss": 0.175,
      "step": 66400
    },
    {
      "epoch": 199.43,
      "grad_norm": 7.6049628257751465,
      "learning_rate": 3.352352352352353e-06,
      "loss": 0.1865,
      "step": 66410
    },
    {
      "epoch": 199.46,
      "grad_norm": 7.027915954589844,
      "learning_rate": 3.351351351351352e-06,
      "loss": 0.1712,
      "step": 66420
    },
    {
      "epoch": 199.49,
      "grad_norm": 6.528171539306641,
      "learning_rate": 3.3503503503503504e-06,
      "loss": 0.2032,
      "step": 66430
    },
    {
      "epoch": 199.52,
      "grad_norm": 9.243008613586426,
      "learning_rate": 3.3493493493493495e-06,
      "loss": 0.2201,
      "step": 66440
    },
    {
      "epoch": 199.55,
      "grad_norm": 14.192416191101074,
      "learning_rate": 3.3483483483483485e-06,
      "loss": 0.2348,
      "step": 66450
    },
    {
      "epoch": 199.58,
      "grad_norm": 10.812016487121582,
      "learning_rate": 3.347347347347348e-06,
      "loss": 0.1968,
      "step": 66460
    },
    {
      "epoch": 199.61,
      "grad_norm": 8.774687767028809,
      "learning_rate": 3.3463463463463465e-06,
      "loss": 0.206,
      "step": 66470
    },
    {
      "epoch": 199.64,
      "grad_norm": 10.574451446533203,
      "learning_rate": 3.3453453453453455e-06,
      "loss": 0.1872,
      "step": 66480
    },
    {
      "epoch": 199.67,
      "grad_norm": 6.8933563232421875,
      "learning_rate": 3.3443443443443445e-06,
      "loss": 0.2041,
      "step": 66490
    },
    {
      "epoch": 199.7,
      "grad_norm": 8.988615989685059,
      "learning_rate": 3.3433433433433436e-06,
      "loss": 0.1777,
      "step": 66500
    },
    {
      "epoch": 199.73,
      "grad_norm": 6.239690780639648,
      "learning_rate": 3.342342342342342e-06,
      "loss": 0.1804,
      "step": 66510
    },
    {
      "epoch": 199.76,
      "grad_norm": 20.342754364013672,
      "learning_rate": 3.3413413413413416e-06,
      "loss": 0.2307,
      "step": 66520
    },
    {
      "epoch": 199.79,
      "grad_norm": 16.23260498046875,
      "learning_rate": 3.3403403403403406e-06,
      "loss": 0.2192,
      "step": 66530
    },
    {
      "epoch": 199.82,
      "grad_norm": 8.545127868652344,
      "learning_rate": 3.3393393393393396e-06,
      "loss": 0.1779,
      "step": 66540
    },
    {
      "epoch": 199.85,
      "grad_norm": 7.973302364349365,
      "learning_rate": 3.338338338338339e-06,
      "loss": 0.1887,
      "step": 66550
    },
    {
      "epoch": 199.88,
      "grad_norm": 13.314821243286133,
      "learning_rate": 3.3373373373373373e-06,
      "loss": 0.2237,
      "step": 66560
    },
    {
      "epoch": 199.91,
      "grad_norm": 7.705291271209717,
      "learning_rate": 3.3363363363363367e-06,
      "loss": 0.2227,
      "step": 66570
    },
    {
      "epoch": 199.94,
      "grad_norm": 7.75813102722168,
      "learning_rate": 3.3353353353353357e-06,
      "loss": 0.2191,
      "step": 66580
    },
    {
      "epoch": 199.97,
      "grad_norm": 10.53028392791748,
      "learning_rate": 3.3343343343343347e-06,
      "loss": 0.1945,
      "step": 66590
    },
    {
      "epoch": 200.0,
      "grad_norm": 0.004936050623655319,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.1899,
      "step": 66600
    },
    {
      "epoch": 200.0,
      "eval_accuracy": 0.9229,
      "eval_loss": 0.34333765506744385,
      "eval_runtime": 30.1363,
      "eval_samples_per_second": 331.825,
      "eval_steps_per_second": 1.327,
      "step": 66600
    },
    {
      "epoch": 200.03,
      "grad_norm": 8.859107971191406,
      "learning_rate": 3.3323323323323323e-06,
      "loss": 0.2169,
      "step": 66610
    },
    {
      "epoch": 200.06,
      "grad_norm": 6.678070545196533,
      "learning_rate": 3.3313313313313318e-06,
      "loss": 0.1924,
      "step": 66620
    },
    {
      "epoch": 200.09,
      "grad_norm": 8.137578964233398,
      "learning_rate": 3.330330330330331e-06,
      "loss": 0.2235,
      "step": 66630
    },
    {
      "epoch": 200.12,
      "grad_norm": 8.310884475708008,
      "learning_rate": 3.3293293293293294e-06,
      "loss": 0.1913,
      "step": 66640
    },
    {
      "epoch": 200.15,
      "grad_norm": 9.3788480758667,
      "learning_rate": 3.3283283283283284e-06,
      "loss": 0.1909,
      "step": 66650
    },
    {
      "epoch": 200.18,
      "grad_norm": 6.447203159332275,
      "learning_rate": 3.327327327327328e-06,
      "loss": 0.2127,
      "step": 66660
    },
    {
      "epoch": 200.21,
      "grad_norm": 13.3621826171875,
      "learning_rate": 3.326326326326327e-06,
      "loss": 0.2161,
      "step": 66670
    },
    {
      "epoch": 200.24,
      "grad_norm": 10.807412147521973,
      "learning_rate": 3.3253253253253255e-06,
      "loss": 0.1927,
      "step": 66680
    },
    {
      "epoch": 200.27,
      "grad_norm": 9.319847106933594,
      "learning_rate": 3.3243243243243245e-06,
      "loss": 0.1868,
      "step": 66690
    },
    {
      "epoch": 200.3,
      "grad_norm": 13.919051170349121,
      "learning_rate": 3.3233233233233235e-06,
      "loss": 0.2213,
      "step": 66700
    },
    {
      "epoch": 200.33,
      "grad_norm": 9.84500789642334,
      "learning_rate": 3.322322322322323e-06,
      "loss": 0.2282,
      "step": 66710
    },
    {
      "epoch": 200.36,
      "grad_norm": 11.435735702514648,
      "learning_rate": 3.3213213213213215e-06,
      "loss": 0.23,
      "step": 66720
    },
    {
      "epoch": 200.39,
      "grad_norm": 6.1106486320495605,
      "learning_rate": 3.3203203203203206e-06,
      "loss": 0.2089,
      "step": 66730
    },
    {
      "epoch": 200.42,
      "grad_norm": 6.477038383483887,
      "learning_rate": 3.3193193193193196e-06,
      "loss": 0.1965,
      "step": 66740
    },
    {
      "epoch": 200.45,
      "grad_norm": 10.375381469726562,
      "learning_rate": 3.3183183183183186e-06,
      "loss": 0.2088,
      "step": 66750
    },
    {
      "epoch": 200.48,
      "grad_norm": 15.585830688476562,
      "learning_rate": 3.317317317317317e-06,
      "loss": 0.1959,
      "step": 66760
    },
    {
      "epoch": 200.51,
      "grad_norm": 6.875393867492676,
      "learning_rate": 3.3163163163163166e-06,
      "loss": 0.1893,
      "step": 66770
    },
    {
      "epoch": 200.54,
      "grad_norm": 7.31072998046875,
      "learning_rate": 3.3153153153153157e-06,
      "loss": 0.181,
      "step": 66780
    },
    {
      "epoch": 200.57,
      "grad_norm": 10.342710494995117,
      "learning_rate": 3.3143143143143147e-06,
      "loss": 0.1975,
      "step": 66790
    },
    {
      "epoch": 200.6,
      "grad_norm": 6.240497589111328,
      "learning_rate": 3.3133133133133137e-06,
      "loss": 0.1858,
      "step": 66800
    },
    {
      "epoch": 200.63,
      "grad_norm": 4.9630513191223145,
      "learning_rate": 3.3123123123123123e-06,
      "loss": 0.2031,
      "step": 66810
    },
    {
      "epoch": 200.66,
      "grad_norm": 24.101116180419922,
      "learning_rate": 3.3113113113113117e-06,
      "loss": 0.2114,
      "step": 66820
    },
    {
      "epoch": 200.69,
      "grad_norm": 8.5513916015625,
      "learning_rate": 3.3103103103103107e-06,
      "loss": 0.2668,
      "step": 66830
    },
    {
      "epoch": 200.72,
      "grad_norm": 10.798471450805664,
      "learning_rate": 3.3093093093093098e-06,
      "loss": 0.2448,
      "step": 66840
    },
    {
      "epoch": 200.75,
      "grad_norm": 18.340370178222656,
      "learning_rate": 3.3083083083083084e-06,
      "loss": 0.2318,
      "step": 66850
    },
    {
      "epoch": 200.78,
      "grad_norm": 7.375101089477539,
      "learning_rate": 3.3073073073073074e-06,
      "loss": 0.1843,
      "step": 66860
    },
    {
      "epoch": 200.81,
      "grad_norm": 8.373695373535156,
      "learning_rate": 3.306306306306307e-06,
      "loss": 0.2365,
      "step": 66870
    },
    {
      "epoch": 200.84,
      "grad_norm": 9.265231132507324,
      "learning_rate": 3.305305305305306e-06,
      "loss": 0.21,
      "step": 66880
    },
    {
      "epoch": 200.87,
      "grad_norm": 10.590888977050781,
      "learning_rate": 3.3043043043043044e-06,
      "loss": 0.1978,
      "step": 66890
    },
    {
      "epoch": 200.9,
      "grad_norm": 13.649727821350098,
      "learning_rate": 3.3033033033033035e-06,
      "loss": 0.2152,
      "step": 66900
    },
    {
      "epoch": 200.93,
      "grad_norm": 5.864605903625488,
      "learning_rate": 3.3023023023023025e-06,
      "loss": 0.2558,
      "step": 66910
    },
    {
      "epoch": 200.96,
      "grad_norm": 4.596671104431152,
      "learning_rate": 3.301301301301302e-06,
      "loss": 0.2217,
      "step": 66920
    },
    {
      "epoch": 200.99,
      "grad_norm": 6.2244744300842285,
      "learning_rate": 3.3003003003003005e-06,
      "loss": 0.231,
      "step": 66930
    },
    {
      "epoch": 201.0,
      "eval_accuracy": 0.9218,
      "eval_loss": 0.34348437190055847,
      "eval_runtime": 30.0429,
      "eval_samples_per_second": 332.857,
      "eval_steps_per_second": 1.331,
      "step": 66933
    },
    {
      "epoch": 201.02,
      "grad_norm": 5.585843563079834,
      "learning_rate": 3.2992992992992995e-06,
      "loss": 0.1795,
      "step": 66940
    },
    {
      "epoch": 201.05,
      "grad_norm": 9.8064603805542,
      "learning_rate": 3.2982982982982985e-06,
      "loss": 0.205,
      "step": 66950
    },
    {
      "epoch": 201.08,
      "grad_norm": 14.857874870300293,
      "learning_rate": 3.2972972972972976e-06,
      "loss": 0.2063,
      "step": 66960
    },
    {
      "epoch": 201.11,
      "grad_norm": 9.070259094238281,
      "learning_rate": 3.296296296296296e-06,
      "loss": 0.206,
      "step": 66970
    },
    {
      "epoch": 201.14,
      "grad_norm": 8.166922569274902,
      "learning_rate": 3.2952952952952956e-06,
      "loss": 0.2353,
      "step": 66980
    },
    {
      "epoch": 201.17,
      "grad_norm": 8.346697807312012,
      "learning_rate": 3.2942942942942946e-06,
      "loss": 0.2073,
      "step": 66990
    },
    {
      "epoch": 201.2,
      "grad_norm": 6.806495189666748,
      "learning_rate": 3.2932932932932936e-06,
      "loss": 0.2024,
      "step": 67000
    },
    {
      "epoch": 201.23,
      "grad_norm": 5.175416946411133,
      "learning_rate": 3.2922922922922922e-06,
      "loss": 0.1944,
      "step": 67010
    },
    {
      "epoch": 201.26,
      "grad_norm": 4.9541425704956055,
      "learning_rate": 3.2912912912912912e-06,
      "loss": 0.1983,
      "step": 67020
    },
    {
      "epoch": 201.29,
      "grad_norm": 10.685269355773926,
      "learning_rate": 3.2902902902902907e-06,
      "loss": 0.2429,
      "step": 67030
    },
    {
      "epoch": 201.32,
      "grad_norm": 3.89286470413208,
      "learning_rate": 3.2892892892892897e-06,
      "loss": 0.1864,
      "step": 67040
    },
    {
      "epoch": 201.35,
      "grad_norm": 24.30112075805664,
      "learning_rate": 3.2882882882882887e-06,
      "loss": 0.2073,
      "step": 67050
    },
    {
      "epoch": 201.38,
      "grad_norm": 10.195809364318848,
      "learning_rate": 3.2872872872872873e-06,
      "loss": 0.1762,
      "step": 67060
    },
    {
      "epoch": 201.41,
      "grad_norm": 4.417555809020996,
      "learning_rate": 3.2862862862862863e-06,
      "loss": 0.1725,
      "step": 67070
    },
    {
      "epoch": 201.44,
      "grad_norm": 9.258975982666016,
      "learning_rate": 3.2852852852852858e-06,
      "loss": 0.1775,
      "step": 67080
    },
    {
      "epoch": 201.47,
      "grad_norm": 7.1014909744262695,
      "learning_rate": 3.284284284284285e-06,
      "loss": 0.2069,
      "step": 67090
    },
    {
      "epoch": 201.5,
      "grad_norm": 8.235123634338379,
      "learning_rate": 3.2832832832832834e-06,
      "loss": 0.2155,
      "step": 67100
    },
    {
      "epoch": 201.53,
      "grad_norm": 10.371126174926758,
      "learning_rate": 3.2822822822822824e-06,
      "loss": 0.2042,
      "step": 67110
    },
    {
      "epoch": 201.56,
      "grad_norm": 7.004726409912109,
      "learning_rate": 3.281281281281282e-06,
      "loss": 0.1763,
      "step": 67120
    },
    {
      "epoch": 201.59,
      "grad_norm": 10.16481876373291,
      "learning_rate": 3.280280280280281e-06,
      "loss": 0.1957,
      "step": 67130
    },
    {
      "epoch": 201.62,
      "grad_norm": 7.993920803070068,
      "learning_rate": 3.2792792792792795e-06,
      "loss": 0.201,
      "step": 67140
    },
    {
      "epoch": 201.65,
      "grad_norm": 8.352984428405762,
      "learning_rate": 3.2782782782782785e-06,
      "loss": 0.2305,
      "step": 67150
    },
    {
      "epoch": 201.68,
      "grad_norm": 9.135019302368164,
      "learning_rate": 3.2772772772772775e-06,
      "loss": 0.1769,
      "step": 67160
    },
    {
      "epoch": 201.71,
      "grad_norm": 8.137670516967773,
      "learning_rate": 3.276276276276277e-06,
      "loss": 0.2009,
      "step": 67170
    },
    {
      "epoch": 201.74,
      "grad_norm": 9.089133262634277,
      "learning_rate": 3.2752752752752755e-06,
      "loss": 0.2282,
      "step": 67180
    },
    {
      "epoch": 201.77,
      "grad_norm": 23.728456497192383,
      "learning_rate": 3.2742742742742746e-06,
      "loss": 0.241,
      "step": 67190
    },
    {
      "epoch": 201.8,
      "grad_norm": 13.11851692199707,
      "learning_rate": 3.2732732732732736e-06,
      "loss": 0.2389,
      "step": 67200
    },
    {
      "epoch": 201.83,
      "grad_norm": 6.347787380218506,
      "learning_rate": 3.2722722722722726e-06,
      "loss": 0.2113,
      "step": 67210
    },
    {
      "epoch": 201.86,
      "grad_norm": 6.735030651092529,
      "learning_rate": 3.271271271271271e-06,
      "loss": 0.186,
      "step": 67220
    },
    {
      "epoch": 201.89,
      "grad_norm": 10.286836624145508,
      "learning_rate": 3.2702702702702706e-06,
      "loss": 0.2022,
      "step": 67230
    },
    {
      "epoch": 201.92,
      "grad_norm": 11.07098388671875,
      "learning_rate": 3.2692692692692696e-06,
      "loss": 0.2172,
      "step": 67240
    },
    {
      "epoch": 201.95,
      "grad_norm": 32.070674896240234,
      "learning_rate": 3.2682682682682687e-06,
      "loss": 0.2491,
      "step": 67250
    },
    {
      "epoch": 201.98,
      "grad_norm": 5.501422882080078,
      "learning_rate": 3.2672672672672673e-06,
      "loss": 0.2002,
      "step": 67260
    },
    {
      "epoch": 202.0,
      "eval_accuracy": 0.9219,
      "eval_loss": 0.3429811894893646,
      "eval_runtime": 30.0061,
      "eval_samples_per_second": 333.265,
      "eval_steps_per_second": 1.333,
      "step": 67266
    },
    {
      "epoch": 202.01,
      "grad_norm": 8.718255996704102,
      "learning_rate": 3.2662662662662663e-06,
      "loss": 0.2876,
      "step": 67270
    },
    {
      "epoch": 202.04,
      "grad_norm": 10.62755012512207,
      "learning_rate": 3.2652652652652657e-06,
      "loss": 0.2276,
      "step": 67280
    },
    {
      "epoch": 202.07,
      "grad_norm": 7.26613712310791,
      "learning_rate": 3.2642642642642647e-06,
      "loss": 0.1601,
      "step": 67290
    },
    {
      "epoch": 202.1,
      "grad_norm": 5.3975324630737305,
      "learning_rate": 3.2632632632632633e-06,
      "loss": 0.2221,
      "step": 67300
    },
    {
      "epoch": 202.13,
      "grad_norm": 10.205594062805176,
      "learning_rate": 3.2622622622622624e-06,
      "loss": 0.1769,
      "step": 67310
    },
    {
      "epoch": 202.16,
      "grad_norm": 12.495868682861328,
      "learning_rate": 3.2612612612612614e-06,
      "loss": 0.1723,
      "step": 67320
    },
    {
      "epoch": 202.19,
      "grad_norm": 6.708544731140137,
      "learning_rate": 3.260260260260261e-06,
      "loss": 0.2758,
      "step": 67330
    },
    {
      "epoch": 202.22,
      "grad_norm": 73.14022827148438,
      "learning_rate": 3.25925925925926e-06,
      "loss": 0.1793,
      "step": 67340
    },
    {
      "epoch": 202.25,
      "grad_norm": 10.164395332336426,
      "learning_rate": 3.2582582582582584e-06,
      "loss": 0.2091,
      "step": 67350
    },
    {
      "epoch": 202.28,
      "grad_norm": 6.603264331817627,
      "learning_rate": 3.2572572572572574e-06,
      "loss": 0.1974,
      "step": 67360
    },
    {
      "epoch": 202.31,
      "grad_norm": 5.238640308380127,
      "learning_rate": 3.2562562562562565e-06,
      "loss": 0.2016,
      "step": 67370
    },
    {
      "epoch": 202.34,
      "grad_norm": 8.296351432800293,
      "learning_rate": 3.255255255255256e-06,
      "loss": 0.2126,
      "step": 67380
    },
    {
      "epoch": 202.37,
      "grad_norm": 5.110581398010254,
      "learning_rate": 3.2542542542542545e-06,
      "loss": 0.1699,
      "step": 67390
    },
    {
      "epoch": 202.4,
      "grad_norm": 11.183879852294922,
      "learning_rate": 3.2532532532532535e-06,
      "loss": 0.1769,
      "step": 67400
    },
    {
      "epoch": 202.43,
      "grad_norm": 13.261514663696289,
      "learning_rate": 3.2522522522522525e-06,
      "loss": 0.1974,
      "step": 67410
    },
    {
      "epoch": 202.46,
      "grad_norm": 5.287303447723389,
      "learning_rate": 3.2512512512512516e-06,
      "loss": 0.2223,
      "step": 67420
    },
    {
      "epoch": 202.49,
      "grad_norm": 19.359033584594727,
      "learning_rate": 3.25025025025025e-06,
      "loss": 0.197,
      "step": 67430
    },
    {
      "epoch": 202.52,
      "grad_norm": 10.110488891601562,
      "learning_rate": 3.2492492492492496e-06,
      "loss": 0.1859,
      "step": 67440
    },
    {
      "epoch": 202.55,
      "grad_norm": 16.38164710998535,
      "learning_rate": 3.2482482482482486e-06,
      "loss": 0.1763,
      "step": 67450
    },
    {
      "epoch": 202.58,
      "grad_norm": 12.124810218811035,
      "learning_rate": 3.2472472472472476e-06,
      "loss": 0.1706,
      "step": 67460
    },
    {
      "epoch": 202.61,
      "grad_norm": 5.4137091636657715,
      "learning_rate": 3.2462462462462462e-06,
      "loss": 0.1884,
      "step": 67470
    },
    {
      "epoch": 202.64,
      "grad_norm": 2.667762041091919,
      "learning_rate": 3.2452452452452452e-06,
      "loss": 0.2325,
      "step": 67480
    },
    {
      "epoch": 202.67,
      "grad_norm": 9.29549789428711,
      "learning_rate": 3.2442442442442447e-06,
      "loss": 0.1986,
      "step": 67490
    },
    {
      "epoch": 202.7,
      "grad_norm": 10.691791534423828,
      "learning_rate": 3.2432432432432437e-06,
      "loss": 0.1924,
      "step": 67500
    },
    {
      "epoch": 202.73,
      "grad_norm": 7.661292552947998,
      "learning_rate": 3.2422422422422423e-06,
      "loss": 0.2108,
      "step": 67510
    },
    {
      "epoch": 202.76,
      "grad_norm": 10.93579387664795,
      "learning_rate": 3.2412412412412413e-06,
      "loss": 0.2389,
      "step": 67520
    },
    {
      "epoch": 202.79,
      "grad_norm": 8.473921775817871,
      "learning_rate": 3.2402402402402403e-06,
      "loss": 0.2346,
      "step": 67530
    },
    {
      "epoch": 202.82,
      "grad_norm": 7.481964111328125,
      "learning_rate": 3.2392392392392398e-06,
      "loss": 0.191,
      "step": 67540
    },
    {
      "epoch": 202.85,
      "grad_norm": 9.03524398803711,
      "learning_rate": 3.2382382382382384e-06,
      "loss": 0.2613,
      "step": 67550
    },
    {
      "epoch": 202.88,
      "grad_norm": 11.734621047973633,
      "learning_rate": 3.2372372372372374e-06,
      "loss": 0.2266,
      "step": 67560
    },
    {
      "epoch": 202.91,
      "grad_norm": 9.509786605834961,
      "learning_rate": 3.2362362362362364e-06,
      "loss": 0.2251,
      "step": 67570
    },
    {
      "epoch": 202.94,
      "grad_norm": 9.240423202514648,
      "learning_rate": 3.235235235235236e-06,
      "loss": 0.1949,
      "step": 67580
    },
    {
      "epoch": 202.97,
      "grad_norm": 7.965895652770996,
      "learning_rate": 3.234234234234235e-06,
      "loss": 0.2062,
      "step": 67590
    },
    {
      "epoch": 203.0,
      "eval_accuracy": 0.921,
      "eval_loss": 0.3434547781944275,
      "eval_runtime": 30.9552,
      "eval_samples_per_second": 323.047,
      "eval_steps_per_second": 1.292,
      "step": 67599
    },
    {
      "epoch": 203.0,
      "grad_norm": 10.520033836364746,
      "learning_rate": 3.2332332332332335e-06,
      "loss": 0.3519,
      "step": 67600
    },
    {
      "epoch": 203.03,
      "grad_norm": 6.830593109130859,
      "learning_rate": 3.2322322322322325e-06,
      "loss": 0.2358,
      "step": 67610
    },
    {
      "epoch": 203.06,
      "grad_norm": 7.024171829223633,
      "learning_rate": 3.2312312312312315e-06,
      "loss": 0.1882,
      "step": 67620
    },
    {
      "epoch": 203.09,
      "grad_norm": 7.7735795974731445,
      "learning_rate": 3.230230230230231e-06,
      "loss": 0.1726,
      "step": 67630
    },
    {
      "epoch": 203.12,
      "grad_norm": 11.344868659973145,
      "learning_rate": 3.229229229229229e-06,
      "loss": 0.2038,
      "step": 67640
    },
    {
      "epoch": 203.15,
      "grad_norm": 8.693346977233887,
      "learning_rate": 3.2282282282282286e-06,
      "loss": 0.2087,
      "step": 67650
    },
    {
      "epoch": 203.18,
      "grad_norm": 5.3042802810668945,
      "learning_rate": 3.2272272272272276e-06,
      "loss": 0.1595,
      "step": 67660
    },
    {
      "epoch": 203.21,
      "grad_norm": 10.300958633422852,
      "learning_rate": 3.2262262262262266e-06,
      "loss": 0.2162,
      "step": 67670
    },
    {
      "epoch": 203.24,
      "grad_norm": 12.667335510253906,
      "learning_rate": 3.225225225225225e-06,
      "loss": 0.2021,
      "step": 67680
    },
    {
      "epoch": 203.27,
      "grad_norm": 12.178764343261719,
      "learning_rate": 3.2242242242242246e-06,
      "loss": 0.1913,
      "step": 67690
    },
    {
      "epoch": 203.3,
      "grad_norm": 11.787507057189941,
      "learning_rate": 3.2232232232232236e-06,
      "loss": 0.2109,
      "step": 67700
    },
    {
      "epoch": 203.33,
      "grad_norm": 7.852213382720947,
      "learning_rate": 3.2222222222222227e-06,
      "loss": 0.2229,
      "step": 67710
    },
    {
      "epoch": 203.36,
      "grad_norm": 5.095298767089844,
      "learning_rate": 3.2212212212212213e-06,
      "loss": 0.1835,
      "step": 67720
    },
    {
      "epoch": 203.39,
      "grad_norm": 8.623116493225098,
      "learning_rate": 3.2202202202202203e-06,
      "loss": 0.198,
      "step": 67730
    },
    {
      "epoch": 203.42,
      "grad_norm": 20.116989135742188,
      "learning_rate": 3.2192192192192197e-06,
      "loss": 0.1872,
      "step": 67740
    },
    {
      "epoch": 203.45,
      "grad_norm": 55.0098876953125,
      "learning_rate": 3.2182182182182187e-06,
      "loss": 0.2262,
      "step": 67750
    },
    {
      "epoch": 203.48,
      "grad_norm": 7.781737327575684,
      "learning_rate": 3.2172172172172173e-06,
      "loss": 0.1956,
      "step": 67760
    },
    {
      "epoch": 203.51,
      "grad_norm": 5.256515979766846,
      "learning_rate": 3.2162162162162164e-06,
      "loss": 0.1899,
      "step": 67770
    },
    {
      "epoch": 203.54,
      "grad_norm": 8.341472625732422,
      "learning_rate": 3.2152152152152154e-06,
      "loss": 0.1984,
      "step": 67780
    },
    {
      "epoch": 203.57,
      "grad_norm": 11.650410652160645,
      "learning_rate": 3.214214214214215e-06,
      "loss": 0.1975,
      "step": 67790
    },
    {
      "epoch": 203.6,
      "grad_norm": 13.861037254333496,
      "learning_rate": 3.2132132132132134e-06,
      "loss": 0.2416,
      "step": 67800
    },
    {
      "epoch": 203.63,
      "grad_norm": 5.498610019683838,
      "learning_rate": 3.2122122122122124e-06,
      "loss": 0.209,
      "step": 67810
    },
    {
      "epoch": 203.66,
      "grad_norm": 20.462078094482422,
      "learning_rate": 3.2112112112112114e-06,
      "loss": 0.2335,
      "step": 67820
    },
    {
      "epoch": 203.69,
      "grad_norm": 7.965959072113037,
      "learning_rate": 3.2102102102102105e-06,
      "loss": 0.2223,
      "step": 67830
    },
    {
      "epoch": 203.72,
      "grad_norm": 10.079707145690918,
      "learning_rate": 3.209209209209209e-06,
      "loss": 0.2148,
      "step": 67840
    },
    {
      "epoch": 203.75,
      "grad_norm": 12.665164947509766,
      "learning_rate": 3.2082082082082085e-06,
      "loss": 0.2182,
      "step": 67850
    },
    {
      "epoch": 203.78,
      "grad_norm": 7.6347808837890625,
      "learning_rate": 3.2072072072072075e-06,
      "loss": 0.2156,
      "step": 67860
    },
    {
      "epoch": 203.81,
      "grad_norm": 8.215681076049805,
      "learning_rate": 3.2062062062062065e-06,
      "loss": 0.1907,
      "step": 67870
    },
    {
      "epoch": 203.84,
      "grad_norm": 8.97498607635498,
      "learning_rate": 3.2052052052052056e-06,
      "loss": 0.22,
      "step": 67880
    },
    {
      "epoch": 203.87,
      "grad_norm": 8.686421394348145,
      "learning_rate": 3.204204204204204e-06,
      "loss": 0.2119,
      "step": 67890
    },
    {
      "epoch": 203.9,
      "grad_norm": 6.8936028480529785,
      "learning_rate": 3.2032032032032036e-06,
      "loss": 0.1732,
      "step": 67900
    },
    {
      "epoch": 203.93,
      "grad_norm": 9.670003890991211,
      "learning_rate": 3.2022022022022026e-06,
      "loss": 0.2004,
      "step": 67910
    },
    {
      "epoch": 203.96,
      "grad_norm": 8.618924140930176,
      "learning_rate": 3.2012012012012016e-06,
      "loss": 0.2247,
      "step": 67920
    },
    {
      "epoch": 203.99,
      "grad_norm": 8.400858879089355,
      "learning_rate": 3.2002002002002002e-06,
      "loss": 0.2138,
      "step": 67930
    },
    {
      "epoch": 204.0,
      "eval_accuracy": 0.9214,
      "eval_loss": 0.348578542470932,
      "eval_runtime": 30.1847,
      "eval_samples_per_second": 331.294,
      "eval_steps_per_second": 1.325,
      "step": 67932
    },
    {
      "epoch": 204.02,
      "grad_norm": 7.251631259918213,
      "learning_rate": 3.1991991991991992e-06,
      "loss": 0.1747,
      "step": 67940
    },
    {
      "epoch": 204.05,
      "grad_norm": 7.994751453399658,
      "learning_rate": 3.1981981981981987e-06,
      "loss": 0.2139,
      "step": 67950
    },
    {
      "epoch": 204.08,
      "grad_norm": 10.708035469055176,
      "learning_rate": 3.1971971971971977e-06,
      "loss": 0.2133,
      "step": 67960
    },
    {
      "epoch": 204.11,
      "grad_norm": 12.008417129516602,
      "learning_rate": 3.1961961961961963e-06,
      "loss": 0.2111,
      "step": 67970
    },
    {
      "epoch": 204.14,
      "grad_norm": 7.354037761688232,
      "learning_rate": 3.1951951951951953e-06,
      "loss": 0.1975,
      "step": 67980
    },
    {
      "epoch": 204.17,
      "grad_norm": 5.586855411529541,
      "learning_rate": 3.1941941941941943e-06,
      "loss": 0.2456,
      "step": 67990
    },
    {
      "epoch": 204.2,
      "grad_norm": 6.365147113800049,
      "learning_rate": 3.1931931931931938e-06,
      "loss": 0.221,
      "step": 68000
    },
    {
      "epoch": 204.23,
      "grad_norm": 9.83021354675293,
      "learning_rate": 3.1921921921921924e-06,
      "loss": 0.1844,
      "step": 68010
    },
    {
      "epoch": 204.26,
      "grad_norm": 8.63125228881836,
      "learning_rate": 3.1911911911911914e-06,
      "loss": 0.2273,
      "step": 68020
    },
    {
      "epoch": 204.29,
      "grad_norm": 4.584152698516846,
      "learning_rate": 3.1901901901901904e-06,
      "loss": 0.1928,
      "step": 68030
    },
    {
      "epoch": 204.32,
      "grad_norm": 7.843583583831787,
      "learning_rate": 3.1891891891891894e-06,
      "loss": 0.2027,
      "step": 68040
    },
    {
      "epoch": 204.35,
      "grad_norm": 7.518090724945068,
      "learning_rate": 3.188188188188188e-06,
      "loss": 0.2321,
      "step": 68050
    },
    {
      "epoch": 204.38,
      "grad_norm": 15.306051254272461,
      "learning_rate": 3.1871871871871875e-06,
      "loss": 0.2259,
      "step": 68060
    },
    {
      "epoch": 204.41,
      "grad_norm": 9.591769218444824,
      "learning_rate": 3.1861861861861865e-06,
      "loss": 0.1874,
      "step": 68070
    },
    {
      "epoch": 204.44,
      "grad_norm": 13.327679634094238,
      "learning_rate": 3.1851851851851855e-06,
      "loss": 0.201,
      "step": 68080
    },
    {
      "epoch": 204.47,
      "grad_norm": 9.915436744689941,
      "learning_rate": 3.184184184184184e-06,
      "loss": 0.2284,
      "step": 68090
    },
    {
      "epoch": 204.5,
      "grad_norm": 8.393474578857422,
      "learning_rate": 3.183183183183183e-06,
      "loss": 0.1653,
      "step": 68100
    },
    {
      "epoch": 204.53,
      "grad_norm": 8.712867736816406,
      "learning_rate": 3.1821821821821826e-06,
      "loss": 0.1695,
      "step": 68110
    },
    {
      "epoch": 204.56,
      "grad_norm": 9.78408432006836,
      "learning_rate": 3.1811811811811816e-06,
      "loss": 0.2103,
      "step": 68120
    },
    {
      "epoch": 204.59,
      "grad_norm": 10.859975814819336,
      "learning_rate": 3.1801801801801806e-06,
      "loss": 0.2421,
      "step": 68130
    },
    {
      "epoch": 204.62,
      "grad_norm": 9.492353439331055,
      "learning_rate": 3.179179179179179e-06,
      "loss": 0.2112,
      "step": 68140
    },
    {
      "epoch": 204.65,
      "grad_norm": 8.4496431350708,
      "learning_rate": 3.1781781781781786e-06,
      "loss": 0.2326,
      "step": 68150
    },
    {
      "epoch": 204.68,
      "grad_norm": 5.3056511878967285,
      "learning_rate": 3.1771771771771776e-06,
      "loss": 0.1863,
      "step": 68160
    },
    {
      "epoch": 204.71,
      "grad_norm": 8.66458511352539,
      "learning_rate": 3.1761761761761767e-06,
      "loss": 0.2263,
      "step": 68170
    },
    {
      "epoch": 204.74,
      "grad_norm": 12.74610424041748,
      "learning_rate": 3.1751751751751753e-06,
      "loss": 0.19,
      "step": 68180
    },
    {
      "epoch": 204.77,
      "grad_norm": 9.532842636108398,
      "learning_rate": 3.1741741741741743e-06,
      "loss": 0.1902,
      "step": 68190
    },
    {
      "epoch": 204.8,
      "grad_norm": 8.059785842895508,
      "learning_rate": 3.1731731731731737e-06,
      "loss": 0.1739,
      "step": 68200
    },
    {
      "epoch": 204.83,
      "grad_norm": 7.7359724044799805,
      "learning_rate": 3.1721721721721727e-06,
      "loss": 0.215,
      "step": 68210
    },
    {
      "epoch": 204.86,
      "grad_norm": 9.996238708496094,
      "learning_rate": 3.1711711711711713e-06,
      "loss": 0.2487,
      "step": 68220
    },
    {
      "epoch": 204.89,
      "grad_norm": 9.121658325195312,
      "learning_rate": 3.1701701701701703e-06,
      "loss": 0.2669,
      "step": 68230
    },
    {
      "epoch": 204.92,
      "grad_norm": 9.57058334350586,
      "learning_rate": 3.1691691691691694e-06,
      "loss": 0.1695,
      "step": 68240
    },
    {
      "epoch": 204.95,
      "grad_norm": 8.241745948791504,
      "learning_rate": 3.168168168168169e-06,
      "loss": 0.171,
      "step": 68250
    },
    {
      "epoch": 204.98,
      "grad_norm": 16.624746322631836,
      "learning_rate": 3.1671671671671674e-06,
      "loss": 0.1742,
      "step": 68260
    },
    {
      "epoch": 205.0,
      "eval_accuracy": 0.9224,
      "eval_loss": 0.3453855812549591,
      "eval_runtime": 30.1946,
      "eval_samples_per_second": 331.185,
      "eval_steps_per_second": 1.325,
      "step": 68265
    },
    {
      "epoch": 205.02,
      "grad_norm": 9.602413177490234,
      "learning_rate": 3.1661661661661664e-06,
      "loss": 0.2467,
      "step": 68270
    },
    {
      "epoch": 205.05,
      "grad_norm": 7.113770008087158,
      "learning_rate": 3.1651651651651654e-06,
      "loss": 0.1756,
      "step": 68280
    },
    {
      "epoch": 205.08,
      "grad_norm": 9.36727523803711,
      "learning_rate": 3.1641641641641645e-06,
      "loss": 0.2468,
      "step": 68290
    },
    {
      "epoch": 205.11,
      "grad_norm": 7.920475959777832,
      "learning_rate": 3.163163163163163e-06,
      "loss": 0.2097,
      "step": 68300
    },
    {
      "epoch": 205.14,
      "grad_norm": 17.55813217163086,
      "learning_rate": 3.1621621621621625e-06,
      "loss": 0.2068,
      "step": 68310
    },
    {
      "epoch": 205.17,
      "grad_norm": 8.1986665725708,
      "learning_rate": 3.1611611611611615e-06,
      "loss": 0.2084,
      "step": 68320
    },
    {
      "epoch": 205.2,
      "grad_norm": 7.943564414978027,
      "learning_rate": 3.1601601601601605e-06,
      "loss": 0.1825,
      "step": 68330
    },
    {
      "epoch": 205.23,
      "grad_norm": 6.722221374511719,
      "learning_rate": 3.159159159159159e-06,
      "loss": 0.2263,
      "step": 68340
    },
    {
      "epoch": 205.26,
      "grad_norm": 5.99729585647583,
      "learning_rate": 3.158158158158158e-06,
      "loss": 0.2198,
      "step": 68350
    },
    {
      "epoch": 205.29,
      "grad_norm": 7.836442947387695,
      "learning_rate": 3.1571571571571576e-06,
      "loss": 0.2033,
      "step": 68360
    },
    {
      "epoch": 205.32,
      "grad_norm": 9.03475284576416,
      "learning_rate": 3.1561561561561566e-06,
      "loss": 0.226,
      "step": 68370
    },
    {
      "epoch": 205.35,
      "grad_norm": 7.008472919464111,
      "learning_rate": 3.155155155155155e-06,
      "loss": 0.2043,
      "step": 68380
    },
    {
      "epoch": 205.38,
      "grad_norm": 7.130331039428711,
      "learning_rate": 3.1541541541541542e-06,
      "loss": 0.2091,
      "step": 68390
    },
    {
      "epoch": 205.41,
      "grad_norm": 5.682414531707764,
      "learning_rate": 3.1531531531531532e-06,
      "loss": 0.1976,
      "step": 68400
    },
    {
      "epoch": 205.44,
      "grad_norm": 10.798087120056152,
      "learning_rate": 3.1521521521521527e-06,
      "loss": 0.2045,
      "step": 68410
    },
    {
      "epoch": 205.47,
      "grad_norm": 7.6159257888793945,
      "learning_rate": 3.1511511511511517e-06,
      "loss": 0.1903,
      "step": 68420
    },
    {
      "epoch": 205.5,
      "grad_norm": 44.628604888916016,
      "learning_rate": 3.1501501501501503e-06,
      "loss": 0.2026,
      "step": 68430
    },
    {
      "epoch": 205.53,
      "grad_norm": 5.834968566894531,
      "learning_rate": 3.1491491491491493e-06,
      "loss": 0.2134,
      "step": 68440
    },
    {
      "epoch": 205.56,
      "grad_norm": 14.142244338989258,
      "learning_rate": 3.1481481481481483e-06,
      "loss": 0.1683,
      "step": 68450
    },
    {
      "epoch": 205.59,
      "grad_norm": 7.717212677001953,
      "learning_rate": 3.1471471471471478e-06,
      "loss": 0.2134,
      "step": 68460
    },
    {
      "epoch": 205.62,
      "grad_norm": 7.716274261474609,
      "learning_rate": 3.1461461461461464e-06,
      "loss": 0.2141,
      "step": 68470
    },
    {
      "epoch": 205.65,
      "grad_norm": 6.664702415466309,
      "learning_rate": 3.1451451451451454e-06,
      "loss": 0.2084,
      "step": 68480
    },
    {
      "epoch": 205.68,
      "grad_norm": 8.9114990234375,
      "learning_rate": 3.1441441441441444e-06,
      "loss": 0.1927,
      "step": 68490
    },
    {
      "epoch": 205.71,
      "grad_norm": 7.253476142883301,
      "learning_rate": 3.1431431431431434e-06,
      "loss": 0.2279,
      "step": 68500
    },
    {
      "epoch": 205.74,
      "grad_norm": 6.188289165496826,
      "learning_rate": 3.142142142142142e-06,
      "loss": 0.1908,
      "step": 68510
    },
    {
      "epoch": 205.77,
      "grad_norm": 7.8627519607543945,
      "learning_rate": 3.1411411411411415e-06,
      "loss": 0.2141,
      "step": 68520
    },
    {
      "epoch": 205.8,
      "grad_norm": 6.97731351852417,
      "learning_rate": 3.1401401401401405e-06,
      "loss": 0.2232,
      "step": 68530
    },
    {
      "epoch": 205.83,
      "grad_norm": 9.98652458190918,
      "learning_rate": 3.1391391391391395e-06,
      "loss": 0.1902,
      "step": 68540
    },
    {
      "epoch": 205.86,
      "grad_norm": 9.972131729125977,
      "learning_rate": 3.138138138138138e-06,
      "loss": 0.1951,
      "step": 68550
    },
    {
      "epoch": 205.89,
      "grad_norm": 11.02843189239502,
      "learning_rate": 3.137137137137137e-06,
      "loss": 0.2157,
      "step": 68560
    },
    {
      "epoch": 205.92,
      "grad_norm": 6.5551910400390625,
      "learning_rate": 3.1361361361361365e-06,
      "loss": 0.2084,
      "step": 68570
    },
    {
      "epoch": 205.95,
      "grad_norm": 6.231637954711914,
      "learning_rate": 3.1351351351351356e-06,
      "loss": 0.2005,
      "step": 68580
    },
    {
      "epoch": 205.98,
      "grad_norm": 10.181565284729004,
      "learning_rate": 3.134134134134134e-06,
      "loss": 0.2116,
      "step": 68590
    },
    {
      "epoch": 206.0,
      "eval_accuracy": 0.9212,
      "eval_loss": 0.34586644172668457,
      "eval_runtime": 30.3177,
      "eval_samples_per_second": 329.84,
      "eval_steps_per_second": 1.319,
      "step": 68598
    },
    {
      "epoch": 206.01,
      "grad_norm": 9.235222816467285,
      "learning_rate": 3.133133133133133e-06,
      "loss": 0.2663,
      "step": 68600
    },
    {
      "epoch": 206.04,
      "grad_norm": 6.275533676147461,
      "learning_rate": 3.132132132132132e-06,
      "loss": 0.1769,
      "step": 68610
    },
    {
      "epoch": 206.07,
      "grad_norm": 11.625263214111328,
      "learning_rate": 3.1311311311311316e-06,
      "loss": 0.2152,
      "step": 68620
    },
    {
      "epoch": 206.1,
      "grad_norm": 8.174838066101074,
      "learning_rate": 3.1301301301301302e-06,
      "loss": 0.2069,
      "step": 68630
    },
    {
      "epoch": 206.13,
      "grad_norm": 11.70566177368164,
      "learning_rate": 3.1291291291291293e-06,
      "loss": 0.1938,
      "step": 68640
    },
    {
      "epoch": 206.16,
      "grad_norm": 7.651664733886719,
      "learning_rate": 3.1281281281281283e-06,
      "loss": 0.2082,
      "step": 68650
    },
    {
      "epoch": 206.19,
      "grad_norm": 8.114742279052734,
      "learning_rate": 3.1271271271271277e-06,
      "loss": 0.2333,
      "step": 68660
    },
    {
      "epoch": 206.22,
      "grad_norm": 3.929366111755371,
      "learning_rate": 3.1261261261261267e-06,
      "loss": 0.1761,
      "step": 68670
    },
    {
      "epoch": 206.25,
      "grad_norm": 13.208675384521484,
      "learning_rate": 3.1251251251251253e-06,
      "loss": 0.2087,
      "step": 68680
    },
    {
      "epoch": 206.28,
      "grad_norm": 11.83393669128418,
      "learning_rate": 3.1241241241241243e-06,
      "loss": 0.1992,
      "step": 68690
    },
    {
      "epoch": 206.31,
      "grad_norm": 13.24660587310791,
      "learning_rate": 3.1231231231231234e-06,
      "loss": 0.2033,
      "step": 68700
    },
    {
      "epoch": 206.34,
      "grad_norm": 9.719695091247559,
      "learning_rate": 3.122122122122123e-06,
      "loss": 0.2095,
      "step": 68710
    },
    {
      "epoch": 206.37,
      "grad_norm": 10.940682411193848,
      "learning_rate": 3.1211211211211214e-06,
      "loss": 0.1689,
      "step": 68720
    },
    {
      "epoch": 206.4,
      "grad_norm": 14.1677827835083,
      "learning_rate": 3.1201201201201204e-06,
      "loss": 0.2098,
      "step": 68730
    },
    {
      "epoch": 206.43,
      "grad_norm": 15.953267097473145,
      "learning_rate": 3.1191191191191194e-06,
      "loss": 0.2153,
      "step": 68740
    },
    {
      "epoch": 206.46,
      "grad_norm": 8.892033576965332,
      "learning_rate": 3.1181181181181185e-06,
      "loss": 0.201,
      "step": 68750
    },
    {
      "epoch": 206.49,
      "grad_norm": 9.107186317443848,
      "learning_rate": 3.117117117117117e-06,
      "loss": 0.2101,
      "step": 68760
    },
    {
      "epoch": 206.52,
      "grad_norm": 9.353983879089355,
      "learning_rate": 3.1161161161161165e-06,
      "loss": 0.2265,
      "step": 68770
    },
    {
      "epoch": 206.55,
      "grad_norm": 5.846837520599365,
      "learning_rate": 3.1151151151151155e-06,
      "loss": 0.2063,
      "step": 68780
    },
    {
      "epoch": 206.58,
      "grad_norm": 11.786069869995117,
      "learning_rate": 3.1141141141141145e-06,
      "loss": 0.2005,
      "step": 68790
    },
    {
      "epoch": 206.61,
      "grad_norm": 9.795258522033691,
      "learning_rate": 3.113113113113113e-06,
      "loss": 0.2046,
      "step": 68800
    },
    {
      "epoch": 206.64,
      "grad_norm": 29.182819366455078,
      "learning_rate": 3.112112112112112e-06,
      "loss": 0.2002,
      "step": 68810
    },
    {
      "epoch": 206.67,
      "grad_norm": 7.303112030029297,
      "learning_rate": 3.1111111111111116e-06,
      "loss": 0.2126,
      "step": 68820
    },
    {
      "epoch": 206.7,
      "grad_norm": 11.245390892028809,
      "learning_rate": 3.1101101101101106e-06,
      "loss": 0.206,
      "step": 68830
    },
    {
      "epoch": 206.73,
      "grad_norm": 10.535725593566895,
      "learning_rate": 3.109109109109109e-06,
      "loss": 0.2446,
      "step": 68840
    },
    {
      "epoch": 206.76,
      "grad_norm": 10.453160285949707,
      "learning_rate": 3.1081081081081082e-06,
      "loss": 0.2332,
      "step": 68850
    },
    {
      "epoch": 206.79,
      "grad_norm": 21.370594024658203,
      "learning_rate": 3.1071071071071072e-06,
      "loss": 0.209,
      "step": 68860
    },
    {
      "epoch": 206.82,
      "grad_norm": 8.737125396728516,
      "learning_rate": 3.1061061061061067e-06,
      "loss": 0.1944,
      "step": 68870
    },
    {
      "epoch": 206.85,
      "grad_norm": 51.51033401489258,
      "learning_rate": 3.1051051051051053e-06,
      "loss": 0.175,
      "step": 68880
    },
    {
      "epoch": 206.88,
      "grad_norm": 5.918121814727783,
      "learning_rate": 3.1041041041041043e-06,
      "loss": 0.2061,
      "step": 68890
    },
    {
      "epoch": 206.91,
      "grad_norm": 6.032999038696289,
      "learning_rate": 3.1031031031031033e-06,
      "loss": 0.2145,
      "step": 68900
    },
    {
      "epoch": 206.94,
      "grad_norm": 8.406412124633789,
      "learning_rate": 3.1021021021021023e-06,
      "loss": 0.1783,
      "step": 68910
    },
    {
      "epoch": 206.97,
      "grad_norm": 6.8791656494140625,
      "learning_rate": 3.1011011011011018e-06,
      "loss": 0.2238,
      "step": 68920
    },
    {
      "epoch": 207.0,
      "grad_norm": 8.280485153198242,
      "learning_rate": 3.1001001001001004e-06,
      "loss": 0.1906,
      "step": 68930
    },
    {
      "epoch": 207.0,
      "eval_accuracy": 0.9202,
      "eval_loss": 0.3474779725074768,
      "eval_runtime": 30.3929,
      "eval_samples_per_second": 329.025,
      "eval_steps_per_second": 1.316,
      "step": 68931
    },
    {
      "epoch": 207.03,
      "grad_norm": 10.287189483642578,
      "learning_rate": 3.0990990990990994e-06,
      "loss": 0.2451,
      "step": 68940
    },
    {
      "epoch": 207.06,
      "grad_norm": 8.182289123535156,
      "learning_rate": 3.0980980980980984e-06,
      "loss": 0.1981,
      "step": 68950
    },
    {
      "epoch": 207.09,
      "grad_norm": 7.457107067108154,
      "learning_rate": 3.0970970970970974e-06,
      "loss": 0.229,
      "step": 68960
    },
    {
      "epoch": 207.12,
      "grad_norm": 8.606640815734863,
      "learning_rate": 3.096096096096096e-06,
      "loss": 0.2255,
      "step": 68970
    },
    {
      "epoch": 207.15,
      "grad_norm": 10.467392921447754,
      "learning_rate": 3.0950950950950955e-06,
      "loss": 0.2157,
      "step": 68980
    },
    {
      "epoch": 207.18,
      "grad_norm": 4.431227684020996,
      "learning_rate": 3.0940940940940945e-06,
      "loss": 0.2113,
      "step": 68990
    },
    {
      "epoch": 207.21,
      "grad_norm": 4.394383907318115,
      "learning_rate": 3.0930930930930935e-06,
      "loss": 0.2236,
      "step": 69000
    },
    {
      "epoch": 207.24,
      "grad_norm": 5.106213569641113,
      "learning_rate": 3.092092092092092e-06,
      "loss": 0.1776,
      "step": 69010
    },
    {
      "epoch": 207.27,
      "grad_norm": 10.5862455368042,
      "learning_rate": 3.091091091091091e-06,
      "loss": 0.2375,
      "step": 69020
    },
    {
      "epoch": 207.3,
      "grad_norm": 12.437689781188965,
      "learning_rate": 3.0900900900900905e-06,
      "loss": 0.2124,
      "step": 69030
    },
    {
      "epoch": 207.33,
      "grad_norm": 7.362451553344727,
      "learning_rate": 3.0890890890890896e-06,
      "loss": 0.1865,
      "step": 69040
    },
    {
      "epoch": 207.36,
      "grad_norm": 5.777377605438232,
      "learning_rate": 3.088088088088088e-06,
      "loss": 0.1866,
      "step": 69050
    },
    {
      "epoch": 207.39,
      "grad_norm": 9.140380859375,
      "learning_rate": 3.087087087087087e-06,
      "loss": 0.1874,
      "step": 69060
    },
    {
      "epoch": 207.42,
      "grad_norm": 5.589869022369385,
      "learning_rate": 3.086086086086086e-06,
      "loss": 0.188,
      "step": 69070
    },
    {
      "epoch": 207.45,
      "grad_norm": 8.92371654510498,
      "learning_rate": 3.0850850850850856e-06,
      "loss": 0.2202,
      "step": 69080
    },
    {
      "epoch": 207.48,
      "grad_norm": 8.389839172363281,
      "learning_rate": 3.0840840840840842e-06,
      "loss": 0.1988,
      "step": 69090
    },
    {
      "epoch": 207.51,
      "grad_norm": 9.888059616088867,
      "learning_rate": 3.0830830830830832e-06,
      "loss": 0.2054,
      "step": 69100
    },
    {
      "epoch": 207.54,
      "grad_norm": 12.49099349975586,
      "learning_rate": 3.0820820820820823e-06,
      "loss": 0.1985,
      "step": 69110
    },
    {
      "epoch": 207.57,
      "grad_norm": 6.890304088592529,
      "learning_rate": 3.0810810810810817e-06,
      "loss": 0.1996,
      "step": 69120
    },
    {
      "epoch": 207.6,
      "grad_norm": 6.663242816925049,
      "learning_rate": 3.08008008008008e-06,
      "loss": 0.1742,
      "step": 69130
    },
    {
      "epoch": 207.63,
      "grad_norm": 11.290828704833984,
      "learning_rate": 3.0790790790790793e-06,
      "loss": 0.2021,
      "step": 69140
    },
    {
      "epoch": 207.66,
      "grad_norm": 8.279558181762695,
      "learning_rate": 3.0780780780780783e-06,
      "loss": 0.2111,
      "step": 69150
    },
    {
      "epoch": 207.69,
      "grad_norm": 17.012474060058594,
      "learning_rate": 3.0770770770770774e-06,
      "loss": 0.2473,
      "step": 69160
    },
    {
      "epoch": 207.72,
      "grad_norm": 7.478008270263672,
      "learning_rate": 3.076076076076076e-06,
      "loss": 0.2468,
      "step": 69170
    },
    {
      "epoch": 207.75,
      "grad_norm": 7.471030235290527,
      "learning_rate": 3.0750750750750754e-06,
      "loss": 0.1856,
      "step": 69180
    },
    {
      "epoch": 207.78,
      "grad_norm": 8.613855361938477,
      "learning_rate": 3.0740740740740744e-06,
      "loss": 0.212,
      "step": 69190
    },
    {
      "epoch": 207.81,
      "grad_norm": 7.684292316436768,
      "learning_rate": 3.0730730730730734e-06,
      "loss": 0.1941,
      "step": 69200
    },
    {
      "epoch": 207.84,
      "grad_norm": 10.045713424682617,
      "learning_rate": 3.0720720720720725e-06,
      "loss": 0.1884,
      "step": 69210
    },
    {
      "epoch": 207.87,
      "grad_norm": 8.311060905456543,
      "learning_rate": 3.071071071071071e-06,
      "loss": 0.2113,
      "step": 69220
    },
    {
      "epoch": 207.9,
      "grad_norm": 9.759926795959473,
      "learning_rate": 3.0700700700700705e-06,
      "loss": 0.22,
      "step": 69230
    },
    {
      "epoch": 207.93,
      "grad_norm": 11.26575756072998,
      "learning_rate": 3.0690690690690695e-06,
      "loss": 0.2064,
      "step": 69240
    },
    {
      "epoch": 207.96,
      "grad_norm": 13.82739543914795,
      "learning_rate": 3.0680680680680685e-06,
      "loss": 0.2111,
      "step": 69250
    },
    {
      "epoch": 207.99,
      "grad_norm": 11.98332691192627,
      "learning_rate": 3.067067067067067e-06,
      "loss": 0.2253,
      "step": 69260
    },
    {
      "epoch": 208.0,
      "eval_accuracy": 0.9227,
      "eval_loss": 0.34467118978500366,
      "eval_runtime": 30.1381,
      "eval_samples_per_second": 331.805,
      "eval_steps_per_second": 1.327,
      "step": 69264
    },
    {
      "epoch": 208.02,
      "grad_norm": 8.039475440979004,
      "learning_rate": 3.066066066066066e-06,
      "loss": 0.2116,
      "step": 69270
    },
    {
      "epoch": 208.05,
      "grad_norm": 6.659489631652832,
      "learning_rate": 3.0650650650650656e-06,
      "loss": 0.1548,
      "step": 69280
    },
    {
      "epoch": 208.08,
      "grad_norm": 5.740638732910156,
      "learning_rate": 3.0640640640640646e-06,
      "loss": 0.2026,
      "step": 69290
    },
    {
      "epoch": 208.11,
      "grad_norm": 16.331350326538086,
      "learning_rate": 3.063063063063063e-06,
      "loss": 0.1746,
      "step": 69300
    },
    {
      "epoch": 208.14,
      "grad_norm": 7.077906131744385,
      "learning_rate": 3.062062062062062e-06,
      "loss": 0.2259,
      "step": 69310
    },
    {
      "epoch": 208.17,
      "grad_norm": 9.825617790222168,
      "learning_rate": 3.0610610610610612e-06,
      "loss": 0.177,
      "step": 69320
    },
    {
      "epoch": 208.2,
      "grad_norm": 24.363231658935547,
      "learning_rate": 3.0600600600600607e-06,
      "loss": 0.2099,
      "step": 69330
    },
    {
      "epoch": 208.23,
      "grad_norm": 12.086694717407227,
      "learning_rate": 3.0590590590590593e-06,
      "loss": 0.2342,
      "step": 69340
    },
    {
      "epoch": 208.26,
      "grad_norm": 6.557600975036621,
      "learning_rate": 3.0580580580580583e-06,
      "loss": 0.1922,
      "step": 69350
    },
    {
      "epoch": 208.29,
      "grad_norm": 8.346308708190918,
      "learning_rate": 3.0570570570570573e-06,
      "loss": 0.1869,
      "step": 69360
    },
    {
      "epoch": 208.32,
      "grad_norm": 25.540225982666016,
      "learning_rate": 3.0560560560560563e-06,
      "loss": 0.2072,
      "step": 69370
    },
    {
      "epoch": 208.35,
      "grad_norm": 7.531724452972412,
      "learning_rate": 3.055055055055055e-06,
      "loss": 0.2166,
      "step": 69380
    },
    {
      "epoch": 208.38,
      "grad_norm": 22.228979110717773,
      "learning_rate": 3.0540540540540544e-06,
      "loss": 0.2049,
      "step": 69390
    },
    {
      "epoch": 208.41,
      "grad_norm": 8.810423851013184,
      "learning_rate": 3.0530530530530534e-06,
      "loss": 0.2424,
      "step": 69400
    },
    {
      "epoch": 208.44,
      "grad_norm": 7.928361892700195,
      "learning_rate": 3.0520520520520524e-06,
      "loss": 0.1783,
      "step": 69410
    },
    {
      "epoch": 208.47,
      "grad_norm": 8.047750473022461,
      "learning_rate": 3.051051051051051e-06,
      "loss": 0.1773,
      "step": 69420
    },
    {
      "epoch": 208.5,
      "grad_norm": 11.566629409790039,
      "learning_rate": 3.05005005005005e-06,
      "loss": 0.1867,
      "step": 69430
    },
    {
      "epoch": 208.53,
      "grad_norm": 5.5922136306762695,
      "learning_rate": 3.0490490490490494e-06,
      "loss": 0.1682,
      "step": 69440
    },
    {
      "epoch": 208.56,
      "grad_norm": 10.460928916931152,
      "learning_rate": 3.0480480480480485e-06,
      "loss": 0.1975,
      "step": 69450
    },
    {
      "epoch": 208.59,
      "grad_norm": 17.044843673706055,
      "learning_rate": 3.0470470470470475e-06,
      "loss": 0.1986,
      "step": 69460
    },
    {
      "epoch": 208.62,
      "grad_norm": 9.960391998291016,
      "learning_rate": 3.046046046046046e-06,
      "loss": 0.2116,
      "step": 69470
    },
    {
      "epoch": 208.65,
      "grad_norm": 11.07605266571045,
      "learning_rate": 3.045045045045045e-06,
      "loss": 0.2318,
      "step": 69480
    },
    {
      "epoch": 208.68,
      "grad_norm": 12.375711441040039,
      "learning_rate": 3.0440440440440445e-06,
      "loss": 0.1967,
      "step": 69490
    },
    {
      "epoch": 208.71,
      "grad_norm": 12.97611141204834,
      "learning_rate": 3.0430430430430436e-06,
      "loss": 0.2048,
      "step": 69500
    },
    {
      "epoch": 208.74,
      "grad_norm": 15.956262588500977,
      "learning_rate": 3.042042042042042e-06,
      "loss": 0.2123,
      "step": 69510
    },
    {
      "epoch": 208.77,
      "grad_norm": 8.807023048400879,
      "learning_rate": 3.041041041041041e-06,
      "loss": 0.2012,
      "step": 69520
    },
    {
      "epoch": 208.8,
      "grad_norm": 11.12279987335205,
      "learning_rate": 3.04004004004004e-06,
      "loss": 0.1931,
      "step": 69530
    },
    {
      "epoch": 208.83,
      "grad_norm": 11.185975074768066,
      "learning_rate": 3.0390390390390396e-06,
      "loss": 0.1789,
      "step": 69540
    },
    {
      "epoch": 208.86,
      "grad_norm": 10.935491561889648,
      "learning_rate": 3.0380380380380382e-06,
      "loss": 0.1802,
      "step": 69550
    },
    {
      "epoch": 208.89,
      "grad_norm": 7.59798526763916,
      "learning_rate": 3.0370370370370372e-06,
      "loss": 0.1891,
      "step": 69560
    },
    {
      "epoch": 208.92,
      "grad_norm": 14.1504545211792,
      "learning_rate": 3.0360360360360363e-06,
      "loss": 0.2487,
      "step": 69570
    },
    {
      "epoch": 208.95,
      "grad_norm": 8.360812187194824,
      "learning_rate": 3.0350350350350353e-06,
      "loss": 0.1796,
      "step": 69580
    },
    {
      "epoch": 208.98,
      "grad_norm": 6.125962257385254,
      "learning_rate": 3.034034034034034e-06,
      "loss": 0.1944,
      "step": 69590
    },
    {
      "epoch": 209.0,
      "eval_accuracy": 0.9209,
      "eval_loss": 0.3476316034793854,
      "eval_runtime": 30.3671,
      "eval_samples_per_second": 329.304,
      "eval_steps_per_second": 1.317,
      "step": 69597
    },
    {
      "epoch": 209.01,
      "grad_norm": 7.430843830108643,
      "learning_rate": 3.0330330330330333e-06,
      "loss": 0.1593,
      "step": 69600
    },
    {
      "epoch": 209.04,
      "grad_norm": 8.702662467956543,
      "learning_rate": 3.0320320320320323e-06,
      "loss": 0.1755,
      "step": 69610
    },
    {
      "epoch": 209.07,
      "grad_norm": 10.767953872680664,
      "learning_rate": 3.0310310310310314e-06,
      "loss": 0.23,
      "step": 69620
    },
    {
      "epoch": 209.1,
      "grad_norm": 6.538099765777588,
      "learning_rate": 3.03003003003003e-06,
      "loss": 0.1945,
      "step": 69630
    },
    {
      "epoch": 209.13,
      "grad_norm": 5.935338497161865,
      "learning_rate": 3.029029029029029e-06,
      "loss": 0.184,
      "step": 69640
    },
    {
      "epoch": 209.16,
      "grad_norm": 9.932296752929688,
      "learning_rate": 3.0280280280280284e-06,
      "loss": 0.1871,
      "step": 69650
    },
    {
      "epoch": 209.19,
      "grad_norm": 13.118521690368652,
      "learning_rate": 3.0270270270270274e-06,
      "loss": 0.2397,
      "step": 69660
    },
    {
      "epoch": 209.22,
      "grad_norm": 8.255109786987305,
      "learning_rate": 3.026026026026026e-06,
      "loss": 0.1855,
      "step": 69670
    },
    {
      "epoch": 209.25,
      "grad_norm": 5.949951171875,
      "learning_rate": 3.025025025025025e-06,
      "loss": 0.2013,
      "step": 69680
    },
    {
      "epoch": 209.28,
      "grad_norm": 6.14366340637207,
      "learning_rate": 3.0240240240240245e-06,
      "loss": 0.208,
      "step": 69690
    },
    {
      "epoch": 209.31,
      "grad_norm": 9.711809158325195,
      "learning_rate": 3.0230230230230235e-06,
      "loss": 0.2066,
      "step": 69700
    },
    {
      "epoch": 209.34,
      "grad_norm": 12.036486625671387,
      "learning_rate": 3.022022022022022e-06,
      "loss": 0.2155,
      "step": 69710
    },
    {
      "epoch": 209.37,
      "grad_norm": 8.757012367248535,
      "learning_rate": 3.021021021021021e-06,
      "loss": 0.2054,
      "step": 69720
    },
    {
      "epoch": 209.4,
      "grad_norm": 8.647063255310059,
      "learning_rate": 3.02002002002002e-06,
      "loss": 0.2168,
      "step": 69730
    },
    {
      "epoch": 209.43,
      "grad_norm": 142.7971954345703,
      "learning_rate": 3.0190190190190196e-06,
      "loss": 0.2024,
      "step": 69740
    },
    {
      "epoch": 209.46,
      "grad_norm": 23.4234561920166,
      "learning_rate": 3.0180180180180186e-06,
      "loss": 0.1948,
      "step": 69750
    },
    {
      "epoch": 209.49,
      "grad_norm": 15.356817245483398,
      "learning_rate": 3.017017017017017e-06,
      "loss": 0.2377,
      "step": 69760
    },
    {
      "epoch": 209.52,
      "grad_norm": 6.864319801330566,
      "learning_rate": 3.016016016016016e-06,
      "loss": 0.2036,
      "step": 69770
    },
    {
      "epoch": 209.55,
      "grad_norm": 7.54796028137207,
      "learning_rate": 3.0150150150150152e-06,
      "loss": 0.1921,
      "step": 69780
    },
    {
      "epoch": 209.58,
      "grad_norm": 12.432341575622559,
      "learning_rate": 3.0140140140140147e-06,
      "loss": 0.1946,
      "step": 69790
    },
    {
      "epoch": 209.61,
      "grad_norm": 9.492752075195312,
      "learning_rate": 3.0130130130130133e-06,
      "loss": 0.1714,
      "step": 69800
    },
    {
      "epoch": 209.64,
      "grad_norm": 6.0655107498168945,
      "learning_rate": 3.0120120120120123e-06,
      "loss": 0.1525,
      "step": 69810
    },
    {
      "epoch": 209.67,
      "grad_norm": 78.12712860107422,
      "learning_rate": 3.0110110110110113e-06,
      "loss": 0.2146,
      "step": 69820
    },
    {
      "epoch": 209.7,
      "grad_norm": 11.135631561279297,
      "learning_rate": 3.0100100100100103e-06,
      "loss": 0.162,
      "step": 69830
    },
    {
      "epoch": 209.73,
      "grad_norm": 10.186637878417969,
      "learning_rate": 3.009009009009009e-06,
      "loss": 0.2146,
      "step": 69840
    },
    {
      "epoch": 209.76,
      "grad_norm": 8.121334075927734,
      "learning_rate": 3.0080080080080084e-06,
      "loss": 0.2145,
      "step": 69850
    },
    {
      "epoch": 209.79,
      "grad_norm": 5.9976959228515625,
      "learning_rate": 3.0070070070070074e-06,
      "loss": 0.1918,
      "step": 69860
    },
    {
      "epoch": 209.82,
      "grad_norm": 15.697314262390137,
      "learning_rate": 3.0060060060060064e-06,
      "loss": 0.2105,
      "step": 69870
    },
    {
      "epoch": 209.85,
      "grad_norm": 6.640625476837158,
      "learning_rate": 3.005005005005005e-06,
      "loss": 0.2059,
      "step": 69880
    },
    {
      "epoch": 209.88,
      "grad_norm": 13.738353729248047,
      "learning_rate": 3.004004004004004e-06,
      "loss": 0.1623,
      "step": 69890
    },
    {
      "epoch": 209.91,
      "grad_norm": 6.439333438873291,
      "learning_rate": 3.0030030030030034e-06,
      "loss": 0.1955,
      "step": 69900
    },
    {
      "epoch": 209.94,
      "grad_norm": 8.461752891540527,
      "learning_rate": 3.0020020020020025e-06,
      "loss": 0.2288,
      "step": 69910
    },
    {
      "epoch": 209.97,
      "grad_norm": 8.837239265441895,
      "learning_rate": 3.001001001001001e-06,
      "loss": 0.2261,
      "step": 69920
    },
    {
      "epoch": 210.0,
      "grad_norm": 4.344946384429932,
      "learning_rate": 3e-06,
      "loss": 0.2114,
      "step": 69930
    },
    {
      "epoch": 210.0,
      "eval_accuracy": 0.9211,
      "eval_loss": 0.34734922647476196,
      "eval_runtime": 29.7978,
      "eval_samples_per_second": 335.595,
      "eval_steps_per_second": 1.342,
      "step": 69930
    },
    {
      "epoch": 210.03,
      "grad_norm": 10.93993091583252,
      "learning_rate": 2.998998998998999e-06,
      "loss": 0.2043,
      "step": 69940
    },
    {
      "epoch": 210.06,
      "grad_norm": 19.829133987426758,
      "learning_rate": 2.9979979979979985e-06,
      "loss": 0.2448,
      "step": 69950
    },
    {
      "epoch": 210.09,
      "grad_norm": 5.756875514984131,
      "learning_rate": 2.996996996996997e-06,
      "loss": 0.1992,
      "step": 69960
    },
    {
      "epoch": 210.12,
      "grad_norm": 9.509207725524902,
      "learning_rate": 2.995995995995996e-06,
      "loss": 0.2257,
      "step": 69970
    },
    {
      "epoch": 210.15,
      "grad_norm": 10.189516067504883,
      "learning_rate": 2.994994994994995e-06,
      "loss": 0.2327,
      "step": 69980
    },
    {
      "epoch": 210.18,
      "grad_norm": 18.809640884399414,
      "learning_rate": 2.993993993993994e-06,
      "loss": 0.2115,
      "step": 69990
    },
    {
      "epoch": 210.21,
      "grad_norm": 6.157955646514893,
      "learning_rate": 2.9929929929929936e-06,
      "loss": 0.1923,
      "step": 70000
    },
    {
      "epoch": 210.24,
      "grad_norm": 6.132051467895508,
      "learning_rate": 2.9919919919919922e-06,
      "loss": 0.189,
      "step": 70010
    },
    {
      "epoch": 210.27,
      "grad_norm": 6.394039630889893,
      "learning_rate": 2.9909909909909912e-06,
      "loss": 0.1871,
      "step": 70020
    },
    {
      "epoch": 210.3,
      "grad_norm": 5.475592136383057,
      "learning_rate": 2.9899899899899903e-06,
      "loss": 0.2036,
      "step": 70030
    },
    {
      "epoch": 210.33,
      "grad_norm": 9.245697975158691,
      "learning_rate": 2.9889889889889893e-06,
      "loss": 0.2321,
      "step": 70040
    },
    {
      "epoch": 210.36,
      "grad_norm": 7.29138708114624,
      "learning_rate": 2.987987987987988e-06,
      "loss": 0.2261,
      "step": 70050
    },
    {
      "epoch": 210.39,
      "grad_norm": 10.879846572875977,
      "learning_rate": 2.9869869869869873e-06,
      "loss": 0.2199,
      "step": 70060
    },
    {
      "epoch": 210.42,
      "grad_norm": 14.651310920715332,
      "learning_rate": 2.9859859859859863e-06,
      "loss": 0.186,
      "step": 70070
    },
    {
      "epoch": 210.45,
      "grad_norm": 11.868749618530273,
      "learning_rate": 2.9849849849849854e-06,
      "loss": 0.201,
      "step": 70080
    },
    {
      "epoch": 210.48,
      "grad_norm": 6.608259201049805,
      "learning_rate": 2.983983983983984e-06,
      "loss": 0.2258,
      "step": 70090
    },
    {
      "epoch": 210.51,
      "grad_norm": 6.677424907684326,
      "learning_rate": 2.982982982982983e-06,
      "loss": 0.2181,
      "step": 70100
    },
    {
      "epoch": 210.54,
      "grad_norm": 6.744799613952637,
      "learning_rate": 2.9819819819819824e-06,
      "loss": 0.2333,
      "step": 70110
    },
    {
      "epoch": 210.57,
      "grad_norm": 11.171833992004395,
      "learning_rate": 2.9809809809809814e-06,
      "loss": 0.2484,
      "step": 70120
    },
    {
      "epoch": 210.6,
      "grad_norm": 6.993897438049316,
      "learning_rate": 2.97997997997998e-06,
      "loss": 0.2329,
      "step": 70130
    },
    {
      "epoch": 210.63,
      "grad_norm": 10.105130195617676,
      "learning_rate": 2.978978978978979e-06,
      "loss": 0.226,
      "step": 70140
    },
    {
      "epoch": 210.66,
      "grad_norm": 8.83067512512207,
      "learning_rate": 2.9779779779779785e-06,
      "loss": 0.1931,
      "step": 70150
    },
    {
      "epoch": 210.69,
      "grad_norm": 9.080909729003906,
      "learning_rate": 2.9769769769769775e-06,
      "loss": 0.2069,
      "step": 70160
    },
    {
      "epoch": 210.72,
      "grad_norm": 8.153776168823242,
      "learning_rate": 2.975975975975976e-06,
      "loss": 0.1909,
      "step": 70170
    },
    {
      "epoch": 210.75,
      "grad_norm": 10.623549461364746,
      "learning_rate": 2.974974974974975e-06,
      "loss": 0.2104,
      "step": 70180
    },
    {
      "epoch": 210.78,
      "grad_norm": 5.49294900894165,
      "learning_rate": 2.973973973973974e-06,
      "loss": 0.2152,
      "step": 70190
    },
    {
      "epoch": 210.81,
      "grad_norm": 16.034284591674805,
      "learning_rate": 2.9729729729729736e-06,
      "loss": 0.1848,
      "step": 70200
    },
    {
      "epoch": 210.84,
      "grad_norm": 7.7901129722595215,
      "learning_rate": 2.971971971971972e-06,
      "loss": 0.1757,
      "step": 70210
    },
    {
      "epoch": 210.87,
      "grad_norm": 4.1843180656433105,
      "learning_rate": 2.970970970970971e-06,
      "loss": 0.205,
      "step": 70220
    },
    {
      "epoch": 210.9,
      "grad_norm": 8.070104598999023,
      "learning_rate": 2.96996996996997e-06,
      "loss": 0.2059,
      "step": 70230
    },
    {
      "epoch": 210.93,
      "grad_norm": 6.9925103187561035,
      "learning_rate": 2.9689689689689692e-06,
      "loss": 0.2086,
      "step": 70240
    },
    {
      "epoch": 210.96,
      "grad_norm": 4.306607246398926,
      "learning_rate": 2.967967967967968e-06,
      "loss": 0.1736,
      "step": 70250
    },
    {
      "epoch": 210.99,
      "grad_norm": 9.622013092041016,
      "learning_rate": 2.9669669669669673e-06,
      "loss": 0.1959,
      "step": 70260
    },
    {
      "epoch": 211.0,
      "eval_accuracy": 0.9213,
      "eval_loss": 0.34627532958984375,
      "eval_runtime": 30.4734,
      "eval_samples_per_second": 328.155,
      "eval_steps_per_second": 1.313,
      "step": 70263
    },
    {
      "epoch": 211.02,
      "grad_norm": 14.039822578430176,
      "learning_rate": 2.9659659659659663e-06,
      "loss": 0.1677,
      "step": 70270
    },
    {
      "epoch": 211.05,
      "grad_norm": 22.424680709838867,
      "learning_rate": 2.9649649649649653e-06,
      "loss": 0.1678,
      "step": 70280
    },
    {
      "epoch": 211.08,
      "grad_norm": 10.206035614013672,
      "learning_rate": 2.9639639639639643e-06,
      "loss": 0.2247,
      "step": 70290
    },
    {
      "epoch": 211.11,
      "grad_norm": 7.583370208740234,
      "learning_rate": 2.962962962962963e-06,
      "loss": 0.2512,
      "step": 70300
    },
    {
      "epoch": 211.14,
      "grad_norm": 13.69864273071289,
      "learning_rate": 2.9619619619619623e-06,
      "loss": 0.2479,
      "step": 70310
    },
    {
      "epoch": 211.17,
      "grad_norm": 12.130060195922852,
      "learning_rate": 2.9609609609609614e-06,
      "loss": 0.1771,
      "step": 70320
    },
    {
      "epoch": 211.2,
      "grad_norm": 11.583346366882324,
      "learning_rate": 2.9599599599599604e-06,
      "loss": 0.2024,
      "step": 70330
    },
    {
      "epoch": 211.23,
      "grad_norm": 10.837929725646973,
      "learning_rate": 2.958958958958959e-06,
      "loss": 0.2136,
      "step": 70340
    },
    {
      "epoch": 211.26,
      "grad_norm": 6.780359745025635,
      "learning_rate": 2.957957957957958e-06,
      "loss": 0.2036,
      "step": 70350
    },
    {
      "epoch": 211.29,
      "grad_norm": 8.665677070617676,
      "learning_rate": 2.9569569569569574e-06,
      "loss": 0.2607,
      "step": 70360
    },
    {
      "epoch": 211.32,
      "grad_norm": 10.008548736572266,
      "learning_rate": 2.9559559559559565e-06,
      "loss": 0.187,
      "step": 70370
    },
    {
      "epoch": 211.35,
      "grad_norm": 13.72387409210205,
      "learning_rate": 2.954954954954955e-06,
      "loss": 0.1296,
      "step": 70380
    },
    {
      "epoch": 211.38,
      "grad_norm": 7.62874174118042,
      "learning_rate": 2.953953953953954e-06,
      "loss": 0.1964,
      "step": 70390
    },
    {
      "epoch": 211.41,
      "grad_norm": 12.509716987609863,
      "learning_rate": 2.952952952952953e-06,
      "loss": 0.1922,
      "step": 70400
    },
    {
      "epoch": 211.44,
      "grad_norm": 8.657240867614746,
      "learning_rate": 2.9519519519519525e-06,
      "loss": 0.2024,
      "step": 70410
    },
    {
      "epoch": 211.47,
      "grad_norm": 36.341064453125,
      "learning_rate": 2.950950950950951e-06,
      "loss": 0.2399,
      "step": 70420
    },
    {
      "epoch": 211.5,
      "grad_norm": 10.542356491088867,
      "learning_rate": 2.94994994994995e-06,
      "loss": 0.1806,
      "step": 70430
    },
    {
      "epoch": 211.53,
      "grad_norm": 5.965512275695801,
      "learning_rate": 2.948948948948949e-06,
      "loss": 0.1963,
      "step": 70440
    },
    {
      "epoch": 211.56,
      "grad_norm": 9.04621410369873,
      "learning_rate": 2.947947947947948e-06,
      "loss": 0.2256,
      "step": 70450
    },
    {
      "epoch": 211.59,
      "grad_norm": 8.210007667541504,
      "learning_rate": 2.9469469469469468e-06,
      "loss": 0.2511,
      "step": 70460
    },
    {
      "epoch": 211.62,
      "grad_norm": 9.948978424072266,
      "learning_rate": 2.9459459459459462e-06,
      "loss": 0.202,
      "step": 70470
    },
    {
      "epoch": 211.65,
      "grad_norm": 6.913282871246338,
      "learning_rate": 2.9449449449449452e-06,
      "loss": 0.1921,
      "step": 70480
    },
    {
      "epoch": 211.68,
      "grad_norm": 8.861976623535156,
      "learning_rate": 2.9439439439439443e-06,
      "loss": 0.1773,
      "step": 70490
    },
    {
      "epoch": 211.71,
      "grad_norm": 7.092284679412842,
      "learning_rate": 2.942942942942943e-06,
      "loss": 0.2052,
      "step": 70500
    },
    {
      "epoch": 211.74,
      "grad_norm": 6.368556976318359,
      "learning_rate": 2.941941941941942e-06,
      "loss": 0.1936,
      "step": 70510
    },
    {
      "epoch": 211.77,
      "grad_norm": 16.630992889404297,
      "learning_rate": 2.9409409409409413e-06,
      "loss": 0.2032,
      "step": 70520
    },
    {
      "epoch": 211.8,
      "grad_norm": 5.999650478363037,
      "learning_rate": 2.9399399399399403e-06,
      "loss": 0.1452,
      "step": 70530
    },
    {
      "epoch": 211.83,
      "grad_norm": 10.432190895080566,
      "learning_rate": 2.9389389389389393e-06,
      "loss": 0.2125,
      "step": 70540
    },
    {
      "epoch": 211.86,
      "grad_norm": 9.564906120300293,
      "learning_rate": 2.937937937937938e-06,
      "loss": 0.2116,
      "step": 70550
    },
    {
      "epoch": 211.89,
      "grad_norm": 6.764892101287842,
      "learning_rate": 2.936936936936937e-06,
      "loss": 0.2185,
      "step": 70560
    },
    {
      "epoch": 211.92,
      "grad_norm": 7.846088886260986,
      "learning_rate": 2.9359359359359364e-06,
      "loss": 0.2184,
      "step": 70570
    },
    {
      "epoch": 211.95,
      "grad_norm": 8.219886779785156,
      "learning_rate": 2.9349349349349354e-06,
      "loss": 0.1956,
      "step": 70580
    },
    {
      "epoch": 211.98,
      "grad_norm": 10.829766273498535,
      "learning_rate": 2.933933933933934e-06,
      "loss": 0.1967,
      "step": 70590
    },
    {
      "epoch": 212.0,
      "eval_accuracy": 0.9235,
      "eval_loss": 0.34535786509513855,
      "eval_runtime": 30.0471,
      "eval_samples_per_second": 332.811,
      "eval_steps_per_second": 1.331,
      "step": 70596
    },
    {
      "epoch": 212.01,
      "grad_norm": 20.395723342895508,
      "learning_rate": 2.932932932932933e-06,
      "loss": 0.213,
      "step": 70600
    },
    {
      "epoch": 212.04,
      "grad_norm": 8.814101219177246,
      "learning_rate": 2.931931931931932e-06,
      "loss": 0.2175,
      "step": 70610
    },
    {
      "epoch": 212.07,
      "grad_norm": 23.89792251586914,
      "learning_rate": 2.9309309309309315e-06,
      "loss": 0.2119,
      "step": 70620
    },
    {
      "epoch": 212.1,
      "grad_norm": 6.568928241729736,
      "learning_rate": 2.92992992992993e-06,
      "loss": 0.2249,
      "step": 70630
    },
    {
      "epoch": 212.13,
      "grad_norm": 9.626646995544434,
      "learning_rate": 2.928928928928929e-06,
      "loss": 0.2021,
      "step": 70640
    },
    {
      "epoch": 212.16,
      "grad_norm": 6.2540459632873535,
      "learning_rate": 2.927927927927928e-06,
      "loss": 0.238,
      "step": 70650
    },
    {
      "epoch": 212.19,
      "grad_norm": 24.524145126342773,
      "learning_rate": 2.9269269269269276e-06,
      "loss": 0.1944,
      "step": 70660
    },
    {
      "epoch": 212.22,
      "grad_norm": 13.55052375793457,
      "learning_rate": 2.9259259259259257e-06,
      "loss": 0.2257,
      "step": 70670
    },
    {
      "epoch": 212.25,
      "grad_norm": 7.477406978607178,
      "learning_rate": 2.924924924924925e-06,
      "loss": 0.2073,
      "step": 70680
    },
    {
      "epoch": 212.28,
      "grad_norm": 9.072762489318848,
      "learning_rate": 2.923923923923924e-06,
      "loss": 0.165,
      "step": 70690
    },
    {
      "epoch": 212.31,
      "grad_norm": 8.283666610717773,
      "learning_rate": 2.9229229229229232e-06,
      "loss": 0.2046,
      "step": 70700
    },
    {
      "epoch": 212.34,
      "grad_norm": 7.929000377655029,
      "learning_rate": 2.921921921921922e-06,
      "loss": 0.1849,
      "step": 70710
    },
    {
      "epoch": 212.37,
      "grad_norm": 6.452295780181885,
      "learning_rate": 2.9209209209209213e-06,
      "loss": 0.2082,
      "step": 70720
    },
    {
      "epoch": 212.4,
      "grad_norm": 4.6625566482543945,
      "learning_rate": 2.9199199199199203e-06,
      "loss": 0.1711,
      "step": 70730
    },
    {
      "epoch": 212.43,
      "grad_norm": 8.12144947052002,
      "learning_rate": 2.9189189189189193e-06,
      "loss": 0.2164,
      "step": 70740
    },
    {
      "epoch": 212.46,
      "grad_norm": 7.450501918792725,
      "learning_rate": 2.917917917917918e-06,
      "loss": 0.2027,
      "step": 70750
    },
    {
      "epoch": 212.49,
      "grad_norm": 10.880239486694336,
      "learning_rate": 2.916916916916917e-06,
      "loss": 0.2173,
      "step": 70760
    },
    {
      "epoch": 212.52,
      "grad_norm": 8.497297286987305,
      "learning_rate": 2.9159159159159163e-06,
      "loss": 0.2303,
      "step": 70770
    },
    {
      "epoch": 212.55,
      "grad_norm": 10.547518730163574,
      "learning_rate": 2.9149149149149154e-06,
      "loss": 0.2479,
      "step": 70780
    },
    {
      "epoch": 212.58,
      "grad_norm": 10.81995677947998,
      "learning_rate": 2.9139139139139144e-06,
      "loss": 0.1986,
      "step": 70790
    },
    {
      "epoch": 212.61,
      "grad_norm": 13.778059005737305,
      "learning_rate": 2.912912912912913e-06,
      "loss": 0.185,
      "step": 70800
    },
    {
      "epoch": 212.64,
      "grad_norm": 8.473325729370117,
      "learning_rate": 2.911911911911912e-06,
      "loss": 0.2133,
      "step": 70810
    },
    {
      "epoch": 212.67,
      "grad_norm": 4.058024883270264,
      "learning_rate": 2.9109109109109114e-06,
      "loss": 0.2038,
      "step": 70820
    },
    {
      "epoch": 212.7,
      "grad_norm": 10.112835884094238,
      "learning_rate": 2.9099099099099105e-06,
      "loss": 0.215,
      "step": 70830
    },
    {
      "epoch": 212.73,
      "grad_norm": 8.044441223144531,
      "learning_rate": 2.908908908908909e-06,
      "loss": 0.2135,
      "step": 70840
    },
    {
      "epoch": 212.76,
      "grad_norm": 25.646636962890625,
      "learning_rate": 2.907907907907908e-06,
      "loss": 0.2531,
      "step": 70850
    },
    {
      "epoch": 212.79,
      "grad_norm": 7.205240726470947,
      "learning_rate": 2.906906906906907e-06,
      "loss": 0.2286,
      "step": 70860
    },
    {
      "epoch": 212.82,
      "grad_norm": 7.134045600891113,
      "learning_rate": 2.9059059059059065e-06,
      "loss": 0.2008,
      "step": 70870
    },
    {
      "epoch": 212.85,
      "grad_norm": 12.322681427001953,
      "learning_rate": 2.904904904904905e-06,
      "loss": 0.1884,
      "step": 70880
    },
    {
      "epoch": 212.88,
      "grad_norm": 7.105944633483887,
      "learning_rate": 2.903903903903904e-06,
      "loss": 0.2379,
      "step": 70890
    },
    {
      "epoch": 212.91,
      "grad_norm": 8.589242935180664,
      "learning_rate": 2.902902902902903e-06,
      "loss": 0.1756,
      "step": 70900
    },
    {
      "epoch": 212.94,
      "grad_norm": 11.212593078613281,
      "learning_rate": 2.901901901901902e-06,
      "loss": 0.1708,
      "step": 70910
    },
    {
      "epoch": 212.97,
      "grad_norm": 8.9379301071167,
      "learning_rate": 2.9009009009009008e-06,
      "loss": 0.221,
      "step": 70920
    },
    {
      "epoch": 213.0,
      "eval_accuracy": 0.9239,
      "eval_loss": 0.3499828577041626,
      "eval_runtime": 30.2068,
      "eval_samples_per_second": 331.052,
      "eval_steps_per_second": 1.324,
      "step": 70929
    },
    {
      "epoch": 213.0,
      "grad_norm": 4.507487773895264,
      "learning_rate": 2.8998998998999002e-06,
      "loss": 0.1606,
      "step": 70930
    },
    {
      "epoch": 213.03,
      "grad_norm": 20.41989517211914,
      "learning_rate": 2.8988988988988992e-06,
      "loss": 0.2422,
      "step": 70940
    },
    {
      "epoch": 213.06,
      "grad_norm": 36.32375717163086,
      "learning_rate": 2.8978978978978983e-06,
      "loss": 0.2689,
      "step": 70950
    },
    {
      "epoch": 213.09,
      "grad_norm": 7.120980262756348,
      "learning_rate": 2.896896896896897e-06,
      "loss": 0.2072,
      "step": 70960
    },
    {
      "epoch": 213.12,
      "grad_norm": 4.925622463226318,
      "learning_rate": 2.895895895895896e-06,
      "loss": 0.1787,
      "step": 70970
    },
    {
      "epoch": 213.15,
      "grad_norm": 6.040926933288574,
      "learning_rate": 2.8948948948948953e-06,
      "loss": 0.1923,
      "step": 70980
    },
    {
      "epoch": 213.18,
      "grad_norm": 7.413207054138184,
      "learning_rate": 2.8938938938938943e-06,
      "loss": 0.2043,
      "step": 70990
    },
    {
      "epoch": 213.21,
      "grad_norm": 9.666970252990723,
      "learning_rate": 2.892892892892893e-06,
      "loss": 0.1723,
      "step": 71000
    },
    {
      "epoch": 213.24,
      "grad_norm": 8.363574028015137,
      "learning_rate": 2.891891891891892e-06,
      "loss": 0.1903,
      "step": 71010
    },
    {
      "epoch": 213.27,
      "grad_norm": 4.613482475280762,
      "learning_rate": 2.890890890890891e-06,
      "loss": 0.2157,
      "step": 71020
    },
    {
      "epoch": 213.3,
      "grad_norm": 8.76285171508789,
      "learning_rate": 2.8898898898898904e-06,
      "loss": 0.2298,
      "step": 71030
    },
    {
      "epoch": 213.33,
      "grad_norm": 7.378128528594971,
      "learning_rate": 2.888888888888889e-06,
      "loss": 0.1947,
      "step": 71040
    },
    {
      "epoch": 213.36,
      "grad_norm": 9.30163288116455,
      "learning_rate": 2.887887887887888e-06,
      "loss": 0.208,
      "step": 71050
    },
    {
      "epoch": 213.39,
      "grad_norm": 8.834739685058594,
      "learning_rate": 2.886886886886887e-06,
      "loss": 0.1704,
      "step": 71060
    },
    {
      "epoch": 213.42,
      "grad_norm": 8.515447616577148,
      "learning_rate": 2.885885885885886e-06,
      "loss": 0.2254,
      "step": 71070
    },
    {
      "epoch": 213.45,
      "grad_norm": 8.63796329498291,
      "learning_rate": 2.8848848848848855e-06,
      "loss": 0.1885,
      "step": 71080
    },
    {
      "epoch": 213.48,
      "grad_norm": 17.191904067993164,
      "learning_rate": 2.883883883883884e-06,
      "loss": 0.1909,
      "step": 71090
    },
    {
      "epoch": 213.51,
      "grad_norm": 6.440315246582031,
      "learning_rate": 2.882882882882883e-06,
      "loss": 0.1878,
      "step": 71100
    },
    {
      "epoch": 213.54,
      "grad_norm": 15.28896427154541,
      "learning_rate": 2.881881881881882e-06,
      "loss": 0.183,
      "step": 71110
    },
    {
      "epoch": 213.57,
      "grad_norm": 8.12579345703125,
      "learning_rate": 2.8808808808808816e-06,
      "loss": 0.1936,
      "step": 71120
    },
    {
      "epoch": 213.6,
      "grad_norm": 13.280617713928223,
      "learning_rate": 2.8798798798798797e-06,
      "loss": 0.2174,
      "step": 71130
    },
    {
      "epoch": 213.63,
      "grad_norm": 13.622649192810059,
      "learning_rate": 2.878878878878879e-06,
      "loss": 0.195,
      "step": 71140
    },
    {
      "epoch": 213.66,
      "grad_norm": 7.373317241668701,
      "learning_rate": 2.877877877877878e-06,
      "loss": 0.2277,
      "step": 71150
    },
    {
      "epoch": 213.69,
      "grad_norm": 11.941675186157227,
      "learning_rate": 2.8768768768768772e-06,
      "loss": 0.2025,
      "step": 71160
    },
    {
      "epoch": 213.72,
      "grad_norm": 11.626638412475586,
      "learning_rate": 2.875875875875876e-06,
      "loss": 0.1961,
      "step": 71170
    },
    {
      "epoch": 213.75,
      "grad_norm": 12.83227825164795,
      "learning_rate": 2.8748748748748753e-06,
      "loss": 0.1973,
      "step": 71180
    },
    {
      "epoch": 213.78,
      "grad_norm": 10.523051261901855,
      "learning_rate": 2.8738738738738743e-06,
      "loss": 0.2102,
      "step": 71190
    },
    {
      "epoch": 213.81,
      "grad_norm": 7.714756488800049,
      "learning_rate": 2.8728728728728733e-06,
      "loss": 0.1964,
      "step": 71200
    },
    {
      "epoch": 213.84,
      "grad_norm": 7.046169757843018,
      "learning_rate": 2.871871871871872e-06,
      "loss": 0.2089,
      "step": 71210
    },
    {
      "epoch": 213.87,
      "grad_norm": 8.68962287902832,
      "learning_rate": 2.870870870870871e-06,
      "loss": 0.2278,
      "step": 71220
    },
    {
      "epoch": 213.9,
      "grad_norm": 26.473052978515625,
      "learning_rate": 2.8698698698698703e-06,
      "loss": 0.201,
      "step": 71230
    },
    {
      "epoch": 213.93,
      "grad_norm": 7.375164031982422,
      "learning_rate": 2.8688688688688694e-06,
      "loss": 0.1951,
      "step": 71240
    },
    {
      "epoch": 213.96,
      "grad_norm": 8.632165908813477,
      "learning_rate": 2.867867867867868e-06,
      "loss": 0.2117,
      "step": 71250
    },
    {
      "epoch": 213.99,
      "grad_norm": 10.628283500671387,
      "learning_rate": 2.866866866866867e-06,
      "loss": 0.194,
      "step": 71260
    },
    {
      "epoch": 214.0,
      "eval_accuracy": 0.9231,
      "eval_loss": 0.34677863121032715,
      "eval_runtime": 30.4848,
      "eval_samples_per_second": 328.032,
      "eval_steps_per_second": 1.312,
      "step": 71262
    },
    {
      "epoch": 214.02,
      "grad_norm": 23.461536407470703,
      "learning_rate": 2.865865865865866e-06,
      "loss": 0.1977,
      "step": 71270
    },
    {
      "epoch": 214.05,
      "grad_norm": 11.352763175964355,
      "learning_rate": 2.8648648648648654e-06,
      "loss": 0.2108,
      "step": 71280
    },
    {
      "epoch": 214.08,
      "grad_norm": 11.891490936279297,
      "learning_rate": 2.863863863863864e-06,
      "loss": 0.2148,
      "step": 71290
    },
    {
      "epoch": 214.11,
      "grad_norm": 6.188478469848633,
      "learning_rate": 2.862862862862863e-06,
      "loss": 0.1899,
      "step": 71300
    },
    {
      "epoch": 214.14,
      "grad_norm": 9.143193244934082,
      "learning_rate": 2.861861861861862e-06,
      "loss": 0.2071,
      "step": 71310
    },
    {
      "epoch": 214.17,
      "grad_norm": 7.544147968292236,
      "learning_rate": 2.860860860860861e-06,
      "loss": 0.2265,
      "step": 71320
    },
    {
      "epoch": 214.2,
      "grad_norm": 7.824271202087402,
      "learning_rate": 2.8598598598598605e-06,
      "loss": 0.1988,
      "step": 71330
    },
    {
      "epoch": 214.23,
      "grad_norm": 10.61258316040039,
      "learning_rate": 2.858858858858859e-06,
      "loss": 0.2407,
      "step": 71340
    },
    {
      "epoch": 214.26,
      "grad_norm": 7.024418354034424,
      "learning_rate": 2.857857857857858e-06,
      "loss": 0.184,
      "step": 71350
    },
    {
      "epoch": 214.29,
      "grad_norm": 6.606484413146973,
      "learning_rate": 2.856856856856857e-06,
      "loss": 0.2065,
      "step": 71360
    },
    {
      "epoch": 214.32,
      "grad_norm": 9.904562950134277,
      "learning_rate": 2.855855855855856e-06,
      "loss": 0.2292,
      "step": 71370
    },
    {
      "epoch": 214.35,
      "grad_norm": 9.729063987731934,
      "learning_rate": 2.8548548548548548e-06,
      "loss": 0.2291,
      "step": 71380
    },
    {
      "epoch": 214.38,
      "grad_norm": 11.775930404663086,
      "learning_rate": 2.853853853853854e-06,
      "loss": 0.2376,
      "step": 71390
    },
    {
      "epoch": 214.41,
      "grad_norm": 7.784823894500732,
      "learning_rate": 2.8528528528528532e-06,
      "loss": 0.1917,
      "step": 71400
    },
    {
      "epoch": 214.44,
      "grad_norm": 8.19719409942627,
      "learning_rate": 2.8518518518518522e-06,
      "loss": 0.1843,
      "step": 71410
    },
    {
      "epoch": 214.47,
      "grad_norm": 23.48727035522461,
      "learning_rate": 2.850850850850851e-06,
      "loss": 0.2255,
      "step": 71420
    },
    {
      "epoch": 214.5,
      "grad_norm": 3.8752529621124268,
      "learning_rate": 2.84984984984985e-06,
      "loss": 0.2045,
      "step": 71430
    },
    {
      "epoch": 214.53,
      "grad_norm": 6.7470383644104,
      "learning_rate": 2.8488488488488493e-06,
      "loss": 0.2258,
      "step": 71440
    },
    {
      "epoch": 214.56,
      "grad_norm": 7.743106842041016,
      "learning_rate": 2.8478478478478483e-06,
      "loss": 0.2305,
      "step": 71450
    },
    {
      "epoch": 214.59,
      "grad_norm": 7.937587261199951,
      "learning_rate": 2.846846846846847e-06,
      "loss": 0.2171,
      "step": 71460
    },
    {
      "epoch": 214.62,
      "grad_norm": 10.845245361328125,
      "learning_rate": 2.845845845845846e-06,
      "loss": 0.2019,
      "step": 71470
    },
    {
      "epoch": 214.65,
      "grad_norm": 5.829017639160156,
      "learning_rate": 2.844844844844845e-06,
      "loss": 0.199,
      "step": 71480
    },
    {
      "epoch": 214.68,
      "grad_norm": 10.80936050415039,
      "learning_rate": 2.8438438438438444e-06,
      "loss": 0.1993,
      "step": 71490
    },
    {
      "epoch": 214.71,
      "grad_norm": 8.909308433532715,
      "learning_rate": 2.842842842842843e-06,
      "loss": 0.1738,
      "step": 71500
    },
    {
      "epoch": 214.74,
      "grad_norm": 5.768959999084473,
      "learning_rate": 2.841841841841842e-06,
      "loss": 0.1789,
      "step": 71510
    },
    {
      "epoch": 214.77,
      "grad_norm": 9.296338081359863,
      "learning_rate": 2.840840840840841e-06,
      "loss": 0.2099,
      "step": 71520
    },
    {
      "epoch": 214.8,
      "grad_norm": 10.856645584106445,
      "learning_rate": 2.83983983983984e-06,
      "loss": 0.174,
      "step": 71530
    },
    {
      "epoch": 214.83,
      "grad_norm": 13.668378829956055,
      "learning_rate": 2.8388388388388386e-06,
      "loss": 0.2526,
      "step": 71540
    },
    {
      "epoch": 214.86,
      "grad_norm": 6.775784015655518,
      "learning_rate": 2.837837837837838e-06,
      "loss": 0.171,
      "step": 71550
    },
    {
      "epoch": 214.89,
      "grad_norm": 9.731255531311035,
      "learning_rate": 2.836836836836837e-06,
      "loss": 0.1775,
      "step": 71560
    },
    {
      "epoch": 214.92,
      "grad_norm": 8.46754264831543,
      "learning_rate": 2.835835835835836e-06,
      "loss": 0.1951,
      "step": 71570
    },
    {
      "epoch": 214.95,
      "grad_norm": 6.424610614776611,
      "learning_rate": 2.8348348348348347e-06,
      "loss": 0.1825,
      "step": 71580
    },
    {
      "epoch": 214.98,
      "grad_norm": 4.4361653327941895,
      "learning_rate": 2.8338338338338337e-06,
      "loss": 0.2055,
      "step": 71590
    },
    {
      "epoch": 215.0,
      "eval_accuracy": 0.9221,
      "eval_loss": 0.3454572260379791,
      "eval_runtime": 30.0246,
      "eval_samples_per_second": 333.06,
      "eval_steps_per_second": 1.332,
      "step": 71595
    },
    {
      "epoch": 215.02,
      "grad_norm": 8.404500007629395,
      "learning_rate": 2.832832832832833e-06,
      "loss": 0.1714,
      "step": 71600
    },
    {
      "epoch": 215.05,
      "grad_norm": 32.28673553466797,
      "learning_rate": 2.831831831831832e-06,
      "loss": 0.2298,
      "step": 71610
    },
    {
      "epoch": 215.08,
      "grad_norm": 7.313044548034668,
      "learning_rate": 2.830830830830831e-06,
      "loss": 0.196,
      "step": 71620
    },
    {
      "epoch": 215.11,
      "grad_norm": 8.27155876159668,
      "learning_rate": 2.82982982982983e-06,
      "loss": 0.2379,
      "step": 71630
    },
    {
      "epoch": 215.14,
      "grad_norm": 11.385709762573242,
      "learning_rate": 2.828828828828829e-06,
      "loss": 0.2182,
      "step": 71640
    },
    {
      "epoch": 215.17,
      "grad_norm": 12.198932647705078,
      "learning_rate": 2.8278278278278283e-06,
      "loss": 0.2391,
      "step": 71650
    },
    {
      "epoch": 215.2,
      "grad_norm": 6.899421691894531,
      "learning_rate": 2.8268268268268273e-06,
      "loss": 0.2005,
      "step": 71660
    },
    {
      "epoch": 215.23,
      "grad_norm": 7.768916606903076,
      "learning_rate": 2.825825825825826e-06,
      "loss": 0.2245,
      "step": 71670
    },
    {
      "epoch": 215.26,
      "grad_norm": 7.303072452545166,
      "learning_rate": 2.824824824824825e-06,
      "loss": 0.1803,
      "step": 71680
    },
    {
      "epoch": 215.29,
      "grad_norm": 7.1324591636657715,
      "learning_rate": 2.8238238238238243e-06,
      "loss": 0.1849,
      "step": 71690
    },
    {
      "epoch": 215.32,
      "grad_norm": 15.254097938537598,
      "learning_rate": 2.8228228228228234e-06,
      "loss": 0.2061,
      "step": 71700
    },
    {
      "epoch": 215.35,
      "grad_norm": 6.591416835784912,
      "learning_rate": 2.821821821821822e-06,
      "loss": 0.2241,
      "step": 71710
    },
    {
      "epoch": 215.38,
      "grad_norm": 11.163851737976074,
      "learning_rate": 2.820820820820821e-06,
      "loss": 0.2515,
      "step": 71720
    },
    {
      "epoch": 215.41,
      "grad_norm": 5.331653594970703,
      "learning_rate": 2.81981981981982e-06,
      "loss": 0.1844,
      "step": 71730
    },
    {
      "epoch": 215.44,
      "grad_norm": 4.547163963317871,
      "learning_rate": 2.8188188188188194e-06,
      "loss": 0.1984,
      "step": 71740
    },
    {
      "epoch": 215.47,
      "grad_norm": 10.927351951599121,
      "learning_rate": 2.817817817817818e-06,
      "loss": 0.1766,
      "step": 71750
    },
    {
      "epoch": 215.5,
      "grad_norm": 8.041626930236816,
      "learning_rate": 2.816816816816817e-06,
      "loss": 0.2597,
      "step": 71760
    },
    {
      "epoch": 215.53,
      "grad_norm": 8.665907859802246,
      "learning_rate": 2.815815815815816e-06,
      "loss": 0.2095,
      "step": 71770
    },
    {
      "epoch": 215.56,
      "grad_norm": 45.45501708984375,
      "learning_rate": 2.814814814814815e-06,
      "loss": 0.1725,
      "step": 71780
    },
    {
      "epoch": 215.59,
      "grad_norm": 9.072709083557129,
      "learning_rate": 2.8138138138138137e-06,
      "loss": 0.237,
      "step": 71790
    },
    {
      "epoch": 215.62,
      "grad_norm": 15.897494316101074,
      "learning_rate": 2.812812812812813e-06,
      "loss": 0.1805,
      "step": 71800
    },
    {
      "epoch": 215.65,
      "grad_norm": 8.495500564575195,
      "learning_rate": 2.811811811811812e-06,
      "loss": 0.1673,
      "step": 71810
    },
    {
      "epoch": 215.68,
      "grad_norm": 7.18415641784668,
      "learning_rate": 2.810810810810811e-06,
      "loss": 0.1994,
      "step": 71820
    },
    {
      "epoch": 215.71,
      "grad_norm": 22.13734245300293,
      "learning_rate": 2.8098098098098097e-06,
      "loss": 0.2088,
      "step": 71830
    },
    {
      "epoch": 215.74,
      "grad_norm": 6.479516506195068,
      "learning_rate": 2.8088088088088088e-06,
      "loss": 0.1988,
      "step": 71840
    },
    {
      "epoch": 215.77,
      "grad_norm": 6.1503424644470215,
      "learning_rate": 2.807807807807808e-06,
      "loss": 0.1823,
      "step": 71850
    },
    {
      "epoch": 215.8,
      "grad_norm": 6.404964923858643,
      "learning_rate": 2.8068068068068072e-06,
      "loss": 0.1972,
      "step": 71860
    },
    {
      "epoch": 215.83,
      "grad_norm": 12.038019180297852,
      "learning_rate": 2.8058058058058062e-06,
      "loss": 0.2223,
      "step": 71870
    },
    {
      "epoch": 215.86,
      "grad_norm": 6.7564544677734375,
      "learning_rate": 2.804804804804805e-06,
      "loss": 0.1508,
      "step": 71880
    },
    {
      "epoch": 215.89,
      "grad_norm": 8.153504371643066,
      "learning_rate": 2.803803803803804e-06,
      "loss": 0.1928,
      "step": 71890
    },
    {
      "epoch": 215.92,
      "grad_norm": 7.201042652130127,
      "learning_rate": 2.8028028028028033e-06,
      "loss": 0.1989,
      "step": 71900
    },
    {
      "epoch": 215.95,
      "grad_norm": 7.942685604095459,
      "learning_rate": 2.8018018018018023e-06,
      "loss": 0.2149,
      "step": 71910
    },
    {
      "epoch": 215.98,
      "grad_norm": 10.716872215270996,
      "learning_rate": 2.800800800800801e-06,
      "loss": 0.216,
      "step": 71920
    },
    {
      "epoch": 216.0,
      "eval_accuracy": 0.9215,
      "eval_loss": 0.3482026159763336,
      "eval_runtime": 30.3414,
      "eval_samples_per_second": 329.582,
      "eval_steps_per_second": 1.318,
      "step": 71928
    },
    {
      "epoch": 216.01,
      "grad_norm": 9.925713539123535,
      "learning_rate": 2.7997997997998e-06,
      "loss": 0.2,
      "step": 71930
    },
    {
      "epoch": 216.04,
      "grad_norm": 13.967159271240234,
      "learning_rate": 2.798798798798799e-06,
      "loss": 0.1802,
      "step": 71940
    },
    {
      "epoch": 216.07,
      "grad_norm": 8.819879531860352,
      "learning_rate": 2.7977977977977984e-06,
      "loss": 0.2154,
      "step": 71950
    },
    {
      "epoch": 216.1,
      "grad_norm": 12.817862510681152,
      "learning_rate": 2.796796796796797e-06,
      "loss": 0.1811,
      "step": 71960
    },
    {
      "epoch": 216.13,
      "grad_norm": 7.065882682800293,
      "learning_rate": 2.795795795795796e-06,
      "loss": 0.2428,
      "step": 71970
    },
    {
      "epoch": 216.16,
      "grad_norm": 33.54391098022461,
      "learning_rate": 2.794794794794795e-06,
      "loss": 0.2447,
      "step": 71980
    },
    {
      "epoch": 216.19,
      "grad_norm": 7.196388244628906,
      "learning_rate": 2.793793793793794e-06,
      "loss": 0.2132,
      "step": 71990
    },
    {
      "epoch": 216.22,
      "grad_norm": 8.765576362609863,
      "learning_rate": 2.7927927927927926e-06,
      "loss": 0.1794,
      "step": 72000
    },
    {
      "epoch": 216.25,
      "grad_norm": 6.354937553405762,
      "learning_rate": 2.791791791791792e-06,
      "loss": 0.1982,
      "step": 72010
    },
    {
      "epoch": 216.28,
      "grad_norm": 7.743326187133789,
      "learning_rate": 2.790790790790791e-06,
      "loss": 0.2246,
      "step": 72020
    },
    {
      "epoch": 216.31,
      "grad_norm": 6.839229583740234,
      "learning_rate": 2.78978978978979e-06,
      "loss": 0.1749,
      "step": 72030
    },
    {
      "epoch": 216.34,
      "grad_norm": 4.4341278076171875,
      "learning_rate": 2.7887887887887887e-06,
      "loss": 0.1586,
      "step": 72040
    },
    {
      "epoch": 216.37,
      "grad_norm": 8.581751823425293,
      "learning_rate": 2.7877877877877877e-06,
      "loss": 0.2273,
      "step": 72050
    },
    {
      "epoch": 216.4,
      "grad_norm": 6.973550796508789,
      "learning_rate": 2.786786786786787e-06,
      "loss": 0.2363,
      "step": 72060
    },
    {
      "epoch": 216.43,
      "grad_norm": 13.864154815673828,
      "learning_rate": 2.785785785785786e-06,
      "loss": 0.2138,
      "step": 72070
    },
    {
      "epoch": 216.46,
      "grad_norm": 7.013706684112549,
      "learning_rate": 2.7847847847847848e-06,
      "loss": 0.2219,
      "step": 72080
    },
    {
      "epoch": 216.49,
      "grad_norm": 9.593807220458984,
      "learning_rate": 2.783783783783784e-06,
      "loss": 0.1933,
      "step": 72090
    },
    {
      "epoch": 216.52,
      "grad_norm": 4.271193504333496,
      "learning_rate": 2.782782782782783e-06,
      "loss": 0.1791,
      "step": 72100
    },
    {
      "epoch": 216.55,
      "grad_norm": 10.384673118591309,
      "learning_rate": 2.7817817817817823e-06,
      "loss": 0.2066,
      "step": 72110
    },
    {
      "epoch": 216.58,
      "grad_norm": 9.062880516052246,
      "learning_rate": 2.780780780780781e-06,
      "loss": 0.2075,
      "step": 72120
    },
    {
      "epoch": 216.61,
      "grad_norm": 14.590088844299316,
      "learning_rate": 2.77977977977978e-06,
      "loss": 0.1628,
      "step": 72130
    },
    {
      "epoch": 216.64,
      "grad_norm": 10.022177696228027,
      "learning_rate": 2.778778778778779e-06,
      "loss": 0.1848,
      "step": 72140
    },
    {
      "epoch": 216.67,
      "grad_norm": 6.810096263885498,
      "learning_rate": 2.7777777777777783e-06,
      "loss": 0.2038,
      "step": 72150
    },
    {
      "epoch": 216.7,
      "grad_norm": 5.736238479614258,
      "learning_rate": 2.7767767767767774e-06,
      "loss": 0.1955,
      "step": 72160
    },
    {
      "epoch": 216.73,
      "grad_norm": 24.79778480529785,
      "learning_rate": 2.775775775775776e-06,
      "loss": 0.1833,
      "step": 72170
    },
    {
      "epoch": 216.76,
      "grad_norm": 7.976933002471924,
      "learning_rate": 2.774774774774775e-06,
      "loss": 0.1689,
      "step": 72180
    },
    {
      "epoch": 216.79,
      "grad_norm": 8.30227279663086,
      "learning_rate": 2.773773773773774e-06,
      "loss": 0.1955,
      "step": 72190
    },
    {
      "epoch": 216.82,
      "grad_norm": 11.25728988647461,
      "learning_rate": 2.7727727727727734e-06,
      "loss": 0.2225,
      "step": 72200
    },
    {
      "epoch": 216.85,
      "grad_norm": 14.823025703430176,
      "learning_rate": 2.771771771771772e-06,
      "loss": 0.2186,
      "step": 72210
    },
    {
      "epoch": 216.88,
      "grad_norm": 11.783392906188965,
      "learning_rate": 2.770770770770771e-06,
      "loss": 0.2016,
      "step": 72220
    },
    {
      "epoch": 216.91,
      "grad_norm": 8.442488670349121,
      "learning_rate": 2.76976976976977e-06,
      "loss": 0.2141,
      "step": 72230
    },
    {
      "epoch": 216.94,
      "grad_norm": 11.220757484436035,
      "learning_rate": 2.768768768768769e-06,
      "loss": 0.2298,
      "step": 72240
    },
    {
      "epoch": 216.97,
      "grad_norm": 9.140406608581543,
      "learning_rate": 2.7677677677677677e-06,
      "loss": 0.2002,
      "step": 72250
    },
    {
      "epoch": 217.0,
      "grad_norm": 9.155285835266113,
      "learning_rate": 2.766766766766767e-06,
      "loss": 0.1887,
      "step": 72260
    },
    {
      "epoch": 217.0,
      "eval_accuracy": 0.9218,
      "eval_loss": 0.3495413661003113,
      "eval_runtime": 30.2462,
      "eval_samples_per_second": 330.62,
      "eval_steps_per_second": 1.322,
      "step": 72261
    },
    {
      "epoch": 217.03,
      "grad_norm": 7.243405818939209,
      "learning_rate": 2.765765765765766e-06,
      "loss": 0.1704,
      "step": 72270
    },
    {
      "epoch": 217.06,
      "grad_norm": 11.209080696105957,
      "learning_rate": 2.764764764764765e-06,
      "loss": 0.1739,
      "step": 72280
    },
    {
      "epoch": 217.09,
      "grad_norm": 8.231606483459473,
      "learning_rate": 2.7637637637637637e-06,
      "loss": 0.2082,
      "step": 72290
    },
    {
      "epoch": 217.12,
      "grad_norm": 11.132699012756348,
      "learning_rate": 2.7627627627627628e-06,
      "loss": 0.2012,
      "step": 72300
    },
    {
      "epoch": 217.15,
      "grad_norm": 9.250426292419434,
      "learning_rate": 2.761761761761762e-06,
      "loss": 0.185,
      "step": 72310
    },
    {
      "epoch": 217.18,
      "grad_norm": 13.694327354431152,
      "learning_rate": 2.7607607607607612e-06,
      "loss": 0.2197,
      "step": 72320
    },
    {
      "epoch": 217.21,
      "grad_norm": 10.875238418579102,
      "learning_rate": 2.75975975975976e-06,
      "loss": 0.2025,
      "step": 72330
    },
    {
      "epoch": 217.24,
      "grad_norm": 8.86931324005127,
      "learning_rate": 2.758758758758759e-06,
      "loss": 0.2148,
      "step": 72340
    },
    {
      "epoch": 217.27,
      "grad_norm": 8.289170265197754,
      "learning_rate": 2.757757757757758e-06,
      "loss": 0.2024,
      "step": 72350
    },
    {
      "epoch": 217.3,
      "grad_norm": 10.484915733337402,
      "learning_rate": 2.7567567567567573e-06,
      "loss": 0.2279,
      "step": 72360
    },
    {
      "epoch": 217.33,
      "grad_norm": 26.137731552124023,
      "learning_rate": 2.755755755755756e-06,
      "loss": 0.1998,
      "step": 72370
    },
    {
      "epoch": 217.36,
      "grad_norm": 10.146907806396484,
      "learning_rate": 2.754754754754755e-06,
      "loss": 0.2021,
      "step": 72380
    },
    {
      "epoch": 217.39,
      "grad_norm": 10.525452613830566,
      "learning_rate": 2.753753753753754e-06,
      "loss": 0.1904,
      "step": 72390
    },
    {
      "epoch": 217.42,
      "grad_norm": 16.731857299804688,
      "learning_rate": 2.752752752752753e-06,
      "loss": 0.1568,
      "step": 72400
    },
    {
      "epoch": 217.45,
      "grad_norm": 7.464110851287842,
      "learning_rate": 2.7517517517517524e-06,
      "loss": 0.2172,
      "step": 72410
    },
    {
      "epoch": 217.48,
      "grad_norm": 9.103891372680664,
      "learning_rate": 2.750750750750751e-06,
      "loss": 0.2135,
      "step": 72420
    },
    {
      "epoch": 217.51,
      "grad_norm": 4.846676826477051,
      "learning_rate": 2.74974974974975e-06,
      "loss": 0.2103,
      "step": 72430
    },
    {
      "epoch": 217.54,
      "grad_norm": 7.851668834686279,
      "learning_rate": 2.748748748748749e-06,
      "loss": 0.1941,
      "step": 72440
    },
    {
      "epoch": 217.57,
      "grad_norm": 7.79166841506958,
      "learning_rate": 2.747747747747748e-06,
      "loss": 0.1924,
      "step": 72450
    },
    {
      "epoch": 217.6,
      "grad_norm": 9.125286102294922,
      "learning_rate": 2.7467467467467466e-06,
      "loss": 0.1882,
      "step": 72460
    },
    {
      "epoch": 217.63,
      "grad_norm": 7.658761501312256,
      "learning_rate": 2.745745745745746e-06,
      "loss": 0.1949,
      "step": 72470
    },
    {
      "epoch": 217.66,
      "grad_norm": 5.133977890014648,
      "learning_rate": 2.744744744744745e-06,
      "loss": 0.1886,
      "step": 72480
    },
    {
      "epoch": 217.69,
      "grad_norm": 8.532540321350098,
      "learning_rate": 2.743743743743744e-06,
      "loss": 0.2056,
      "step": 72490
    },
    {
      "epoch": 217.72,
      "grad_norm": 10.203731536865234,
      "learning_rate": 2.7427427427427427e-06,
      "loss": 0.2089,
      "step": 72500
    },
    {
      "epoch": 217.75,
      "grad_norm": 9.485994338989258,
      "learning_rate": 2.7417417417417417e-06,
      "loss": 0.2018,
      "step": 72510
    },
    {
      "epoch": 217.78,
      "grad_norm": 5.906448841094971,
      "learning_rate": 2.740740740740741e-06,
      "loss": 0.2507,
      "step": 72520
    },
    {
      "epoch": 217.81,
      "grad_norm": 5.6497883796691895,
      "learning_rate": 2.73973973973974e-06,
      "loss": 0.2248,
      "step": 72530
    },
    {
      "epoch": 217.84,
      "grad_norm": 13.750214576721191,
      "learning_rate": 2.7387387387387388e-06,
      "loss": 0.2197,
      "step": 72540
    },
    {
      "epoch": 217.87,
      "grad_norm": 10.23741626739502,
      "learning_rate": 2.737737737737738e-06,
      "loss": 0.1963,
      "step": 72550
    },
    {
      "epoch": 217.9,
      "grad_norm": 9.745584487915039,
      "learning_rate": 2.736736736736737e-06,
      "loss": 0.1875,
      "step": 72560
    },
    {
      "epoch": 217.93,
      "grad_norm": 9.018815994262695,
      "learning_rate": 2.7357357357357363e-06,
      "loss": 0.2012,
      "step": 72570
    },
    {
      "epoch": 217.96,
      "grad_norm": 6.561100006103516,
      "learning_rate": 2.734734734734735e-06,
      "loss": 0.209,
      "step": 72580
    },
    {
      "epoch": 217.99,
      "grad_norm": 9.230961799621582,
      "learning_rate": 2.733733733733734e-06,
      "loss": 0.2043,
      "step": 72590
    },
    {
      "epoch": 218.0,
      "eval_accuracy": 0.9237,
      "eval_loss": 0.3457511365413666,
      "eval_runtime": 30.2864,
      "eval_samples_per_second": 330.181,
      "eval_steps_per_second": 1.321,
      "step": 72594
    },
    {
      "epoch": 218.02,
      "grad_norm": 7.157597064971924,
      "learning_rate": 2.732732732732733e-06,
      "loss": 0.2471,
      "step": 72600
    },
    {
      "epoch": 218.05,
      "grad_norm": 8.788311004638672,
      "learning_rate": 2.731731731731732e-06,
      "loss": 0.1557,
      "step": 72610
    },
    {
      "epoch": 218.08,
      "grad_norm": 14.939733505249023,
      "learning_rate": 2.7307307307307305e-06,
      "loss": 0.1997,
      "step": 72620
    },
    {
      "epoch": 218.11,
      "grad_norm": 8.823548316955566,
      "learning_rate": 2.72972972972973e-06,
      "loss": 0.1929,
      "step": 72630
    },
    {
      "epoch": 218.14,
      "grad_norm": 16.27056121826172,
      "learning_rate": 2.728728728728729e-06,
      "loss": 0.21,
      "step": 72640
    },
    {
      "epoch": 218.17,
      "grad_norm": 7.9817070960998535,
      "learning_rate": 2.727727727727728e-06,
      "loss": 0.2515,
      "step": 72650
    },
    {
      "epoch": 218.2,
      "grad_norm": 10.349863052368164,
      "learning_rate": 2.7267267267267274e-06,
      "loss": 0.2064,
      "step": 72660
    },
    {
      "epoch": 218.23,
      "grad_norm": 8.65876579284668,
      "learning_rate": 2.7257257257257256e-06,
      "loss": 0.2168,
      "step": 72670
    },
    {
      "epoch": 218.26,
      "grad_norm": 6.729831695556641,
      "learning_rate": 2.724724724724725e-06,
      "loss": 0.1743,
      "step": 72680
    },
    {
      "epoch": 218.29,
      "grad_norm": 6.149238586425781,
      "learning_rate": 2.723723723723724e-06,
      "loss": 0.197,
      "step": 72690
    },
    {
      "epoch": 218.32,
      "grad_norm": 9.750221252441406,
      "learning_rate": 2.722722722722723e-06,
      "loss": 0.1648,
      "step": 72700
    },
    {
      "epoch": 218.35,
      "grad_norm": 7.370594024658203,
      "learning_rate": 2.7217217217217217e-06,
      "loss": 0.2009,
      "step": 72710
    },
    {
      "epoch": 218.38,
      "grad_norm": 8.326759338378906,
      "learning_rate": 2.720720720720721e-06,
      "loss": 0.2178,
      "step": 72720
    },
    {
      "epoch": 218.41,
      "grad_norm": 10.32857894897461,
      "learning_rate": 2.71971971971972e-06,
      "loss": 0.2136,
      "step": 72730
    },
    {
      "epoch": 218.44,
      "grad_norm": 10.034354209899902,
      "learning_rate": 2.718718718718719e-06,
      "loss": 0.1994,
      "step": 72740
    },
    {
      "epoch": 218.47,
      "grad_norm": 9.013570785522461,
      "learning_rate": 2.7177177177177177e-06,
      "loss": 0.2192,
      "step": 72750
    },
    {
      "epoch": 218.5,
      "grad_norm": 4.903061389923096,
      "learning_rate": 2.7167167167167168e-06,
      "loss": 0.1889,
      "step": 72760
    },
    {
      "epoch": 218.53,
      "grad_norm": 6.754368782043457,
      "learning_rate": 2.715715715715716e-06,
      "loss": 0.1688,
      "step": 72770
    },
    {
      "epoch": 218.56,
      "grad_norm": 11.72839069366455,
      "learning_rate": 2.7147147147147152e-06,
      "loss": 0.1951,
      "step": 72780
    },
    {
      "epoch": 218.59,
      "grad_norm": 10.756854057312012,
      "learning_rate": 2.713713713713714e-06,
      "loss": 0.2441,
      "step": 72790
    },
    {
      "epoch": 218.62,
      "grad_norm": 5.835324764251709,
      "learning_rate": 2.712712712712713e-06,
      "loss": 0.2035,
      "step": 72800
    },
    {
      "epoch": 218.65,
      "grad_norm": 10.61216926574707,
      "learning_rate": 2.711711711711712e-06,
      "loss": 0.2234,
      "step": 72810
    },
    {
      "epoch": 218.68,
      "grad_norm": 6.754578113555908,
      "learning_rate": 2.7107107107107113e-06,
      "loss": 0.2105,
      "step": 72820
    },
    {
      "epoch": 218.71,
      "grad_norm": 9.073264122009277,
      "learning_rate": 2.70970970970971e-06,
      "loss": 0.2285,
      "step": 72830
    },
    {
      "epoch": 218.74,
      "grad_norm": 5.594026565551758,
      "learning_rate": 2.708708708708709e-06,
      "loss": 0.1739,
      "step": 72840
    },
    {
      "epoch": 218.77,
      "grad_norm": 6.108695030212402,
      "learning_rate": 2.707707707707708e-06,
      "loss": 0.1705,
      "step": 72850
    },
    {
      "epoch": 218.8,
      "grad_norm": 8.277609825134277,
      "learning_rate": 2.706706706706707e-06,
      "loss": 0.2068,
      "step": 72860
    },
    {
      "epoch": 218.83,
      "grad_norm": 7.87473201751709,
      "learning_rate": 2.7057057057057055e-06,
      "loss": 0.2083,
      "step": 72870
    },
    {
      "epoch": 218.86,
      "grad_norm": 6.828376770019531,
      "learning_rate": 2.704704704704705e-06,
      "loss": 0.2167,
      "step": 72880
    },
    {
      "epoch": 218.89,
      "grad_norm": 8.389116287231445,
      "learning_rate": 2.703703703703704e-06,
      "loss": 0.1794,
      "step": 72890
    },
    {
      "epoch": 218.92,
      "grad_norm": 5.8197126388549805,
      "learning_rate": 2.702702702702703e-06,
      "loss": 0.2173,
      "step": 72900
    },
    {
      "epoch": 218.95,
      "grad_norm": 11.83924388885498,
      "learning_rate": 2.7017017017017016e-06,
      "loss": 0.241,
      "step": 72910
    },
    {
      "epoch": 218.98,
      "grad_norm": 9.426558494567871,
      "learning_rate": 2.7007007007007006e-06,
      "loss": 0.2022,
      "step": 72920
    },
    {
      "epoch": 219.0,
      "eval_accuracy": 0.9224,
      "eval_loss": 0.34281840920448303,
      "eval_runtime": 30.3954,
      "eval_samples_per_second": 328.997,
      "eval_steps_per_second": 1.316,
      "step": 72927
    },
    {
      "epoch": 219.01,
      "grad_norm": 11.196281433105469,
      "learning_rate": 2.6996996996997e-06,
      "loss": 0.2259,
      "step": 72930
    },
    {
      "epoch": 219.04,
      "grad_norm": 7.248385429382324,
      "learning_rate": 2.698698698698699e-06,
      "loss": 0.2126,
      "step": 72940
    },
    {
      "epoch": 219.07,
      "grad_norm": 6.20833158493042,
      "learning_rate": 2.697697697697698e-06,
      "loss": 0.2081,
      "step": 72950
    },
    {
      "epoch": 219.1,
      "grad_norm": 8.519920349121094,
      "learning_rate": 2.6966966966966967e-06,
      "loss": 0.172,
      "step": 72960
    },
    {
      "epoch": 219.13,
      "grad_norm": 6.8859357833862305,
      "learning_rate": 2.6956956956956957e-06,
      "loss": 0.187,
      "step": 72970
    },
    {
      "epoch": 219.16,
      "grad_norm": 8.50240421295166,
      "learning_rate": 2.694694694694695e-06,
      "loss": 0.1914,
      "step": 72980
    },
    {
      "epoch": 219.19,
      "grad_norm": 8.029461860656738,
      "learning_rate": 2.693693693693694e-06,
      "loss": 0.1969,
      "step": 72990
    },
    {
      "epoch": 219.22,
      "grad_norm": 6.866386413574219,
      "learning_rate": 2.6926926926926928e-06,
      "loss": 0.1899,
      "step": 73000
    },
    {
      "epoch": 219.25,
      "grad_norm": 10.542679786682129,
      "learning_rate": 2.691691691691692e-06,
      "loss": 0.2137,
      "step": 73010
    },
    {
      "epoch": 219.28,
      "grad_norm": 16.749271392822266,
      "learning_rate": 2.690690690690691e-06,
      "loss": 0.2215,
      "step": 73020
    },
    {
      "epoch": 219.31,
      "grad_norm": 8.110306739807129,
      "learning_rate": 2.6896896896896903e-06,
      "loss": 0.1955,
      "step": 73030
    },
    {
      "epoch": 219.34,
      "grad_norm": 7.513057231903076,
      "learning_rate": 2.688688688688689e-06,
      "loss": 0.171,
      "step": 73040
    },
    {
      "epoch": 219.37,
      "grad_norm": 5.497344017028809,
      "learning_rate": 2.687687687687688e-06,
      "loss": 0.1818,
      "step": 73050
    },
    {
      "epoch": 219.4,
      "grad_norm": 7.946734428405762,
      "learning_rate": 2.686686686686687e-06,
      "loss": 0.1792,
      "step": 73060
    },
    {
      "epoch": 219.43,
      "grad_norm": 12.142611503601074,
      "learning_rate": 2.685685685685686e-06,
      "loss": 0.1905,
      "step": 73070
    },
    {
      "epoch": 219.46,
      "grad_norm": 8.013559341430664,
      "learning_rate": 2.6846846846846845e-06,
      "loss": 0.2019,
      "step": 73080
    },
    {
      "epoch": 219.49,
      "grad_norm": 10.399627685546875,
      "learning_rate": 2.683683683683684e-06,
      "loss": 0.1848,
      "step": 73090
    },
    {
      "epoch": 219.52,
      "grad_norm": 14.279886245727539,
      "learning_rate": 2.682682682682683e-06,
      "loss": 0.2095,
      "step": 73100
    },
    {
      "epoch": 219.55,
      "grad_norm": 12.945940971374512,
      "learning_rate": 2.681681681681682e-06,
      "loss": 0.1939,
      "step": 73110
    },
    {
      "epoch": 219.58,
      "grad_norm": 10.047242164611816,
      "learning_rate": 2.6806806806806806e-06,
      "loss": 0.1996,
      "step": 73120
    },
    {
      "epoch": 219.61,
      "grad_norm": 4.870552062988281,
      "learning_rate": 2.6796796796796796e-06,
      "loss": 0.2132,
      "step": 73130
    },
    {
      "epoch": 219.64,
      "grad_norm": 11.649288177490234,
      "learning_rate": 2.678678678678679e-06,
      "loss": 0.1909,
      "step": 73140
    },
    {
      "epoch": 219.67,
      "grad_norm": 10.487556457519531,
      "learning_rate": 2.677677677677678e-06,
      "loss": 0.2297,
      "step": 73150
    },
    {
      "epoch": 219.7,
      "grad_norm": 11.702948570251465,
      "learning_rate": 2.6766766766766766e-06,
      "loss": 0.1846,
      "step": 73160
    },
    {
      "epoch": 219.73,
      "grad_norm": 6.217169284820557,
      "learning_rate": 2.6756756756756757e-06,
      "loss": 0.2077,
      "step": 73170
    },
    {
      "epoch": 219.76,
      "grad_norm": 37.68998336791992,
      "learning_rate": 2.674674674674675e-06,
      "loss": 0.1983,
      "step": 73180
    },
    {
      "epoch": 219.79,
      "grad_norm": 11.19469928741455,
      "learning_rate": 2.673673673673674e-06,
      "loss": 0.2137,
      "step": 73190
    },
    {
      "epoch": 219.82,
      "grad_norm": 9.017317771911621,
      "learning_rate": 2.672672672672673e-06,
      "loss": 0.2111,
      "step": 73200
    },
    {
      "epoch": 219.85,
      "grad_norm": 8.523950576782227,
      "learning_rate": 2.6716716716716717e-06,
      "loss": 0.2264,
      "step": 73210
    },
    {
      "epoch": 219.88,
      "grad_norm": 6.524570465087891,
      "learning_rate": 2.6706706706706708e-06,
      "loss": 0.2416,
      "step": 73220
    },
    {
      "epoch": 219.91,
      "grad_norm": 4.292471408843994,
      "learning_rate": 2.66966966966967e-06,
      "loss": 0.1719,
      "step": 73230
    },
    {
      "epoch": 219.94,
      "grad_norm": 9.695042610168457,
      "learning_rate": 2.6686686686686692e-06,
      "loss": 0.2111,
      "step": 73240
    },
    {
      "epoch": 219.97,
      "grad_norm": 10.235475540161133,
      "learning_rate": 2.667667667667668e-06,
      "loss": 0.241,
      "step": 73250
    },
    {
      "epoch": 220.0,
      "grad_norm": 0.6149976253509521,
      "learning_rate": 2.666666666666667e-06,
      "loss": 0.1834,
      "step": 73260
    },
    {
      "epoch": 220.0,
      "eval_accuracy": 0.9222,
      "eval_loss": 0.3419274091720581,
      "eval_runtime": 30.3071,
      "eval_samples_per_second": 329.956,
      "eval_steps_per_second": 1.32,
      "step": 73260
    },
    {
      "epoch": 220.03,
      "grad_norm": 7.489799499511719,
      "learning_rate": 2.665665665665666e-06,
      "loss": 0.1553,
      "step": 73270
    },
    {
      "epoch": 220.06,
      "grad_norm": 10.625242233276367,
      "learning_rate": 2.6646646646646653e-06,
      "loss": 0.2027,
      "step": 73280
    },
    {
      "epoch": 220.09,
      "grad_norm": 30.546960830688477,
      "learning_rate": 2.663663663663664e-06,
      "loss": 0.2283,
      "step": 73290
    },
    {
      "epoch": 220.12,
      "grad_norm": 6.642341613769531,
      "learning_rate": 2.662662662662663e-06,
      "loss": 0.1829,
      "step": 73300
    },
    {
      "epoch": 220.15,
      "grad_norm": 6.654412746429443,
      "learning_rate": 2.661661661661662e-06,
      "loss": 0.1797,
      "step": 73310
    },
    {
      "epoch": 220.18,
      "grad_norm": 5.4989142417907715,
      "learning_rate": 2.660660660660661e-06,
      "loss": 0.2505,
      "step": 73320
    },
    {
      "epoch": 220.21,
      "grad_norm": 8.788012504577637,
      "learning_rate": 2.6596596596596595e-06,
      "loss": 0.1801,
      "step": 73330
    },
    {
      "epoch": 220.24,
      "grad_norm": 6.398429870605469,
      "learning_rate": 2.658658658658659e-06,
      "loss": 0.1937,
      "step": 73340
    },
    {
      "epoch": 220.27,
      "grad_norm": 9.85134506225586,
      "learning_rate": 2.657657657657658e-06,
      "loss": 0.205,
      "step": 73350
    },
    {
      "epoch": 220.3,
      "grad_norm": 11.559126853942871,
      "learning_rate": 2.656656656656657e-06,
      "loss": 0.1622,
      "step": 73360
    },
    {
      "epoch": 220.33,
      "grad_norm": 10.084897994995117,
      "learning_rate": 2.6556556556556556e-06,
      "loss": 0.1936,
      "step": 73370
    },
    {
      "epoch": 220.36,
      "grad_norm": 6.885250568389893,
      "learning_rate": 2.6546546546546546e-06,
      "loss": 0.2313,
      "step": 73380
    },
    {
      "epoch": 220.39,
      "grad_norm": 7.507399082183838,
      "learning_rate": 2.653653653653654e-06,
      "loss": 0.1504,
      "step": 73390
    },
    {
      "epoch": 220.42,
      "grad_norm": 12.702912330627441,
      "learning_rate": 2.652652652652653e-06,
      "loss": 0.2185,
      "step": 73400
    },
    {
      "epoch": 220.45,
      "grad_norm": 11.054252624511719,
      "learning_rate": 2.6516516516516517e-06,
      "loss": 0.234,
      "step": 73410
    },
    {
      "epoch": 220.48,
      "grad_norm": 12.074002265930176,
      "learning_rate": 2.6506506506506507e-06,
      "loss": 0.1954,
      "step": 73420
    },
    {
      "epoch": 220.51,
      "grad_norm": 8.050591468811035,
      "learning_rate": 2.6496496496496497e-06,
      "loss": 0.1946,
      "step": 73430
    },
    {
      "epoch": 220.54,
      "grad_norm": 15.576664924621582,
      "learning_rate": 2.648648648648649e-06,
      "loss": 0.1947,
      "step": 73440
    },
    {
      "epoch": 220.57,
      "grad_norm": 10.104372024536133,
      "learning_rate": 2.6476476476476478e-06,
      "loss": 0.1942,
      "step": 73450
    },
    {
      "epoch": 220.6,
      "grad_norm": 8.637282371520996,
      "learning_rate": 2.6466466466466468e-06,
      "loss": 0.1969,
      "step": 73460
    },
    {
      "epoch": 220.63,
      "grad_norm": 9.18704891204834,
      "learning_rate": 2.645645645645646e-06,
      "loss": 0.2271,
      "step": 73470
    },
    {
      "epoch": 220.66,
      "grad_norm": 7.51023530960083,
      "learning_rate": 2.644644644644645e-06,
      "loss": 0.2123,
      "step": 73480
    },
    {
      "epoch": 220.69,
      "grad_norm": 6.351064205169678,
      "learning_rate": 2.6436436436436443e-06,
      "loss": 0.2389,
      "step": 73490
    },
    {
      "epoch": 220.72,
      "grad_norm": 7.588408470153809,
      "learning_rate": 2.642642642642643e-06,
      "loss": 0.1883,
      "step": 73500
    },
    {
      "epoch": 220.75,
      "grad_norm": 5.762750625610352,
      "learning_rate": 2.641641641641642e-06,
      "loss": 0.2035,
      "step": 73510
    },
    {
      "epoch": 220.78,
      "grad_norm": 7.591121673583984,
      "learning_rate": 2.640640640640641e-06,
      "loss": 0.2011,
      "step": 73520
    },
    {
      "epoch": 220.81,
      "grad_norm": 9.735515594482422,
      "learning_rate": 2.63963963963964e-06,
      "loss": 0.2324,
      "step": 73530
    },
    {
      "epoch": 220.84,
      "grad_norm": 12.19159984588623,
      "learning_rate": 2.6386386386386385e-06,
      "loss": 0.207,
      "step": 73540
    },
    {
      "epoch": 220.87,
      "grad_norm": 15.916427612304688,
      "learning_rate": 2.637637637637638e-06,
      "loss": 0.1739,
      "step": 73550
    },
    {
      "epoch": 220.9,
      "grad_norm": 10.165000915527344,
      "learning_rate": 2.636636636636637e-06,
      "loss": 0.1875,
      "step": 73560
    },
    {
      "epoch": 220.93,
      "grad_norm": 8.480216979980469,
      "learning_rate": 2.635635635635636e-06,
      "loss": 0.1845,
      "step": 73570
    },
    {
      "epoch": 220.96,
      "grad_norm": 5.643638610839844,
      "learning_rate": 2.6346346346346346e-06,
      "loss": 0.2025,
      "step": 73580
    },
    {
      "epoch": 220.99,
      "grad_norm": 16.773773193359375,
      "learning_rate": 2.6336336336336336e-06,
      "loss": 0.1835,
      "step": 73590
    },
    {
      "epoch": 221.0,
      "eval_accuracy": 0.9231,
      "eval_loss": 0.3446863293647766,
      "eval_runtime": 29.8531,
      "eval_samples_per_second": 334.973,
      "eval_steps_per_second": 1.34,
      "step": 73593
    },
    {
      "epoch": 221.02,
      "grad_norm": 9.704315185546875,
      "learning_rate": 2.632632632632633e-06,
      "loss": 0.1795,
      "step": 73600
    },
    {
      "epoch": 221.05,
      "grad_norm": 8.895734786987305,
      "learning_rate": 2.631631631631632e-06,
      "loss": 0.2405,
      "step": 73610
    },
    {
      "epoch": 221.08,
      "grad_norm": 9.687232971191406,
      "learning_rate": 2.6306306306306306e-06,
      "loss": 0.2201,
      "step": 73620
    },
    {
      "epoch": 221.11,
      "grad_norm": 27.024507522583008,
      "learning_rate": 2.6296296296296297e-06,
      "loss": 0.2155,
      "step": 73630
    },
    {
      "epoch": 221.14,
      "grad_norm": 7.7666521072387695,
      "learning_rate": 2.6286286286286287e-06,
      "loss": 0.2139,
      "step": 73640
    },
    {
      "epoch": 221.17,
      "grad_norm": 13.77450942993164,
      "learning_rate": 2.627627627627628e-06,
      "loss": 0.1947,
      "step": 73650
    },
    {
      "epoch": 221.2,
      "grad_norm": 5.5461883544921875,
      "learning_rate": 2.6266266266266267e-06,
      "loss": 0.1714,
      "step": 73660
    },
    {
      "epoch": 221.23,
      "grad_norm": 9.810296058654785,
      "learning_rate": 2.6256256256256257e-06,
      "loss": 0.1797,
      "step": 73670
    },
    {
      "epoch": 221.26,
      "grad_norm": 4.586328983306885,
      "learning_rate": 2.6246246246246248e-06,
      "loss": 0.1697,
      "step": 73680
    },
    {
      "epoch": 221.29,
      "grad_norm": 13.49818229675293,
      "learning_rate": 2.623623623623624e-06,
      "loss": 0.2145,
      "step": 73690
    },
    {
      "epoch": 221.32,
      "grad_norm": 9.978578567504883,
      "learning_rate": 2.6226226226226224e-06,
      "loss": 0.2013,
      "step": 73700
    },
    {
      "epoch": 221.35,
      "grad_norm": 9.913068771362305,
      "learning_rate": 2.621621621621622e-06,
      "loss": 0.2119,
      "step": 73710
    },
    {
      "epoch": 221.38,
      "grad_norm": 8.233881950378418,
      "learning_rate": 2.620620620620621e-06,
      "loss": 0.1919,
      "step": 73720
    },
    {
      "epoch": 221.41,
      "grad_norm": 10.886078834533691,
      "learning_rate": 2.61961961961962e-06,
      "loss": 0.2099,
      "step": 73730
    },
    {
      "epoch": 221.44,
      "grad_norm": 10.008456230163574,
      "learning_rate": 2.6186186186186193e-06,
      "loss": 0.1842,
      "step": 73740
    },
    {
      "epoch": 221.47,
      "grad_norm": 9.288802146911621,
      "learning_rate": 2.617617617617618e-06,
      "loss": 0.2107,
      "step": 73750
    },
    {
      "epoch": 221.5,
      "grad_norm": 9.149421691894531,
      "learning_rate": 2.616616616616617e-06,
      "loss": 0.2102,
      "step": 73760
    },
    {
      "epoch": 221.53,
      "grad_norm": 6.744452476501465,
      "learning_rate": 2.615615615615616e-06,
      "loss": 0.1779,
      "step": 73770
    },
    {
      "epoch": 221.56,
      "grad_norm": 10.806241035461426,
      "learning_rate": 2.614614614614615e-06,
      "loss": 0.192,
      "step": 73780
    },
    {
      "epoch": 221.59,
      "grad_norm": 7.6963887214660645,
      "learning_rate": 2.6136136136136135e-06,
      "loss": 0.1784,
      "step": 73790
    },
    {
      "epoch": 221.62,
      "grad_norm": 6.816718578338623,
      "learning_rate": 2.612612612612613e-06,
      "loss": 0.2468,
      "step": 73800
    },
    {
      "epoch": 221.65,
      "grad_norm": 5.7830986976623535,
      "learning_rate": 2.611611611611612e-06,
      "loss": 0.1858,
      "step": 73810
    },
    {
      "epoch": 221.68,
      "grad_norm": 5.219381332397461,
      "learning_rate": 2.610610610610611e-06,
      "loss": 0.1855,
      "step": 73820
    },
    {
      "epoch": 221.71,
      "grad_norm": 5.6388468742370605,
      "learning_rate": 2.6096096096096096e-06,
      "loss": 0.2142,
      "step": 73830
    },
    {
      "epoch": 221.74,
      "grad_norm": 17.077388763427734,
      "learning_rate": 2.6086086086086086e-06,
      "loss": 0.2279,
      "step": 73840
    },
    {
      "epoch": 221.77,
      "grad_norm": 10.342958450317383,
      "learning_rate": 2.607607607607608e-06,
      "loss": 0.209,
      "step": 73850
    },
    {
      "epoch": 221.8,
      "grad_norm": 5.337224960327148,
      "learning_rate": 2.606606606606607e-06,
      "loss": 0.1772,
      "step": 73860
    },
    {
      "epoch": 221.83,
      "grad_norm": 14.546438217163086,
      "learning_rate": 2.6056056056056057e-06,
      "loss": 0.2085,
      "step": 73870
    },
    {
      "epoch": 221.86,
      "grad_norm": 15.704980850219727,
      "learning_rate": 2.6046046046046047e-06,
      "loss": 0.228,
      "step": 73880
    },
    {
      "epoch": 221.89,
      "grad_norm": 9.408868789672852,
      "learning_rate": 2.6036036036036037e-06,
      "loss": 0.2196,
      "step": 73890
    },
    {
      "epoch": 221.92,
      "grad_norm": 7.382274150848389,
      "learning_rate": 2.602602602602603e-06,
      "loss": 0.1986,
      "step": 73900
    },
    {
      "epoch": 221.95,
      "grad_norm": 9.318102836608887,
      "learning_rate": 2.6016016016016017e-06,
      "loss": 0.1889,
      "step": 73910
    },
    {
      "epoch": 221.98,
      "grad_norm": 7.4415483474731445,
      "learning_rate": 2.6006006006006008e-06,
      "loss": 0.2194,
      "step": 73920
    },
    {
      "epoch": 222.0,
      "eval_accuracy": 0.9218,
      "eval_loss": 0.34721747040748596,
      "eval_runtime": 30.292,
      "eval_samples_per_second": 330.12,
      "eval_steps_per_second": 1.32,
      "step": 73926
    },
    {
      "epoch": 222.01,
      "grad_norm": 6.644683837890625,
      "learning_rate": 2.5995995995995998e-06,
      "loss": 0.1757,
      "step": 73930
    },
    {
      "epoch": 222.04,
      "grad_norm": 9.237369537353516,
      "learning_rate": 2.598598598598599e-06,
      "loss": 0.1922,
      "step": 73940
    },
    {
      "epoch": 222.07,
      "grad_norm": 8.667475700378418,
      "learning_rate": 2.5975975975975974e-06,
      "loss": 0.1878,
      "step": 73950
    },
    {
      "epoch": 222.1,
      "grad_norm": 12.66580867767334,
      "learning_rate": 2.596596596596597e-06,
      "loss": 0.2279,
      "step": 73960
    },
    {
      "epoch": 222.13,
      "grad_norm": 4.927701473236084,
      "learning_rate": 2.595595595595596e-06,
      "loss": 0.1698,
      "step": 73970
    },
    {
      "epoch": 222.16,
      "grad_norm": 9.017887115478516,
      "learning_rate": 2.594594594594595e-06,
      "loss": 0.1896,
      "step": 73980
    },
    {
      "epoch": 222.19,
      "grad_norm": 7.997684955596924,
      "learning_rate": 2.5935935935935935e-06,
      "loss": 0.1895,
      "step": 73990
    },
    {
      "epoch": 222.22,
      "grad_norm": 6.602138519287109,
      "learning_rate": 2.5925925925925925e-06,
      "loss": 0.1798,
      "step": 74000
    },
    {
      "epoch": 222.25,
      "grad_norm": 10.07337474822998,
      "learning_rate": 2.591591591591592e-06,
      "loss": 0.1734,
      "step": 74010
    },
    {
      "epoch": 222.28,
      "grad_norm": 7.65058708190918,
      "learning_rate": 2.590590590590591e-06,
      "loss": 0.2267,
      "step": 74020
    },
    {
      "epoch": 222.31,
      "grad_norm": 13.038058280944824,
      "learning_rate": 2.58958958958959e-06,
      "loss": 0.2358,
      "step": 74030
    },
    {
      "epoch": 222.34,
      "grad_norm": 6.917391777038574,
      "learning_rate": 2.5885885885885886e-06,
      "loss": 0.1978,
      "step": 74040
    },
    {
      "epoch": 222.37,
      "grad_norm": 8.488529205322266,
      "learning_rate": 2.5875875875875876e-06,
      "loss": 0.1852,
      "step": 74050
    },
    {
      "epoch": 222.4,
      "grad_norm": 5.486719131469727,
      "learning_rate": 2.586586586586587e-06,
      "loss": 0.148,
      "step": 74060
    },
    {
      "epoch": 222.43,
      "grad_norm": 12.247118949890137,
      "learning_rate": 2.585585585585586e-06,
      "loss": 0.248,
      "step": 74070
    },
    {
      "epoch": 222.46,
      "grad_norm": 11.278295516967773,
      "learning_rate": 2.5845845845845846e-06,
      "loss": 0.1996,
      "step": 74080
    },
    {
      "epoch": 222.49,
      "grad_norm": 5.714122295379639,
      "learning_rate": 2.5835835835835837e-06,
      "loss": 0.1658,
      "step": 74090
    },
    {
      "epoch": 222.52,
      "grad_norm": 5.555538177490234,
      "learning_rate": 2.5825825825825827e-06,
      "loss": 0.2035,
      "step": 74100
    },
    {
      "epoch": 222.55,
      "grad_norm": 30.707279205322266,
      "learning_rate": 2.581581581581582e-06,
      "loss": 0.2087,
      "step": 74110
    },
    {
      "epoch": 222.58,
      "grad_norm": 12.643987655639648,
      "learning_rate": 2.5805805805805807e-06,
      "loss": 0.1787,
      "step": 74120
    },
    {
      "epoch": 222.61,
      "grad_norm": 8.386334419250488,
      "learning_rate": 2.5795795795795797e-06,
      "loss": 0.231,
      "step": 74130
    },
    {
      "epoch": 222.64,
      "grad_norm": 112.66641235351562,
      "learning_rate": 2.5785785785785787e-06,
      "loss": 0.1921,
      "step": 74140
    },
    {
      "epoch": 222.67,
      "grad_norm": 10.778056144714355,
      "learning_rate": 2.577577577577578e-06,
      "loss": 0.2185,
      "step": 74150
    },
    {
      "epoch": 222.7,
      "grad_norm": 13.09435749053955,
      "learning_rate": 2.5765765765765764e-06,
      "loss": 0.2196,
      "step": 74160
    },
    {
      "epoch": 222.73,
      "grad_norm": 51.73065948486328,
      "learning_rate": 2.575575575575576e-06,
      "loss": 0.1911,
      "step": 74170
    },
    {
      "epoch": 222.76,
      "grad_norm": 7.6161603927612305,
      "learning_rate": 2.574574574574575e-06,
      "loss": 0.1933,
      "step": 74180
    },
    {
      "epoch": 222.79,
      "grad_norm": 12.05302619934082,
      "learning_rate": 2.573573573573574e-06,
      "loss": 0.2186,
      "step": 74190
    },
    {
      "epoch": 222.82,
      "grad_norm": 13.202037811279297,
      "learning_rate": 2.5725725725725724e-06,
      "loss": 0.1935,
      "step": 74200
    },
    {
      "epoch": 222.85,
      "grad_norm": 7.089195251464844,
      "learning_rate": 2.571571571571572e-06,
      "loss": 0.206,
      "step": 74210
    },
    {
      "epoch": 222.88,
      "grad_norm": 9.69510269165039,
      "learning_rate": 2.570570570570571e-06,
      "loss": 0.2059,
      "step": 74220
    },
    {
      "epoch": 222.91,
      "grad_norm": 12.655362129211426,
      "learning_rate": 2.56956956956957e-06,
      "loss": 0.2338,
      "step": 74230
    },
    {
      "epoch": 222.94,
      "grad_norm": 6.272221088409424,
      "learning_rate": 2.5685685685685685e-06,
      "loss": 0.2077,
      "step": 74240
    },
    {
      "epoch": 222.97,
      "grad_norm": 13.807075500488281,
      "learning_rate": 2.5675675675675675e-06,
      "loss": 0.1775,
      "step": 74250
    },
    {
      "epoch": 223.0,
      "eval_accuracy": 0.922,
      "eval_loss": 0.3465680778026581,
      "eval_runtime": 30.3817,
      "eval_samples_per_second": 329.146,
      "eval_steps_per_second": 1.317,
      "step": 74259
    },
    {
      "epoch": 223.0,
      "grad_norm": 9.5873441696167,
      "learning_rate": 2.566566566566567e-06,
      "loss": 0.2116,
      "step": 74260
    },
    {
      "epoch": 223.03,
      "grad_norm": 6.2512288093566895,
      "learning_rate": 2.565565565565566e-06,
      "loss": 0.1583,
      "step": 74270
    },
    {
      "epoch": 223.06,
      "grad_norm": 6.030568599700928,
      "learning_rate": 2.564564564564565e-06,
      "loss": 0.1827,
      "step": 74280
    },
    {
      "epoch": 223.09,
      "grad_norm": 8.08230209350586,
      "learning_rate": 2.5635635635635636e-06,
      "loss": 0.2043,
      "step": 74290
    },
    {
      "epoch": 223.12,
      "grad_norm": 4.796847343444824,
      "learning_rate": 2.5625625625625626e-06,
      "loss": 0.1856,
      "step": 74300
    },
    {
      "epoch": 223.15,
      "grad_norm": 10.184306144714355,
      "learning_rate": 2.561561561561562e-06,
      "loss": 0.1941,
      "step": 74310
    },
    {
      "epoch": 223.18,
      "grad_norm": 5.808650016784668,
      "learning_rate": 2.560560560560561e-06,
      "loss": 0.1687,
      "step": 74320
    },
    {
      "epoch": 223.21,
      "grad_norm": 10.081612586975098,
      "learning_rate": 2.5595595595595597e-06,
      "loss": 0.2035,
      "step": 74330
    },
    {
      "epoch": 223.24,
      "grad_norm": 21.785295486450195,
      "learning_rate": 2.5585585585585587e-06,
      "loss": 0.1642,
      "step": 74340
    },
    {
      "epoch": 223.27,
      "grad_norm": 11.381464004516602,
      "learning_rate": 2.5575575575575577e-06,
      "loss": 0.2088,
      "step": 74350
    },
    {
      "epoch": 223.3,
      "grad_norm": 11.478914260864258,
      "learning_rate": 2.556556556556557e-06,
      "loss": 0.1886,
      "step": 74360
    },
    {
      "epoch": 223.33,
      "grad_norm": 5.98668098449707,
      "learning_rate": 2.5555555555555557e-06,
      "loss": 0.1842,
      "step": 74370
    },
    {
      "epoch": 223.36,
      "grad_norm": 11.758206367492676,
      "learning_rate": 2.5545545545545548e-06,
      "loss": 0.1833,
      "step": 74380
    },
    {
      "epoch": 223.39,
      "grad_norm": 11.696633338928223,
      "learning_rate": 2.5535535535535538e-06,
      "loss": 0.2221,
      "step": 74390
    },
    {
      "epoch": 223.42,
      "grad_norm": 6.494028091430664,
      "learning_rate": 2.552552552552553e-06,
      "loss": 0.1897,
      "step": 74400
    },
    {
      "epoch": 223.45,
      "grad_norm": 7.899448871612549,
      "learning_rate": 2.5515515515515514e-06,
      "loss": 0.1838,
      "step": 74410
    },
    {
      "epoch": 223.48,
      "grad_norm": 6.147073745727539,
      "learning_rate": 2.550550550550551e-06,
      "loss": 0.1548,
      "step": 74420
    },
    {
      "epoch": 223.51,
      "grad_norm": 7.515016078948975,
      "learning_rate": 2.54954954954955e-06,
      "loss": 0.2401,
      "step": 74430
    },
    {
      "epoch": 223.54,
      "grad_norm": 7.296377658843994,
      "learning_rate": 2.548548548548549e-06,
      "loss": 0.2034,
      "step": 74440
    },
    {
      "epoch": 223.57,
      "grad_norm": 9.166295051574707,
      "learning_rate": 2.5475475475475475e-06,
      "loss": 0.2131,
      "step": 74450
    },
    {
      "epoch": 223.6,
      "grad_norm": 9.167292594909668,
      "learning_rate": 2.5465465465465465e-06,
      "loss": 0.227,
      "step": 74460
    },
    {
      "epoch": 223.63,
      "grad_norm": 19.5406436920166,
      "learning_rate": 2.545545545545546e-06,
      "loss": 0.1979,
      "step": 74470
    },
    {
      "epoch": 223.66,
      "grad_norm": 13.118786811828613,
      "learning_rate": 2.544544544544545e-06,
      "loss": 0.2165,
      "step": 74480
    },
    {
      "epoch": 223.69,
      "grad_norm": 8.188241958618164,
      "learning_rate": 2.5435435435435435e-06,
      "loss": 0.2226,
      "step": 74490
    },
    {
      "epoch": 223.72,
      "grad_norm": 8.490690231323242,
      "learning_rate": 2.5425425425425426e-06,
      "loss": 0.1914,
      "step": 74500
    },
    {
      "epoch": 223.75,
      "grad_norm": 7.575039863586426,
      "learning_rate": 2.5415415415415416e-06,
      "loss": 0.2145,
      "step": 74510
    },
    {
      "epoch": 223.78,
      "grad_norm": 12.27540397644043,
      "learning_rate": 2.540540540540541e-06,
      "loss": 0.2409,
      "step": 74520
    },
    {
      "epoch": 223.81,
      "grad_norm": 9.944217681884766,
      "learning_rate": 2.53953953953954e-06,
      "loss": 0.2067,
      "step": 74530
    },
    {
      "epoch": 223.84,
      "grad_norm": 6.62805700302124,
      "learning_rate": 2.5385385385385386e-06,
      "loss": 0.1862,
      "step": 74540
    },
    {
      "epoch": 223.87,
      "grad_norm": 7.555498123168945,
      "learning_rate": 2.5375375375375377e-06,
      "loss": 0.2015,
      "step": 74550
    },
    {
      "epoch": 223.9,
      "grad_norm": 49.92794418334961,
      "learning_rate": 2.5365365365365367e-06,
      "loss": 0.2184,
      "step": 74560
    },
    {
      "epoch": 223.93,
      "grad_norm": 9.829193115234375,
      "learning_rate": 2.535535535535536e-06,
      "loss": 0.2099,
      "step": 74570
    },
    {
      "epoch": 223.96,
      "grad_norm": 17.51436424255371,
      "learning_rate": 2.5345345345345347e-06,
      "loss": 0.1822,
      "step": 74580
    },
    {
      "epoch": 223.99,
      "grad_norm": 7.9646477699279785,
      "learning_rate": 2.5335335335335337e-06,
      "loss": 0.1781,
      "step": 74590
    },
    {
      "epoch": 224.0,
      "eval_accuracy": 0.9233,
      "eval_loss": 0.3505202829837799,
      "eval_runtime": 29.7815,
      "eval_samples_per_second": 335.779,
      "eval_steps_per_second": 1.343,
      "step": 74592
    },
    {
      "epoch": 224.02,
      "grad_norm": 10.25697135925293,
      "learning_rate": 2.5325325325325327e-06,
      "loss": 0.3234,
      "step": 74600
    },
    {
      "epoch": 224.05,
      "grad_norm": 3.4865219593048096,
      "learning_rate": 2.5315315315315318e-06,
      "loss": 0.2407,
      "step": 74610
    },
    {
      "epoch": 224.08,
      "grad_norm": 5.854301452636719,
      "learning_rate": 2.5305305305305304e-06,
      "loss": 0.1575,
      "step": 74620
    },
    {
      "epoch": 224.11,
      "grad_norm": 7.642549991607666,
      "learning_rate": 2.52952952952953e-06,
      "loss": 0.1983,
      "step": 74630
    },
    {
      "epoch": 224.14,
      "grad_norm": 8.314223289489746,
      "learning_rate": 2.528528528528529e-06,
      "loss": 0.2245,
      "step": 74640
    },
    {
      "epoch": 224.17,
      "grad_norm": 10.357461929321289,
      "learning_rate": 2.527527527527528e-06,
      "loss": 0.199,
      "step": 74650
    },
    {
      "epoch": 224.2,
      "grad_norm": 8.178409576416016,
      "learning_rate": 2.5265265265265264e-06,
      "loss": 0.2063,
      "step": 74660
    },
    {
      "epoch": 224.23,
      "grad_norm": 17.626436233520508,
      "learning_rate": 2.5255255255255254e-06,
      "loss": 0.2072,
      "step": 74670
    },
    {
      "epoch": 224.26,
      "grad_norm": 25.500364303588867,
      "learning_rate": 2.524524524524525e-06,
      "loss": 0.1736,
      "step": 74680
    },
    {
      "epoch": 224.29,
      "grad_norm": 8.849148750305176,
      "learning_rate": 2.523523523523524e-06,
      "loss": 0.2476,
      "step": 74690
    },
    {
      "epoch": 224.32,
      "grad_norm": 3.798586130142212,
      "learning_rate": 2.5225225225225225e-06,
      "loss": 0.1983,
      "step": 74700
    },
    {
      "epoch": 224.35,
      "grad_norm": 6.491689205169678,
      "learning_rate": 2.5215215215215215e-06,
      "loss": 0.1942,
      "step": 74710
    },
    {
      "epoch": 224.38,
      "grad_norm": 14.530172348022461,
      "learning_rate": 2.520520520520521e-06,
      "loss": 0.1952,
      "step": 74720
    },
    {
      "epoch": 224.41,
      "grad_norm": 8.1693754196167,
      "learning_rate": 2.51951951951952e-06,
      "loss": 0.2117,
      "step": 74730
    },
    {
      "epoch": 224.44,
      "grad_norm": 10.478048324584961,
      "learning_rate": 2.5185185185185186e-06,
      "loss": 0.1709,
      "step": 74740
    },
    {
      "epoch": 224.47,
      "grad_norm": 15.083818435668945,
      "learning_rate": 2.5175175175175176e-06,
      "loss": 0.1884,
      "step": 74750
    },
    {
      "epoch": 224.5,
      "grad_norm": 6.968499183654785,
      "learning_rate": 2.5165165165165166e-06,
      "loss": 0.1833,
      "step": 74760
    },
    {
      "epoch": 224.53,
      "grad_norm": 9.864310264587402,
      "learning_rate": 2.515515515515516e-06,
      "loss": 0.2095,
      "step": 74770
    },
    {
      "epoch": 224.56,
      "grad_norm": 5.818115711212158,
      "learning_rate": 2.5145145145145147e-06,
      "loss": 0.1737,
      "step": 74780
    },
    {
      "epoch": 224.59,
      "grad_norm": 7.289052963256836,
      "learning_rate": 2.5135135135135137e-06,
      "loss": 0.2343,
      "step": 74790
    },
    {
      "epoch": 224.62,
      "grad_norm": 12.624438285827637,
      "learning_rate": 2.5125125125125127e-06,
      "loss": 0.1827,
      "step": 74800
    },
    {
      "epoch": 224.65,
      "grad_norm": 8.8141508102417,
      "learning_rate": 2.5115115115115117e-06,
      "loss": 0.1895,
      "step": 74810
    },
    {
      "epoch": 224.68,
      "grad_norm": 41.93707275390625,
      "learning_rate": 2.510510510510511e-06,
      "loss": 0.2188,
      "step": 74820
    },
    {
      "epoch": 224.71,
      "grad_norm": 11.06936264038086,
      "learning_rate": 2.5095095095095097e-06,
      "loss": 0.1704,
      "step": 74830
    },
    {
      "epoch": 224.74,
      "grad_norm": 7.881535530090332,
      "learning_rate": 2.5085085085085088e-06,
      "loss": 0.2479,
      "step": 74840
    },
    {
      "epoch": 224.77,
      "grad_norm": 11.873896598815918,
      "learning_rate": 2.5075075075075078e-06,
      "loss": 0.1763,
      "step": 74850
    },
    {
      "epoch": 224.8,
      "grad_norm": 14.784677505493164,
      "learning_rate": 2.506506506506507e-06,
      "loss": 0.2141,
      "step": 74860
    },
    {
      "epoch": 224.83,
      "grad_norm": 11.823383331298828,
      "learning_rate": 2.5055055055055054e-06,
      "loss": 0.1767,
      "step": 74870
    },
    {
      "epoch": 224.86,
      "grad_norm": 8.350910186767578,
      "learning_rate": 2.504504504504505e-06,
      "loss": 0.2113,
      "step": 74880
    },
    {
      "epoch": 224.89,
      "grad_norm": 8.238725662231445,
      "learning_rate": 2.503503503503504e-06,
      "loss": 0.2007,
      "step": 74890
    },
    {
      "epoch": 224.92,
      "grad_norm": 6.922833442687988,
      "learning_rate": 2.502502502502503e-06,
      "loss": 0.1835,
      "step": 74900
    },
    {
      "epoch": 224.95,
      "grad_norm": 8.142541885375977,
      "learning_rate": 2.5015015015015015e-06,
      "loss": 0.1908,
      "step": 74910
    },
    {
      "epoch": 224.98,
      "grad_norm": 14.21639633178711,
      "learning_rate": 2.5005005005005005e-06,
      "loss": 0.2001,
      "step": 74920
    },
    {
      "epoch": 225.0,
      "eval_accuracy": 0.9226,
      "eval_loss": 0.3476966917514801,
      "eval_runtime": 30.1972,
      "eval_samples_per_second": 331.157,
      "eval_steps_per_second": 1.325,
      "step": 74925
    },
    {
      "epoch": 225.02,
      "grad_norm": 5.387598514556885,
      "learning_rate": 2.4994994994995e-06,
      "loss": 0.1855,
      "step": 74930
    },
    {
      "epoch": 225.05,
      "grad_norm": 5.197170734405518,
      "learning_rate": 2.4984984984984985e-06,
      "loss": 0.1888,
      "step": 74940
    },
    {
      "epoch": 225.08,
      "grad_norm": 15.583074569702148,
      "learning_rate": 2.497497497497498e-06,
      "loss": 0.2093,
      "step": 74950
    },
    {
      "epoch": 225.11,
      "grad_norm": 12.630781173706055,
      "learning_rate": 2.4964964964964966e-06,
      "loss": 0.1959,
      "step": 74960
    },
    {
      "epoch": 225.14,
      "grad_norm": 8.129318237304688,
      "learning_rate": 2.4954954954954956e-06,
      "loss": 0.2094,
      "step": 74970
    },
    {
      "epoch": 225.17,
      "grad_norm": 4.533878803253174,
      "learning_rate": 2.4944944944944946e-06,
      "loss": 0.1889,
      "step": 74980
    },
    {
      "epoch": 225.2,
      "grad_norm": 8.877197265625,
      "learning_rate": 2.4934934934934936e-06,
      "loss": 0.2092,
      "step": 74990
    },
    {
      "epoch": 225.23,
      "grad_norm": 7.808531284332275,
      "learning_rate": 2.4924924924924926e-06,
      "loss": 0.2229,
      "step": 75000
    },
    {
      "epoch": 225.26,
      "grad_norm": 10.788240432739258,
      "learning_rate": 2.4914914914914916e-06,
      "loss": 0.2333,
      "step": 75010
    },
    {
      "epoch": 225.29,
      "grad_norm": 10.756891250610352,
      "learning_rate": 2.4904904904904907e-06,
      "loss": 0.163,
      "step": 75020
    },
    {
      "epoch": 225.32,
      "grad_norm": 9.653883934020996,
      "learning_rate": 2.4894894894894897e-06,
      "loss": 0.2306,
      "step": 75030
    },
    {
      "epoch": 225.35,
      "grad_norm": 15.852289199829102,
      "learning_rate": 2.4884884884884887e-06,
      "loss": 0.2138,
      "step": 75040
    },
    {
      "epoch": 225.38,
      "grad_norm": 8.68993091583252,
      "learning_rate": 2.4874874874874877e-06,
      "loss": 0.1741,
      "step": 75050
    },
    {
      "epoch": 225.41,
      "grad_norm": 11.190282821655273,
      "learning_rate": 2.4864864864864867e-06,
      "loss": 0.1953,
      "step": 75060
    },
    {
      "epoch": 225.44,
      "grad_norm": 8.548380851745605,
      "learning_rate": 2.4854854854854858e-06,
      "loss": 0.1829,
      "step": 75070
    },
    {
      "epoch": 225.47,
      "grad_norm": 14.41917896270752,
      "learning_rate": 2.4844844844844848e-06,
      "loss": 0.2164,
      "step": 75080
    },
    {
      "epoch": 225.5,
      "grad_norm": 8.500130653381348,
      "learning_rate": 2.483483483483484e-06,
      "loss": 0.2111,
      "step": 75090
    },
    {
      "epoch": 225.53,
      "grad_norm": 5.746265411376953,
      "learning_rate": 2.482482482482483e-06,
      "loss": 0.1863,
      "step": 75100
    },
    {
      "epoch": 225.56,
      "grad_norm": 7.01035213470459,
      "learning_rate": 2.481481481481482e-06,
      "loss": 0.2053,
      "step": 75110
    },
    {
      "epoch": 225.59,
      "grad_norm": 20.22051429748535,
      "learning_rate": 2.480480480480481e-06,
      "loss": 0.1925,
      "step": 75120
    },
    {
      "epoch": 225.62,
      "grad_norm": 10.756665229797363,
      "learning_rate": 2.4794794794794794e-06,
      "loss": 0.1749,
      "step": 75130
    },
    {
      "epoch": 225.65,
      "grad_norm": 11.924793243408203,
      "learning_rate": 2.478478478478479e-06,
      "loss": 0.201,
      "step": 75140
    },
    {
      "epoch": 225.68,
      "grad_norm": 16.768310546875,
      "learning_rate": 2.4774774774774775e-06,
      "loss": 0.183,
      "step": 75150
    },
    {
      "epoch": 225.71,
      "grad_norm": 13.273860931396484,
      "learning_rate": 2.476476476476477e-06,
      "loss": 0.2076,
      "step": 75160
    },
    {
      "epoch": 225.74,
      "grad_norm": 7.01507043838501,
      "learning_rate": 2.4754754754754755e-06,
      "loss": 0.1879,
      "step": 75170
    },
    {
      "epoch": 225.77,
      "grad_norm": 8.41706371307373,
      "learning_rate": 2.474474474474475e-06,
      "loss": 0.1861,
      "step": 75180
    },
    {
      "epoch": 225.8,
      "grad_norm": 5.684409141540527,
      "learning_rate": 2.4734734734734736e-06,
      "loss": 0.2085,
      "step": 75190
    },
    {
      "epoch": 225.83,
      "grad_norm": 11.462095260620117,
      "learning_rate": 2.4724724724724726e-06,
      "loss": 0.1969,
      "step": 75200
    },
    {
      "epoch": 225.86,
      "grad_norm": 8.477944374084473,
      "learning_rate": 2.4714714714714716e-06,
      "loss": 0.2086,
      "step": 75210
    },
    {
      "epoch": 225.89,
      "grad_norm": 9.354867935180664,
      "learning_rate": 2.4704704704704706e-06,
      "loss": 0.1644,
      "step": 75220
    },
    {
      "epoch": 225.92,
      "grad_norm": 7.556333541870117,
      "learning_rate": 2.4694694694694696e-06,
      "loss": 0.2,
      "step": 75230
    },
    {
      "epoch": 225.95,
      "grad_norm": 8.64382266998291,
      "learning_rate": 2.4684684684684686e-06,
      "loss": 0.2295,
      "step": 75240
    },
    {
      "epoch": 225.98,
      "grad_norm": 6.866322040557861,
      "learning_rate": 2.4674674674674677e-06,
      "loss": 0.185,
      "step": 75250
    },
    {
      "epoch": 226.0,
      "eval_accuracy": 0.9229,
      "eval_loss": 0.34693443775177,
      "eval_runtime": 29.9308,
      "eval_samples_per_second": 334.104,
      "eval_steps_per_second": 1.336,
      "step": 75258
    },
    {
      "epoch": 226.01,
      "grad_norm": 4.460969924926758,
      "learning_rate": 2.4664664664664667e-06,
      "loss": 0.2253,
      "step": 75260
    },
    {
      "epoch": 226.04,
      "grad_norm": 34.26120376586914,
      "learning_rate": 2.4654654654654657e-06,
      "loss": 0.1945,
      "step": 75270
    },
    {
      "epoch": 226.07,
      "grad_norm": 3.8135986328125,
      "learning_rate": 2.4644644644644647e-06,
      "loss": 0.1929,
      "step": 75280
    },
    {
      "epoch": 226.1,
      "grad_norm": 6.226556301116943,
      "learning_rate": 2.4634634634634637e-06,
      "loss": 0.1734,
      "step": 75290
    },
    {
      "epoch": 226.13,
      "grad_norm": 8.38015365600586,
      "learning_rate": 2.4624624624624628e-06,
      "loss": 0.2235,
      "step": 75300
    },
    {
      "epoch": 226.16,
      "grad_norm": 9.834549903869629,
      "learning_rate": 2.4614614614614614e-06,
      "loss": 0.2026,
      "step": 75310
    },
    {
      "epoch": 226.19,
      "grad_norm": 23.953840255737305,
      "learning_rate": 2.460460460460461e-06,
      "loss": 0.2253,
      "step": 75320
    },
    {
      "epoch": 226.22,
      "grad_norm": 11.594855308532715,
      "learning_rate": 2.45945945945946e-06,
      "loss": 0.2336,
      "step": 75330
    },
    {
      "epoch": 226.25,
      "grad_norm": 12.098176956176758,
      "learning_rate": 2.458458458458459e-06,
      "loss": 0.2257,
      "step": 75340
    },
    {
      "epoch": 226.28,
      "grad_norm": 11.029435157775879,
      "learning_rate": 2.457457457457458e-06,
      "loss": 0.2119,
      "step": 75350
    },
    {
      "epoch": 226.31,
      "grad_norm": 11.778158187866211,
      "learning_rate": 2.4564564564564564e-06,
      "loss": 0.2158,
      "step": 75360
    },
    {
      "epoch": 226.34,
      "grad_norm": 14.672806739807129,
      "learning_rate": 2.455455455455456e-06,
      "loss": 0.2333,
      "step": 75370
    },
    {
      "epoch": 226.37,
      "grad_norm": 6.195869445800781,
      "learning_rate": 2.4544544544544545e-06,
      "loss": 0.1595,
      "step": 75380
    },
    {
      "epoch": 226.4,
      "grad_norm": 7.25065279006958,
      "learning_rate": 2.453453453453454e-06,
      "loss": 0.1847,
      "step": 75390
    },
    {
      "epoch": 226.43,
      "grad_norm": 11.593536376953125,
      "learning_rate": 2.4524524524524525e-06,
      "loss": 0.1935,
      "step": 75400
    },
    {
      "epoch": 226.46,
      "grad_norm": 16.819974899291992,
      "learning_rate": 2.451451451451452e-06,
      "loss": 0.186,
      "step": 75410
    },
    {
      "epoch": 226.49,
      "grad_norm": 4.765963554382324,
      "learning_rate": 2.4504504504504506e-06,
      "loss": 0.1994,
      "step": 75420
    },
    {
      "epoch": 226.52,
      "grad_norm": 8.2737398147583,
      "learning_rate": 2.4494494494494496e-06,
      "loss": 0.182,
      "step": 75430
    },
    {
      "epoch": 226.55,
      "grad_norm": 19.161453247070312,
      "learning_rate": 2.4484484484484486e-06,
      "loss": 0.2077,
      "step": 75440
    },
    {
      "epoch": 226.58,
      "grad_norm": 7.401142120361328,
      "learning_rate": 2.4474474474474476e-06,
      "loss": 0.1637,
      "step": 75450
    },
    {
      "epoch": 226.61,
      "grad_norm": 7.793619155883789,
      "learning_rate": 2.4464464464464466e-06,
      "loss": 0.1721,
      "step": 75460
    },
    {
      "epoch": 226.64,
      "grad_norm": 8.290586471557617,
      "learning_rate": 2.4454454454454456e-06,
      "loss": 0.1945,
      "step": 75470
    },
    {
      "epoch": 226.67,
      "grad_norm": 6.1472907066345215,
      "learning_rate": 2.4444444444444447e-06,
      "loss": 0.2398,
      "step": 75480
    },
    {
      "epoch": 226.7,
      "grad_norm": 5.70947265625,
      "learning_rate": 2.4434434434434437e-06,
      "loss": 0.1926,
      "step": 75490
    },
    {
      "epoch": 226.73,
      "grad_norm": 17.208721160888672,
      "learning_rate": 2.4424424424424427e-06,
      "loss": 0.1918,
      "step": 75500
    },
    {
      "epoch": 226.76,
      "grad_norm": 16.1998233795166,
      "learning_rate": 2.4414414414414417e-06,
      "loss": 0.1829,
      "step": 75510
    },
    {
      "epoch": 226.79,
      "grad_norm": 8.930643081665039,
      "learning_rate": 2.4404404404404407e-06,
      "loss": 0.2076,
      "step": 75520
    },
    {
      "epoch": 226.82,
      "grad_norm": 11.398575782775879,
      "learning_rate": 2.4394394394394398e-06,
      "loss": 0.1859,
      "step": 75530
    },
    {
      "epoch": 226.85,
      "grad_norm": 11.469947814941406,
      "learning_rate": 2.4384384384384383e-06,
      "loss": 0.2028,
      "step": 75540
    },
    {
      "epoch": 226.88,
      "grad_norm": 7.568595886230469,
      "learning_rate": 2.437437437437438e-06,
      "loss": 0.1968,
      "step": 75550
    },
    {
      "epoch": 226.91,
      "grad_norm": 9.093255996704102,
      "learning_rate": 2.4364364364364364e-06,
      "loss": 0.2026,
      "step": 75560
    },
    {
      "epoch": 226.94,
      "grad_norm": 7.9535722732543945,
      "learning_rate": 2.435435435435436e-06,
      "loss": 0.2025,
      "step": 75570
    },
    {
      "epoch": 226.97,
      "grad_norm": 7.016534805297852,
      "learning_rate": 2.4344344344344344e-06,
      "loss": 0.217,
      "step": 75580
    },
    {
      "epoch": 227.0,
      "grad_norm": 8.68912410736084,
      "learning_rate": 2.4334334334334334e-06,
      "loss": 0.2079,
      "step": 75590
    },
    {
      "epoch": 227.0,
      "eval_accuracy": 0.9228,
      "eval_loss": 0.3464837670326233,
      "eval_runtime": 30.4365,
      "eval_samples_per_second": 328.553,
      "eval_steps_per_second": 1.314,
      "step": 75591
    },
    {
      "epoch": 227.03,
      "grad_norm": 4.821264266967773,
      "learning_rate": 2.432432432432433e-06,
      "loss": 0.1874,
      "step": 75600
    },
    {
      "epoch": 227.06,
      "grad_norm": 15.190011978149414,
      "learning_rate": 2.4314314314314315e-06,
      "loss": 0.1862,
      "step": 75610
    },
    {
      "epoch": 227.09,
      "grad_norm": 11.231321334838867,
      "learning_rate": 2.430430430430431e-06,
      "loss": 0.1927,
      "step": 75620
    },
    {
      "epoch": 227.12,
      "grad_norm": 9.437117576599121,
      "learning_rate": 2.4294294294294295e-06,
      "loss": 0.1944,
      "step": 75630
    },
    {
      "epoch": 227.15,
      "grad_norm": 4.821615695953369,
      "learning_rate": 2.4284284284284285e-06,
      "loss": 0.1929,
      "step": 75640
    },
    {
      "epoch": 227.18,
      "grad_norm": 27.8004150390625,
      "learning_rate": 2.4274274274274276e-06,
      "loss": 0.1895,
      "step": 75650
    },
    {
      "epoch": 227.21,
      "grad_norm": 17.749380111694336,
      "learning_rate": 2.4264264264264266e-06,
      "loss": 0.206,
      "step": 75660
    },
    {
      "epoch": 227.24,
      "grad_norm": 14.908825874328613,
      "learning_rate": 2.4254254254254256e-06,
      "loss": 0.1993,
      "step": 75670
    },
    {
      "epoch": 227.27,
      "grad_norm": 11.08859634399414,
      "learning_rate": 2.4244244244244246e-06,
      "loss": 0.1926,
      "step": 75680
    },
    {
      "epoch": 227.3,
      "grad_norm": 13.444189071655273,
      "learning_rate": 2.4234234234234236e-06,
      "loss": 0.1676,
      "step": 75690
    },
    {
      "epoch": 227.33,
      "grad_norm": 7.276403903961182,
      "learning_rate": 2.4224224224224226e-06,
      "loss": 0.1999,
      "step": 75700
    },
    {
      "epoch": 227.36,
      "grad_norm": 4.481171607971191,
      "learning_rate": 2.4214214214214217e-06,
      "loss": 0.2133,
      "step": 75710
    },
    {
      "epoch": 227.39,
      "grad_norm": 15.2894287109375,
      "learning_rate": 2.4204204204204207e-06,
      "loss": 0.2031,
      "step": 75720
    },
    {
      "epoch": 227.42,
      "grad_norm": 12.393250465393066,
      "learning_rate": 2.4194194194194197e-06,
      "loss": 0.2253,
      "step": 75730
    },
    {
      "epoch": 227.45,
      "grad_norm": 10.808647155761719,
      "learning_rate": 2.4184184184184187e-06,
      "loss": 0.1857,
      "step": 75740
    },
    {
      "epoch": 227.48,
      "grad_norm": 8.543875694274902,
      "learning_rate": 2.4174174174174177e-06,
      "loss": 0.2129,
      "step": 75750
    },
    {
      "epoch": 227.51,
      "grad_norm": 14.241586685180664,
      "learning_rate": 2.4164164164164168e-06,
      "loss": 0.1738,
      "step": 75760
    },
    {
      "epoch": 227.54,
      "grad_norm": 13.835046768188477,
      "learning_rate": 2.4154154154154153e-06,
      "loss": 0.1983,
      "step": 75770
    },
    {
      "epoch": 227.57,
      "grad_norm": 8.650141716003418,
      "learning_rate": 2.414414414414415e-06,
      "loss": 0.2211,
      "step": 75780
    },
    {
      "epoch": 227.6,
      "grad_norm": 7.608523368835449,
      "learning_rate": 2.4134134134134134e-06,
      "loss": 0.2373,
      "step": 75790
    },
    {
      "epoch": 227.63,
      "grad_norm": 7.987109184265137,
      "learning_rate": 2.412412412412413e-06,
      "loss": 0.204,
      "step": 75800
    },
    {
      "epoch": 227.66,
      "grad_norm": 8.824383735656738,
      "learning_rate": 2.4114114114114114e-06,
      "loss": 0.2418,
      "step": 75810
    },
    {
      "epoch": 227.69,
      "grad_norm": 9.029004096984863,
      "learning_rate": 2.4104104104104104e-06,
      "loss": 0.1662,
      "step": 75820
    },
    {
      "epoch": 227.72,
      "grad_norm": 22.121503829956055,
      "learning_rate": 2.4094094094094095e-06,
      "loss": 0.2092,
      "step": 75830
    },
    {
      "epoch": 227.75,
      "grad_norm": 6.722419261932373,
      "learning_rate": 2.4084084084084085e-06,
      "loss": 0.2154,
      "step": 75840
    },
    {
      "epoch": 227.78,
      "grad_norm": 8.044062614440918,
      "learning_rate": 2.4074074074074075e-06,
      "loss": 0.174,
      "step": 75850
    },
    {
      "epoch": 227.81,
      "grad_norm": 6.550583362579346,
      "learning_rate": 2.4064064064064065e-06,
      "loss": 0.1976,
      "step": 75860
    },
    {
      "epoch": 227.84,
      "grad_norm": 13.887739181518555,
      "learning_rate": 2.4054054054054055e-06,
      "loss": 0.1887,
      "step": 75870
    },
    {
      "epoch": 227.87,
      "grad_norm": 6.07134485244751,
      "learning_rate": 2.4044044044044045e-06,
      "loss": 0.168,
      "step": 75880
    },
    {
      "epoch": 227.9,
      "grad_norm": 7.120780944824219,
      "learning_rate": 2.4034034034034036e-06,
      "loss": 0.2075,
      "step": 75890
    },
    {
      "epoch": 227.93,
      "grad_norm": 9.222057342529297,
      "learning_rate": 2.4024024024024026e-06,
      "loss": 0.1869,
      "step": 75900
    },
    {
      "epoch": 227.96,
      "grad_norm": 12.663531303405762,
      "learning_rate": 2.4014014014014016e-06,
      "loss": 0.1884,
      "step": 75910
    },
    {
      "epoch": 227.99,
      "grad_norm": 12.278611183166504,
      "learning_rate": 2.4004004004004006e-06,
      "loss": 0.1709,
      "step": 75920
    },
    {
      "epoch": 228.0,
      "eval_accuracy": 0.9217,
      "eval_loss": 0.34851327538490295,
      "eval_runtime": 30.3497,
      "eval_samples_per_second": 329.492,
      "eval_steps_per_second": 1.318,
      "step": 75924
    },
    {
      "epoch": 228.02,
      "grad_norm": 5.223087310791016,
      "learning_rate": 2.3993993993993996e-06,
      "loss": 0.2044,
      "step": 75930
    },
    {
      "epoch": 228.05,
      "grad_norm": 8.285252571105957,
      "learning_rate": 2.3983983983983987e-06,
      "loss": 0.2431,
      "step": 75940
    },
    {
      "epoch": 228.08,
      "grad_norm": 8.095876693725586,
      "learning_rate": 2.3973973973973977e-06,
      "loss": 0.1859,
      "step": 75950
    },
    {
      "epoch": 228.11,
      "grad_norm": 7.669524192810059,
      "learning_rate": 2.3963963963963967e-06,
      "loss": 0.1755,
      "step": 75960
    },
    {
      "epoch": 228.14,
      "grad_norm": 7.505213737487793,
      "learning_rate": 2.3953953953953957e-06,
      "loss": 0.1824,
      "step": 75970
    },
    {
      "epoch": 228.17,
      "grad_norm": 19.451074600219727,
      "learning_rate": 2.3943943943943947e-06,
      "loss": 0.1883,
      "step": 75980
    },
    {
      "epoch": 228.2,
      "grad_norm": 7.598684787750244,
      "learning_rate": 2.3933933933933938e-06,
      "loss": 0.1986,
      "step": 75990
    },
    {
      "epoch": 228.23,
      "grad_norm": 8.656598091125488,
      "learning_rate": 2.3923923923923923e-06,
      "loss": 0.1808,
      "step": 76000
    },
    {
      "epoch": 228.26,
      "grad_norm": 8.365595817565918,
      "learning_rate": 2.3913913913913918e-06,
      "loss": 0.2053,
      "step": 76010
    },
    {
      "epoch": 228.29,
      "grad_norm": 10.554317474365234,
      "learning_rate": 2.3903903903903904e-06,
      "loss": 0.1904,
      "step": 76020
    },
    {
      "epoch": 228.32,
      "grad_norm": 7.168965816497803,
      "learning_rate": 2.38938938938939e-06,
      "loss": 0.1655,
      "step": 76030
    },
    {
      "epoch": 228.35,
      "grad_norm": 6.671514511108398,
      "learning_rate": 2.3883883883883884e-06,
      "loss": 0.1948,
      "step": 76040
    },
    {
      "epoch": 228.38,
      "grad_norm": 6.270973205566406,
      "learning_rate": 2.3873873873873874e-06,
      "loss": 0.2226,
      "step": 76050
    },
    {
      "epoch": 228.41,
      "grad_norm": 7.213583469390869,
      "learning_rate": 2.3863863863863865e-06,
      "loss": 0.1716,
      "step": 76060
    },
    {
      "epoch": 228.44,
      "grad_norm": 7.272107124328613,
      "learning_rate": 2.3853853853853855e-06,
      "loss": 0.1698,
      "step": 76070
    },
    {
      "epoch": 228.47,
      "grad_norm": 11.273940086364746,
      "learning_rate": 2.3843843843843845e-06,
      "loss": 0.1713,
      "step": 76080
    },
    {
      "epoch": 228.5,
      "grad_norm": 7.803356647491455,
      "learning_rate": 2.3833833833833835e-06,
      "loss": 0.2088,
      "step": 76090
    },
    {
      "epoch": 228.53,
      "grad_norm": 8.163289070129395,
      "learning_rate": 2.3823823823823825e-06,
      "loss": 0.1881,
      "step": 76100
    },
    {
      "epoch": 228.56,
      "grad_norm": 12.534679412841797,
      "learning_rate": 2.3813813813813815e-06,
      "loss": 0.1767,
      "step": 76110
    },
    {
      "epoch": 228.59,
      "grad_norm": 8.930296897888184,
      "learning_rate": 2.3803803803803806e-06,
      "loss": 0.2255,
      "step": 76120
    },
    {
      "epoch": 228.62,
      "grad_norm": 4.730624198913574,
      "learning_rate": 2.3793793793793796e-06,
      "loss": 0.1634,
      "step": 76130
    },
    {
      "epoch": 228.65,
      "grad_norm": 10.565033912658691,
      "learning_rate": 2.3783783783783786e-06,
      "loss": 0.1966,
      "step": 76140
    },
    {
      "epoch": 228.68,
      "grad_norm": 3.681523084640503,
      "learning_rate": 2.3773773773773776e-06,
      "loss": 0.2032,
      "step": 76150
    },
    {
      "epoch": 228.71,
      "grad_norm": 5.887136936187744,
      "learning_rate": 2.3763763763763766e-06,
      "loss": 0.1599,
      "step": 76160
    },
    {
      "epoch": 228.74,
      "grad_norm": 8.69440746307373,
      "learning_rate": 2.3753753753753757e-06,
      "loss": 0.1843,
      "step": 76170
    },
    {
      "epoch": 228.77,
      "grad_norm": 8.567039489746094,
      "learning_rate": 2.3743743743743747e-06,
      "loss": 0.1832,
      "step": 76180
    },
    {
      "epoch": 228.8,
      "grad_norm": 6.169824600219727,
      "learning_rate": 2.3733733733733737e-06,
      "loss": 0.1918,
      "step": 76190
    },
    {
      "epoch": 228.83,
      "grad_norm": 5.732800483703613,
      "learning_rate": 2.3723723723723727e-06,
      "loss": 0.2272,
      "step": 76200
    },
    {
      "epoch": 228.86,
      "grad_norm": 6.58015775680542,
      "learning_rate": 2.3713713713713717e-06,
      "loss": 0.1984,
      "step": 76210
    },
    {
      "epoch": 228.89,
      "grad_norm": 7.364563465118408,
      "learning_rate": 2.3703703703703707e-06,
      "loss": 0.2104,
      "step": 76220
    },
    {
      "epoch": 228.92,
      "grad_norm": 11.298601150512695,
      "learning_rate": 2.3693693693693693e-06,
      "loss": 0.1797,
      "step": 76230
    },
    {
      "epoch": 228.95,
      "grad_norm": 12.94999885559082,
      "learning_rate": 2.3683683683683688e-06,
      "loss": 0.1823,
      "step": 76240
    },
    {
      "epoch": 228.98,
      "grad_norm": 9.011898040771484,
      "learning_rate": 2.3673673673673674e-06,
      "loss": 0.2041,
      "step": 76250
    },
    {
      "epoch": 229.0,
      "eval_accuracy": 0.9213,
      "eval_loss": 0.34760624170303345,
      "eval_runtime": 30.2467,
      "eval_samples_per_second": 330.615,
      "eval_steps_per_second": 1.322,
      "step": 76257
    },
    {
      "epoch": 229.01,
      "grad_norm": 7.73806095123291,
      "learning_rate": 2.366366366366367e-06,
      "loss": 0.1723,
      "step": 76260
    },
    {
      "epoch": 229.04,
      "grad_norm": 7.611754894256592,
      "learning_rate": 2.3653653653653654e-06,
      "loss": 0.1777,
      "step": 76270
    },
    {
      "epoch": 229.07,
      "grad_norm": 13.91844367980957,
      "learning_rate": 2.3643643643643644e-06,
      "loss": 0.203,
      "step": 76280
    },
    {
      "epoch": 229.1,
      "grad_norm": 8.50583553314209,
      "learning_rate": 2.3633633633633635e-06,
      "loss": 0.1651,
      "step": 76290
    },
    {
      "epoch": 229.13,
      "grad_norm": 8.610441207885742,
      "learning_rate": 2.3623623623623625e-06,
      "loss": 0.2142,
      "step": 76300
    },
    {
      "epoch": 229.16,
      "grad_norm": 9.762157440185547,
      "learning_rate": 2.3613613613613615e-06,
      "loss": 0.1605,
      "step": 76310
    },
    {
      "epoch": 229.19,
      "grad_norm": 9.749794960021973,
      "learning_rate": 2.3603603603603605e-06,
      "loss": 0.1806,
      "step": 76320
    },
    {
      "epoch": 229.22,
      "grad_norm": 14.731498718261719,
      "learning_rate": 2.3593593593593595e-06,
      "loss": 0.1974,
      "step": 76330
    },
    {
      "epoch": 229.25,
      "grad_norm": 27.00509262084961,
      "learning_rate": 2.3583583583583585e-06,
      "loss": 0.2046,
      "step": 76340
    },
    {
      "epoch": 229.28,
      "grad_norm": 8.396903038024902,
      "learning_rate": 2.3573573573573576e-06,
      "loss": 0.1851,
      "step": 76350
    },
    {
      "epoch": 229.31,
      "grad_norm": 8.726446151733398,
      "learning_rate": 2.3563563563563566e-06,
      "loss": 0.1788,
      "step": 76360
    },
    {
      "epoch": 229.34,
      "grad_norm": 7.764161586761475,
      "learning_rate": 2.3553553553553556e-06,
      "loss": 0.24,
      "step": 76370
    },
    {
      "epoch": 229.37,
      "grad_norm": 9.18239974975586,
      "learning_rate": 2.3543543543543546e-06,
      "loss": 0.2097,
      "step": 76380
    },
    {
      "epoch": 229.4,
      "grad_norm": 6.925025463104248,
      "learning_rate": 2.3533533533533532e-06,
      "loss": 0.1454,
      "step": 76390
    },
    {
      "epoch": 229.43,
      "grad_norm": 9.61048412322998,
      "learning_rate": 2.3523523523523527e-06,
      "loss": 0.2208,
      "step": 76400
    },
    {
      "epoch": 229.46,
      "grad_norm": 7.273800373077393,
      "learning_rate": 2.3513513513513517e-06,
      "loss": 0.2377,
      "step": 76410
    },
    {
      "epoch": 229.49,
      "grad_norm": 5.806123733520508,
      "learning_rate": 2.3503503503503507e-06,
      "loss": 0.1636,
      "step": 76420
    },
    {
      "epoch": 229.52,
      "grad_norm": 6.21291446685791,
      "learning_rate": 2.3493493493493497e-06,
      "loss": 0.201,
      "step": 76430
    },
    {
      "epoch": 229.55,
      "grad_norm": 9.995850563049316,
      "learning_rate": 2.3483483483483483e-06,
      "loss": 0.1662,
      "step": 76440
    },
    {
      "epoch": 229.58,
      "grad_norm": 7.633752346038818,
      "learning_rate": 2.3473473473473477e-06,
      "loss": 0.2043,
      "step": 76450
    },
    {
      "epoch": 229.61,
      "grad_norm": 8.02734088897705,
      "learning_rate": 2.3463463463463463e-06,
      "loss": 0.1643,
      "step": 76460
    },
    {
      "epoch": 229.64,
      "grad_norm": 9.1734037399292,
      "learning_rate": 2.3453453453453458e-06,
      "loss": 0.2101,
      "step": 76470
    },
    {
      "epoch": 229.67,
      "grad_norm": 8.150520324707031,
      "learning_rate": 2.3443443443443444e-06,
      "loss": 0.1845,
      "step": 76480
    },
    {
      "epoch": 229.7,
      "grad_norm": 9.137328147888184,
      "learning_rate": 2.343343343343344e-06,
      "loss": 0.1829,
      "step": 76490
    },
    {
      "epoch": 229.73,
      "grad_norm": 8.66564655303955,
      "learning_rate": 2.3423423423423424e-06,
      "loss": 0.2263,
      "step": 76500
    },
    {
      "epoch": 229.76,
      "grad_norm": 8.800448417663574,
      "learning_rate": 2.3413413413413414e-06,
      "loss": 0.1801,
      "step": 76510
    },
    {
      "epoch": 229.79,
      "grad_norm": 7.406792640686035,
      "learning_rate": 2.3403403403403405e-06,
      "loss": 0.1989,
      "step": 76520
    },
    {
      "epoch": 229.82,
      "grad_norm": 5.5974555015563965,
      "learning_rate": 2.3393393393393395e-06,
      "loss": 0.155,
      "step": 76530
    },
    {
      "epoch": 229.85,
      "grad_norm": 14.025899887084961,
      "learning_rate": 2.3383383383383385e-06,
      "loss": 0.1929,
      "step": 76540
    },
    {
      "epoch": 229.88,
      "grad_norm": 9.799904823303223,
      "learning_rate": 2.3373373373373375e-06,
      "loss": 0.1797,
      "step": 76550
    },
    {
      "epoch": 229.91,
      "grad_norm": 7.078367710113525,
      "learning_rate": 2.3363363363363365e-06,
      "loss": 0.1537,
      "step": 76560
    },
    {
      "epoch": 229.94,
      "grad_norm": 10.776582717895508,
      "learning_rate": 2.3353353353353355e-06,
      "loss": 0.2054,
      "step": 76570
    },
    {
      "epoch": 229.97,
      "grad_norm": 3.896117925643921,
      "learning_rate": 2.3343343343343346e-06,
      "loss": 0.2628,
      "step": 76580
    },
    {
      "epoch": 230.0,
      "grad_norm": 0.02343888022005558,
      "learning_rate": 2.3333333333333336e-06,
      "loss": 0.1793,
      "step": 76590
    },
    {
      "epoch": 230.0,
      "eval_accuracy": 0.9222,
      "eval_loss": 0.3505186140537262,
      "eval_runtime": 30.2858,
      "eval_samples_per_second": 330.188,
      "eval_steps_per_second": 1.321,
      "step": 76590
    },
    {
      "epoch": 230.03,
      "grad_norm": 5.371084213256836,
      "learning_rate": 2.3323323323323326e-06,
      "loss": 0.1515,
      "step": 76600
    },
    {
      "epoch": 230.06,
      "grad_norm": 4.363478660583496,
      "learning_rate": 2.3313313313313316e-06,
      "loss": 0.184,
      "step": 76610
    },
    {
      "epoch": 230.09,
      "grad_norm": 17.097923278808594,
      "learning_rate": 2.3303303303303302e-06,
      "loss": 0.205,
      "step": 76620
    },
    {
      "epoch": 230.12,
      "grad_norm": 5.364786624908447,
      "learning_rate": 2.3293293293293297e-06,
      "loss": 0.2457,
      "step": 76630
    },
    {
      "epoch": 230.15,
      "grad_norm": 10.032120704650879,
      "learning_rate": 2.3283283283283282e-06,
      "loss": 0.1804,
      "step": 76640
    },
    {
      "epoch": 230.18,
      "grad_norm": 6.185202598571777,
      "learning_rate": 2.3273273273273277e-06,
      "loss": 0.1564,
      "step": 76650
    },
    {
      "epoch": 230.21,
      "grad_norm": 9.38698959350586,
      "learning_rate": 2.3263263263263263e-06,
      "loss": 0.1699,
      "step": 76660
    },
    {
      "epoch": 230.24,
      "grad_norm": 11.390268325805664,
      "learning_rate": 2.3253253253253253e-06,
      "loss": 0.165,
      "step": 76670
    },
    {
      "epoch": 230.27,
      "grad_norm": 11.898106575012207,
      "learning_rate": 2.3243243243243247e-06,
      "loss": 0.1923,
      "step": 76680
    },
    {
      "epoch": 230.3,
      "grad_norm": 12.552009582519531,
      "learning_rate": 2.3233233233233233e-06,
      "loss": 0.1812,
      "step": 76690
    },
    {
      "epoch": 230.33,
      "grad_norm": 11.229490280151367,
      "learning_rate": 2.3223223223223228e-06,
      "loss": 0.1996,
      "step": 76700
    },
    {
      "epoch": 230.36,
      "grad_norm": 11.644991874694824,
      "learning_rate": 2.3213213213213214e-06,
      "loss": 0.1594,
      "step": 76710
    },
    {
      "epoch": 230.39,
      "grad_norm": 8.74962043762207,
      "learning_rate": 2.320320320320321e-06,
      "loss": 0.1847,
      "step": 76720
    },
    {
      "epoch": 230.42,
      "grad_norm": 6.968712329864502,
      "learning_rate": 2.3193193193193194e-06,
      "loss": 0.1888,
      "step": 76730
    },
    {
      "epoch": 230.45,
      "grad_norm": 9.760202407836914,
      "learning_rate": 2.3183183183183184e-06,
      "loss": 0.2048,
      "step": 76740
    },
    {
      "epoch": 230.48,
      "grad_norm": 5.242031574249268,
      "learning_rate": 2.3173173173173175e-06,
      "loss": 0.2189,
      "step": 76750
    },
    {
      "epoch": 230.51,
      "grad_norm": 8.781752586364746,
      "learning_rate": 2.3163163163163165e-06,
      "loss": 0.1763,
      "step": 76760
    },
    {
      "epoch": 230.54,
      "grad_norm": 16.051973342895508,
      "learning_rate": 2.3153153153153155e-06,
      "loss": 0.1493,
      "step": 76770
    },
    {
      "epoch": 230.57,
      "grad_norm": 5.0672078132629395,
      "learning_rate": 2.3143143143143145e-06,
      "loss": 0.1894,
      "step": 76780
    },
    {
      "epoch": 230.6,
      "grad_norm": 9.2982816696167,
      "learning_rate": 2.3133133133133135e-06,
      "loss": 0.1927,
      "step": 76790
    },
    {
      "epoch": 230.63,
      "grad_norm": 9.152308464050293,
      "learning_rate": 2.3123123123123125e-06,
      "loss": 0.202,
      "step": 76800
    },
    {
      "epoch": 230.66,
      "grad_norm": 10.022770881652832,
      "learning_rate": 2.3113113113113116e-06,
      "loss": 0.2073,
      "step": 76810
    },
    {
      "epoch": 230.69,
      "grad_norm": 6.455334663391113,
      "learning_rate": 2.3103103103103106e-06,
      "loss": 0.1963,
      "step": 76820
    },
    {
      "epoch": 230.72,
      "grad_norm": 9.132991790771484,
      "learning_rate": 2.3093093093093096e-06,
      "loss": 0.2532,
      "step": 76830
    },
    {
      "epoch": 230.75,
      "grad_norm": 7.9781880378723145,
      "learning_rate": 2.3083083083083086e-06,
      "loss": 0.2099,
      "step": 76840
    },
    {
      "epoch": 230.78,
      "grad_norm": 6.867335319519043,
      "learning_rate": 2.307307307307307e-06,
      "loss": 0.1839,
      "step": 76850
    },
    {
      "epoch": 230.81,
      "grad_norm": 5.954765319824219,
      "learning_rate": 2.3063063063063067e-06,
      "loss": 0.2193,
      "step": 76860
    },
    {
      "epoch": 230.84,
      "grad_norm": 10.151122093200684,
      "learning_rate": 2.3053053053053052e-06,
      "loss": 0.2455,
      "step": 76870
    },
    {
      "epoch": 230.87,
      "grad_norm": 9.759339332580566,
      "learning_rate": 2.3043043043043047e-06,
      "loss": 0.1858,
      "step": 76880
    },
    {
      "epoch": 230.9,
      "grad_norm": 10.373539924621582,
      "learning_rate": 2.3033033033033033e-06,
      "loss": 0.1954,
      "step": 76890
    },
    {
      "epoch": 230.93,
      "grad_norm": 10.346749305725098,
      "learning_rate": 2.3023023023023023e-06,
      "loss": 0.2061,
      "step": 76900
    },
    {
      "epoch": 230.96,
      "grad_norm": 9.29416275024414,
      "learning_rate": 2.3013013013013013e-06,
      "loss": 0.1778,
      "step": 76910
    },
    {
      "epoch": 230.99,
      "grad_norm": 21.40296745300293,
      "learning_rate": 2.3003003003003003e-06,
      "loss": 0.193,
      "step": 76920
    },
    {
      "epoch": 231.0,
      "eval_accuracy": 0.9234,
      "eval_loss": 0.3482620120048523,
      "eval_runtime": 29.8165,
      "eval_samples_per_second": 335.384,
      "eval_steps_per_second": 1.342,
      "step": 76923
    },
    {
      "epoch": 231.02,
      "grad_norm": 9.03520393371582,
      "learning_rate": 2.2992992992992998e-06,
      "loss": 0.1526,
      "step": 76930
    },
    {
      "epoch": 231.05,
      "grad_norm": 7.7270073890686035,
      "learning_rate": 2.2982982982982984e-06,
      "loss": 0.1653,
      "step": 76940
    },
    {
      "epoch": 231.08,
      "grad_norm": 12.016252517700195,
      "learning_rate": 2.297297297297298e-06,
      "loss": 0.2171,
      "step": 76950
    },
    {
      "epoch": 231.11,
      "grad_norm": 9.437795639038086,
      "learning_rate": 2.2962962962962964e-06,
      "loss": 0.2004,
      "step": 76960
    },
    {
      "epoch": 231.14,
      "grad_norm": 8.55064582824707,
      "learning_rate": 2.2952952952952954e-06,
      "loss": 0.1766,
      "step": 76970
    },
    {
      "epoch": 231.17,
      "grad_norm": 23.18525505065918,
      "learning_rate": 2.2942942942942944e-06,
      "loss": 0.222,
      "step": 76980
    },
    {
      "epoch": 231.2,
      "grad_norm": 8.038446426391602,
      "learning_rate": 2.2932932932932935e-06,
      "loss": 0.2106,
      "step": 76990
    },
    {
      "epoch": 231.23,
      "grad_norm": 9.24230670928955,
      "learning_rate": 2.2922922922922925e-06,
      "loss": 0.1796,
      "step": 77000
    },
    {
      "epoch": 231.26,
      "grad_norm": 7.561718940734863,
      "learning_rate": 2.2912912912912915e-06,
      "loss": 0.1885,
      "step": 77010
    },
    {
      "epoch": 231.29,
      "grad_norm": 8.282840728759766,
      "learning_rate": 2.2902902902902905e-06,
      "loss": 0.2138,
      "step": 77020
    },
    {
      "epoch": 231.32,
      "grad_norm": 8.078354835510254,
      "learning_rate": 2.2892892892892895e-06,
      "loss": 0.2022,
      "step": 77030
    },
    {
      "epoch": 231.35,
      "grad_norm": 4.511499404907227,
      "learning_rate": 2.2882882882882886e-06,
      "loss": 0.2197,
      "step": 77040
    },
    {
      "epoch": 231.38,
      "grad_norm": 10.103486061096191,
      "learning_rate": 2.2872872872872876e-06,
      "loss": 0.199,
      "step": 77050
    },
    {
      "epoch": 231.41,
      "grad_norm": 7.792743682861328,
      "learning_rate": 2.2862862862862866e-06,
      "loss": 0.1653,
      "step": 77060
    },
    {
      "epoch": 231.44,
      "grad_norm": 3.5658984184265137,
      "learning_rate": 2.2852852852852856e-06,
      "loss": 0.1755,
      "step": 77070
    },
    {
      "epoch": 231.47,
      "grad_norm": 13.45706558227539,
      "learning_rate": 2.284284284284284e-06,
      "loss": 0.1852,
      "step": 77080
    },
    {
      "epoch": 231.5,
      "grad_norm": 5.999841213226318,
      "learning_rate": 2.2832832832832837e-06,
      "loss": 0.1731,
      "step": 77090
    },
    {
      "epoch": 231.53,
      "grad_norm": 4.771999835968018,
      "learning_rate": 2.2822822822822822e-06,
      "loss": 0.2093,
      "step": 77100
    },
    {
      "epoch": 231.56,
      "grad_norm": 10.13019847869873,
      "learning_rate": 2.2812812812812817e-06,
      "loss": 0.2139,
      "step": 77110
    },
    {
      "epoch": 231.59,
      "grad_norm": 5.555051803588867,
      "learning_rate": 2.2802802802802803e-06,
      "loss": 0.1902,
      "step": 77120
    },
    {
      "epoch": 231.62,
      "grad_norm": 7.312884330749512,
      "learning_rate": 2.2792792792792793e-06,
      "loss": 0.2077,
      "step": 77130
    },
    {
      "epoch": 231.65,
      "grad_norm": 10.161093711853027,
      "learning_rate": 2.2782782782782783e-06,
      "loss": 0.1801,
      "step": 77140
    },
    {
      "epoch": 231.68,
      "grad_norm": 5.468297004699707,
      "learning_rate": 2.2772772772772773e-06,
      "loss": 0.1424,
      "step": 77150
    },
    {
      "epoch": 231.71,
      "grad_norm": 7.211467266082764,
      "learning_rate": 2.2762762762762764e-06,
      "loss": 0.1894,
      "step": 77160
    },
    {
      "epoch": 231.74,
      "grad_norm": 6.839397430419922,
      "learning_rate": 2.2752752752752754e-06,
      "loss": 0.224,
      "step": 77170
    },
    {
      "epoch": 231.77,
      "grad_norm": 9.877079963684082,
      "learning_rate": 2.2742742742742744e-06,
      "loss": 0.2015,
      "step": 77180
    },
    {
      "epoch": 231.8,
      "grad_norm": 7.640049457550049,
      "learning_rate": 2.2732732732732734e-06,
      "loss": 0.2321,
      "step": 77190
    },
    {
      "epoch": 231.83,
      "grad_norm": 7.488707542419434,
      "learning_rate": 2.2722722722722724e-06,
      "loss": 0.1604,
      "step": 77200
    },
    {
      "epoch": 231.86,
      "grad_norm": 5.806524753570557,
      "learning_rate": 2.2712712712712714e-06,
      "loss": 0.172,
      "step": 77210
    },
    {
      "epoch": 231.89,
      "grad_norm": 9.178234100341797,
      "learning_rate": 2.2702702702702705e-06,
      "loss": 0.2289,
      "step": 77220
    },
    {
      "epoch": 231.92,
      "grad_norm": 8.437017440795898,
      "learning_rate": 2.2692692692692695e-06,
      "loss": 0.2068,
      "step": 77230
    },
    {
      "epoch": 231.95,
      "grad_norm": 5.434473037719727,
      "learning_rate": 2.2682682682682685e-06,
      "loss": 0.1796,
      "step": 77240
    },
    {
      "epoch": 231.98,
      "grad_norm": 14.791305541992188,
      "learning_rate": 2.2672672672672675e-06,
      "loss": 0.2192,
      "step": 77250
    },
    {
      "epoch": 232.0,
      "eval_accuracy": 0.9213,
      "eval_loss": 0.3489912152290344,
      "eval_runtime": 30.7282,
      "eval_samples_per_second": 325.434,
      "eval_steps_per_second": 1.302,
      "step": 77256
    },
    {
      "epoch": 232.01,
      "grad_norm": 8.292154312133789,
      "learning_rate": 2.2662662662662665e-06,
      "loss": 0.2211,
      "step": 77260
    },
    {
      "epoch": 232.04,
      "grad_norm": 10.725556373596191,
      "learning_rate": 2.2652652652652656e-06,
      "loss": 0.206,
      "step": 77270
    },
    {
      "epoch": 232.07,
      "grad_norm": 11.501164436340332,
      "learning_rate": 2.2642642642642646e-06,
      "loss": 0.2232,
      "step": 77280
    },
    {
      "epoch": 232.1,
      "grad_norm": 9.01240348815918,
      "learning_rate": 2.2632632632632636e-06,
      "loss": 0.1993,
      "step": 77290
    },
    {
      "epoch": 232.13,
      "grad_norm": 6.064018249511719,
      "learning_rate": 2.2622622622622626e-06,
      "loss": 0.2217,
      "step": 77300
    },
    {
      "epoch": 232.16,
      "grad_norm": 11.418902397155762,
      "learning_rate": 2.261261261261261e-06,
      "loss": 0.2212,
      "step": 77310
    },
    {
      "epoch": 232.19,
      "grad_norm": 8.471709251403809,
      "learning_rate": 2.2602602602602606e-06,
      "loss": 0.1948,
      "step": 77320
    },
    {
      "epoch": 232.22,
      "grad_norm": 5.993159294128418,
      "learning_rate": 2.2592592592592592e-06,
      "loss": 0.1838,
      "step": 77330
    },
    {
      "epoch": 232.25,
      "grad_norm": 8.245962142944336,
      "learning_rate": 2.2582582582582587e-06,
      "loss": 0.2055,
      "step": 77340
    },
    {
      "epoch": 232.28,
      "grad_norm": 6.173013687133789,
      "learning_rate": 2.2572572572572573e-06,
      "loss": 0.156,
      "step": 77350
    },
    {
      "epoch": 232.31,
      "grad_norm": 7.667869567871094,
      "learning_rate": 2.2562562562562563e-06,
      "loss": 0.2401,
      "step": 77360
    },
    {
      "epoch": 232.34,
      "grad_norm": 8.96622085571289,
      "learning_rate": 2.2552552552552553e-06,
      "loss": 0.1806,
      "step": 77370
    },
    {
      "epoch": 232.37,
      "grad_norm": 9.298490524291992,
      "learning_rate": 2.2542542542542543e-06,
      "loss": 0.2163,
      "step": 77380
    },
    {
      "epoch": 232.4,
      "grad_norm": 14.434346199035645,
      "learning_rate": 2.2532532532532534e-06,
      "loss": 0.1623,
      "step": 77390
    },
    {
      "epoch": 232.43,
      "grad_norm": 6.170042991638184,
      "learning_rate": 2.2522522522522524e-06,
      "loss": 0.1644,
      "step": 77400
    },
    {
      "epoch": 232.46,
      "grad_norm": 8.031316757202148,
      "learning_rate": 2.2512512512512514e-06,
      "loss": 0.2378,
      "step": 77410
    },
    {
      "epoch": 232.49,
      "grad_norm": 11.204882621765137,
      "learning_rate": 2.2502502502502504e-06,
      "loss": 0.1969,
      "step": 77420
    },
    {
      "epoch": 232.52,
      "grad_norm": 10.13510513305664,
      "learning_rate": 2.2492492492492494e-06,
      "loss": 0.2521,
      "step": 77430
    },
    {
      "epoch": 232.55,
      "grad_norm": 61.69583511352539,
      "learning_rate": 2.2482482482482484e-06,
      "loss": 0.1765,
      "step": 77440
    },
    {
      "epoch": 232.58,
      "grad_norm": 8.230854988098145,
      "learning_rate": 2.2472472472472475e-06,
      "loss": 0.1911,
      "step": 77450
    },
    {
      "epoch": 232.61,
      "grad_norm": 6.13785457611084,
      "learning_rate": 2.2462462462462465e-06,
      "loss": 0.1831,
      "step": 77460
    },
    {
      "epoch": 232.64,
      "grad_norm": 9.56834888458252,
      "learning_rate": 2.2452452452452455e-06,
      "loss": 0.1793,
      "step": 77470
    },
    {
      "epoch": 232.67,
      "grad_norm": 15.674004554748535,
      "learning_rate": 2.2442442442442445e-06,
      "loss": 0.1609,
      "step": 77480
    },
    {
      "epoch": 232.7,
      "grad_norm": 8.875380516052246,
      "learning_rate": 2.2432432432432435e-06,
      "loss": 0.1708,
      "step": 77490
    },
    {
      "epoch": 232.73,
      "grad_norm": 8.782155990600586,
      "learning_rate": 2.2422422422422426e-06,
      "loss": 0.1754,
      "step": 77500
    },
    {
      "epoch": 232.76,
      "grad_norm": 4.715075969696045,
      "learning_rate": 2.2412412412412416e-06,
      "loss": 0.1751,
      "step": 77510
    },
    {
      "epoch": 232.79,
      "grad_norm": 9.515737533569336,
      "learning_rate": 2.2402402402402406e-06,
      "loss": 0.2443,
      "step": 77520
    },
    {
      "epoch": 232.82,
      "grad_norm": 8.794792175292969,
      "learning_rate": 2.2392392392392396e-06,
      "loss": 0.2235,
      "step": 77530
    },
    {
      "epoch": 232.85,
      "grad_norm": 8.634750366210938,
      "learning_rate": 2.238238238238238e-06,
      "loss": 0.2009,
      "step": 77540
    },
    {
      "epoch": 232.88,
      "grad_norm": 7.033470630645752,
      "learning_rate": 2.2372372372372376e-06,
      "loss": 0.2332,
      "step": 77550
    },
    {
      "epoch": 232.91,
      "grad_norm": 5.513346195220947,
      "learning_rate": 2.2362362362362362e-06,
      "loss": 0.1634,
      "step": 77560
    },
    {
      "epoch": 232.94,
      "grad_norm": 7.526926040649414,
      "learning_rate": 2.2352352352352357e-06,
      "loss": 0.1815,
      "step": 77570
    },
    {
      "epoch": 232.97,
      "grad_norm": 10.892430305480957,
      "learning_rate": 2.2342342342342343e-06,
      "loss": 0.2017,
      "step": 77580
    },
    {
      "epoch": 233.0,
      "eval_accuracy": 0.9219,
      "eval_loss": 0.34811127185821533,
      "eval_runtime": 30.3834,
      "eval_samples_per_second": 329.127,
      "eval_steps_per_second": 1.317,
      "step": 77589
    },
    {
      "epoch": 233.0,
      "grad_norm": 9.773028373718262,
      "learning_rate": 2.2332332332332333e-06,
      "loss": 0.1774,
      "step": 77590
    },
    {
      "epoch": 233.03,
      "grad_norm": 8.507923126220703,
      "learning_rate": 2.2322322322322323e-06,
      "loss": 0.2068,
      "step": 77600
    },
    {
      "epoch": 233.06,
      "grad_norm": 4.806851387023926,
      "learning_rate": 2.2312312312312313e-06,
      "loss": 0.1915,
      "step": 77610
    },
    {
      "epoch": 233.09,
      "grad_norm": 8.063732147216797,
      "learning_rate": 2.2302302302302304e-06,
      "loss": 0.1618,
      "step": 77620
    },
    {
      "epoch": 233.12,
      "grad_norm": 16.068056106567383,
      "learning_rate": 2.2292292292292294e-06,
      "loss": 0.1864,
      "step": 77630
    },
    {
      "epoch": 233.15,
      "grad_norm": 6.210605144500732,
      "learning_rate": 2.2282282282282284e-06,
      "loss": 0.2102,
      "step": 77640
    },
    {
      "epoch": 233.18,
      "grad_norm": 6.056403636932373,
      "learning_rate": 2.2272272272272274e-06,
      "loss": 0.2301,
      "step": 77650
    },
    {
      "epoch": 233.21,
      "grad_norm": 27.828643798828125,
      "learning_rate": 2.2262262262262264e-06,
      "loss": 0.2065,
      "step": 77660
    },
    {
      "epoch": 233.24,
      "grad_norm": 8.4160737991333,
      "learning_rate": 2.2252252252252254e-06,
      "loss": 0.211,
      "step": 77670
    },
    {
      "epoch": 233.27,
      "grad_norm": 9.41533088684082,
      "learning_rate": 2.2242242242242245e-06,
      "loss": 0.1851,
      "step": 77680
    },
    {
      "epoch": 233.3,
      "grad_norm": 10.341246604919434,
      "learning_rate": 2.2232232232232235e-06,
      "loss": 0.1648,
      "step": 77690
    },
    {
      "epoch": 233.33,
      "grad_norm": 8.330281257629395,
      "learning_rate": 2.222222222222222e-06,
      "loss": 0.204,
      "step": 77700
    },
    {
      "epoch": 233.36,
      "grad_norm": 14.169529914855957,
      "learning_rate": 2.2212212212212215e-06,
      "loss": 0.1909,
      "step": 77710
    },
    {
      "epoch": 233.39,
      "grad_norm": 10.802116394042969,
      "learning_rate": 2.22022022022022e-06,
      "loss": 0.2447,
      "step": 77720
    },
    {
      "epoch": 233.42,
      "grad_norm": 11.637665748596191,
      "learning_rate": 2.2192192192192196e-06,
      "loss": 0.1731,
      "step": 77730
    },
    {
      "epoch": 233.45,
      "grad_norm": 13.40414047241211,
      "learning_rate": 2.2182182182182186e-06,
      "loss": 0.2043,
      "step": 77740
    },
    {
      "epoch": 233.48,
      "grad_norm": 12.092058181762695,
      "learning_rate": 2.2172172172172176e-06,
      "loss": 0.1822,
      "step": 77750
    },
    {
      "epoch": 233.51,
      "grad_norm": 13.406065940856934,
      "learning_rate": 2.2162162162162166e-06,
      "loss": 0.2173,
      "step": 77760
    },
    {
      "epoch": 233.54,
      "grad_norm": 13.932661056518555,
      "learning_rate": 2.215215215215215e-06,
      "loss": 0.1849,
      "step": 77770
    },
    {
      "epoch": 233.57,
      "grad_norm": 8.637320518493652,
      "learning_rate": 2.2142142142142146e-06,
      "loss": 0.2016,
      "step": 77780
    },
    {
      "epoch": 233.6,
      "grad_norm": 51.660911560058594,
      "learning_rate": 2.2132132132132132e-06,
      "loss": 0.1454,
      "step": 77790
    },
    {
      "epoch": 233.63,
      "grad_norm": 6.8492279052734375,
      "learning_rate": 2.2122122122122127e-06,
      "loss": 0.164,
      "step": 77800
    },
    {
      "epoch": 233.66,
      "grad_norm": 18.678939819335938,
      "learning_rate": 2.2112112112112113e-06,
      "loss": 0.1994,
      "step": 77810
    },
    {
      "epoch": 233.69,
      "grad_norm": 5.885132312774658,
      "learning_rate": 2.2102102102102103e-06,
      "loss": 0.169,
      "step": 77820
    },
    {
      "epoch": 233.72,
      "grad_norm": 10.849691390991211,
      "learning_rate": 2.2092092092092093e-06,
      "loss": 0.207,
      "step": 77830
    },
    {
      "epoch": 233.75,
      "grad_norm": 8.817901611328125,
      "learning_rate": 2.2082082082082083e-06,
      "loss": 0.2242,
      "step": 77840
    },
    {
      "epoch": 233.78,
      "grad_norm": 13.185579299926758,
      "learning_rate": 2.2072072072072073e-06,
      "loss": 0.2049,
      "step": 77850
    },
    {
      "epoch": 233.81,
      "grad_norm": 7.274044513702393,
      "learning_rate": 2.2062062062062064e-06,
      "loss": 0.1533,
      "step": 77860
    },
    {
      "epoch": 233.84,
      "grad_norm": 11.097867965698242,
      "learning_rate": 2.2052052052052054e-06,
      "loss": 0.2211,
      "step": 77870
    },
    {
      "epoch": 233.87,
      "grad_norm": 18.961280822753906,
      "learning_rate": 2.2042042042042044e-06,
      "loss": 0.1867,
      "step": 77880
    },
    {
      "epoch": 233.9,
      "grad_norm": 13.892167091369629,
      "learning_rate": 2.2032032032032034e-06,
      "loss": 0.2186,
      "step": 77890
    },
    {
      "epoch": 233.93,
      "grad_norm": 4.574970722198486,
      "learning_rate": 2.2022022022022024e-06,
      "loss": 0.1948,
      "step": 77900
    },
    {
      "epoch": 233.96,
      "grad_norm": 8.236494064331055,
      "learning_rate": 2.2012012012012015e-06,
      "loss": 0.1553,
      "step": 77910
    },
    {
      "epoch": 233.99,
      "grad_norm": 13.981201171875,
      "learning_rate": 2.2002002002002005e-06,
      "loss": 0.1883,
      "step": 77920
    },
    {
      "epoch": 234.0,
      "eval_accuracy": 0.9218,
      "eval_loss": 0.3478911519050598,
      "eval_runtime": 30.2686,
      "eval_samples_per_second": 330.375,
      "eval_steps_per_second": 1.322,
      "step": 77922
    },
    {
      "epoch": 234.02,
      "grad_norm": 8.878073692321777,
      "learning_rate": 2.199199199199199e-06,
      "loss": 0.1493,
      "step": 77930
    },
    {
      "epoch": 234.05,
      "grad_norm": 7.709283828735352,
      "learning_rate": 2.1981981981981985e-06,
      "loss": 0.1902,
      "step": 77940
    },
    {
      "epoch": 234.08,
      "grad_norm": 11.465252876281738,
      "learning_rate": 2.197197197197197e-06,
      "loss": 0.2024,
      "step": 77950
    },
    {
      "epoch": 234.11,
      "grad_norm": 8.353307723999023,
      "learning_rate": 2.1961961961961966e-06,
      "loss": 0.1862,
      "step": 77960
    },
    {
      "epoch": 234.14,
      "grad_norm": 16.182514190673828,
      "learning_rate": 2.195195195195195e-06,
      "loss": 0.2146,
      "step": 77970
    },
    {
      "epoch": 234.17,
      "grad_norm": 9.118767738342285,
      "learning_rate": 2.1941941941941946e-06,
      "loss": 0.206,
      "step": 77980
    },
    {
      "epoch": 234.2,
      "grad_norm": 11.01842975616455,
      "learning_rate": 2.193193193193193e-06,
      "loss": 0.188,
      "step": 77990
    },
    {
      "epoch": 234.23,
      "grad_norm": 6.0870819091796875,
      "learning_rate": 2.192192192192192e-06,
      "loss": 0.1647,
      "step": 78000
    },
    {
      "epoch": 234.26,
      "grad_norm": 11.466302871704102,
      "learning_rate": 2.1911911911911916e-06,
      "loss": 0.1921,
      "step": 78010
    },
    {
      "epoch": 234.29,
      "grad_norm": 8.323821067810059,
      "learning_rate": 2.1901901901901902e-06,
      "loss": 0.1644,
      "step": 78020
    },
    {
      "epoch": 234.32,
      "grad_norm": 6.693275451660156,
      "learning_rate": 2.1891891891891897e-06,
      "loss": 0.1596,
      "step": 78030
    },
    {
      "epoch": 234.35,
      "grad_norm": 7.954397678375244,
      "learning_rate": 2.1881881881881883e-06,
      "loss": 0.1962,
      "step": 78040
    },
    {
      "epoch": 234.38,
      "grad_norm": 15.497274398803711,
      "learning_rate": 2.1871871871871873e-06,
      "loss": 0.2079,
      "step": 78050
    },
    {
      "epoch": 234.41,
      "grad_norm": 10.145101547241211,
      "learning_rate": 2.1861861861861863e-06,
      "loss": 0.1693,
      "step": 78060
    },
    {
      "epoch": 234.44,
      "grad_norm": 17.107101440429688,
      "learning_rate": 2.1851851851851853e-06,
      "loss": 0.1883,
      "step": 78070
    },
    {
      "epoch": 234.47,
      "grad_norm": 12.709392547607422,
      "learning_rate": 2.1841841841841843e-06,
      "loss": 0.1853,
      "step": 78080
    },
    {
      "epoch": 234.5,
      "grad_norm": 9.221663475036621,
      "learning_rate": 2.1831831831831834e-06,
      "loss": 0.2073,
      "step": 78090
    },
    {
      "epoch": 234.53,
      "grad_norm": 11.417313575744629,
      "learning_rate": 2.1821821821821824e-06,
      "loss": 0.1622,
      "step": 78100
    },
    {
      "epoch": 234.56,
      "grad_norm": 12.745648384094238,
      "learning_rate": 2.1811811811811814e-06,
      "loss": 0.1829,
      "step": 78110
    },
    {
      "epoch": 234.59,
      "grad_norm": 7.721934795379639,
      "learning_rate": 2.1801801801801804e-06,
      "loss": 0.2292,
      "step": 78120
    },
    {
      "epoch": 234.62,
      "grad_norm": 10.579442977905273,
      "learning_rate": 2.1791791791791794e-06,
      "loss": 0.2058,
      "step": 78130
    },
    {
      "epoch": 234.65,
      "grad_norm": 6.891254425048828,
      "learning_rate": 2.1781781781781785e-06,
      "loss": 0.2222,
      "step": 78140
    },
    {
      "epoch": 234.68,
      "grad_norm": 11.368535995483398,
      "learning_rate": 2.1771771771771775e-06,
      "loss": 0.1902,
      "step": 78150
    },
    {
      "epoch": 234.71,
      "grad_norm": 24.286836624145508,
      "learning_rate": 2.176176176176176e-06,
      "loss": 0.1856,
      "step": 78160
    },
    {
      "epoch": 234.74,
      "grad_norm": 8.479522705078125,
      "learning_rate": 2.1751751751751755e-06,
      "loss": 0.1589,
      "step": 78170
    },
    {
      "epoch": 234.77,
      "grad_norm": 9.470178604125977,
      "learning_rate": 2.174174174174174e-06,
      "loss": 0.205,
      "step": 78180
    },
    {
      "epoch": 234.8,
      "grad_norm": 6.674159526824951,
      "learning_rate": 2.1731731731731735e-06,
      "loss": 0.2211,
      "step": 78190
    },
    {
      "epoch": 234.83,
      "grad_norm": 10.209663391113281,
      "learning_rate": 2.172172172172172e-06,
      "loss": 0.1776,
      "step": 78200
    },
    {
      "epoch": 234.86,
      "grad_norm": 11.971497535705566,
      "learning_rate": 2.1711711711711716e-06,
      "loss": 0.2086,
      "step": 78210
    },
    {
      "epoch": 234.89,
      "grad_norm": 11.3289155960083,
      "learning_rate": 2.17017017017017e-06,
      "loss": 0.1962,
      "step": 78220
    },
    {
      "epoch": 234.92,
      "grad_norm": 12.329174995422363,
      "learning_rate": 2.169169169169169e-06,
      "loss": 0.2303,
      "step": 78230
    },
    {
      "epoch": 234.95,
      "grad_norm": 6.271103382110596,
      "learning_rate": 2.1681681681681682e-06,
      "loss": 0.1899,
      "step": 78240
    },
    {
      "epoch": 234.98,
      "grad_norm": 9.19266414642334,
      "learning_rate": 2.1671671671671672e-06,
      "loss": 0.1682,
      "step": 78250
    },
    {
      "epoch": 235.0,
      "eval_accuracy": 0.9214,
      "eval_loss": 0.3475821316242218,
      "eval_runtime": 30.0669,
      "eval_samples_per_second": 332.592,
      "eval_steps_per_second": 1.33,
      "step": 78255
    },
    {
      "epoch": 235.02,
      "grad_norm": 8.550549507141113,
      "learning_rate": 2.1661661661661663e-06,
      "loss": 0.2188,
      "step": 78260
    },
    {
      "epoch": 235.05,
      "grad_norm": 7.402585506439209,
      "learning_rate": 2.1651651651651653e-06,
      "loss": 0.2028,
      "step": 78270
    },
    {
      "epoch": 235.08,
      "grad_norm": 3.9888172149658203,
      "learning_rate": 2.1641641641641643e-06,
      "loss": 0.1652,
      "step": 78280
    },
    {
      "epoch": 235.11,
      "grad_norm": 22.838623046875,
      "learning_rate": 2.1631631631631633e-06,
      "loss": 0.192,
      "step": 78290
    },
    {
      "epoch": 235.14,
      "grad_norm": 9.974997520446777,
      "learning_rate": 2.1621621621621623e-06,
      "loss": 0.1944,
      "step": 78300
    },
    {
      "epoch": 235.17,
      "grad_norm": 14.093962669372559,
      "learning_rate": 2.1611611611611613e-06,
      "loss": 0.1941,
      "step": 78310
    },
    {
      "epoch": 235.2,
      "grad_norm": 6.051436424255371,
      "learning_rate": 2.1601601601601604e-06,
      "loss": 0.2071,
      "step": 78320
    },
    {
      "epoch": 235.23,
      "grad_norm": 7.633190631866455,
      "learning_rate": 2.1591591591591594e-06,
      "loss": 0.1803,
      "step": 78330
    },
    {
      "epoch": 235.26,
      "grad_norm": 9.410310745239258,
      "learning_rate": 2.1581581581581584e-06,
      "loss": 0.179,
      "step": 78340
    },
    {
      "epoch": 235.29,
      "grad_norm": 6.557682037353516,
      "learning_rate": 2.1571571571571574e-06,
      "loss": 0.1725,
      "step": 78350
    },
    {
      "epoch": 235.32,
      "grad_norm": 10.62854290008545,
      "learning_rate": 2.1561561561561564e-06,
      "loss": 0.2111,
      "step": 78360
    },
    {
      "epoch": 235.35,
      "grad_norm": 8.605103492736816,
      "learning_rate": 2.1551551551551555e-06,
      "loss": 0.1865,
      "step": 78370
    },
    {
      "epoch": 235.38,
      "grad_norm": 7.262552738189697,
      "learning_rate": 2.1541541541541545e-06,
      "loss": 0.1415,
      "step": 78380
    },
    {
      "epoch": 235.41,
      "grad_norm": 8.12015151977539,
      "learning_rate": 2.153153153153153e-06,
      "loss": 0.1833,
      "step": 78390
    },
    {
      "epoch": 235.44,
      "grad_norm": 7.449734210968018,
      "learning_rate": 2.1521521521521525e-06,
      "loss": 0.1969,
      "step": 78400
    },
    {
      "epoch": 235.47,
      "grad_norm": 7.408007621765137,
      "learning_rate": 2.151151151151151e-06,
      "loss": 0.2032,
      "step": 78410
    },
    {
      "epoch": 235.5,
      "grad_norm": 11.683579444885254,
      "learning_rate": 2.1501501501501505e-06,
      "loss": 0.1794,
      "step": 78420
    },
    {
      "epoch": 235.53,
      "grad_norm": 4.7686920166015625,
      "learning_rate": 2.149149149149149e-06,
      "loss": 0.1932,
      "step": 78430
    },
    {
      "epoch": 235.56,
      "grad_norm": 11.894322395324707,
      "learning_rate": 2.148148148148148e-06,
      "loss": 0.2002,
      "step": 78440
    },
    {
      "epoch": 235.59,
      "grad_norm": 8.245132446289062,
      "learning_rate": 2.147147147147147e-06,
      "loss": 0.1869,
      "step": 78450
    },
    {
      "epoch": 235.62,
      "grad_norm": 5.187326908111572,
      "learning_rate": 2.146146146146146e-06,
      "loss": 0.1775,
      "step": 78460
    },
    {
      "epoch": 235.65,
      "grad_norm": 11.751038551330566,
      "learning_rate": 2.1451451451451452e-06,
      "loss": 0.1832,
      "step": 78470
    },
    {
      "epoch": 235.68,
      "grad_norm": 16.29152488708496,
      "learning_rate": 2.1441441441441442e-06,
      "loss": 0.2606,
      "step": 78480
    },
    {
      "epoch": 235.71,
      "grad_norm": 10.798697471618652,
      "learning_rate": 2.1431431431431433e-06,
      "loss": 0.182,
      "step": 78490
    },
    {
      "epoch": 235.74,
      "grad_norm": 8.379005432128906,
      "learning_rate": 2.1421421421421423e-06,
      "loss": 0.1766,
      "step": 78500
    },
    {
      "epoch": 235.77,
      "grad_norm": 8.876402854919434,
      "learning_rate": 2.1411411411411413e-06,
      "loss": 0.2126,
      "step": 78510
    },
    {
      "epoch": 235.8,
      "grad_norm": 14.171756744384766,
      "learning_rate": 2.1401401401401403e-06,
      "loss": 0.1908,
      "step": 78520
    },
    {
      "epoch": 235.83,
      "grad_norm": 5.782662391662598,
      "learning_rate": 2.1391391391391393e-06,
      "loss": 0.194,
      "step": 78530
    },
    {
      "epoch": 235.86,
      "grad_norm": 8.817935943603516,
      "learning_rate": 2.1381381381381383e-06,
      "loss": 0.1872,
      "step": 78540
    },
    {
      "epoch": 235.89,
      "grad_norm": 10.313138008117676,
      "learning_rate": 2.1371371371371374e-06,
      "loss": 0.2196,
      "step": 78550
    },
    {
      "epoch": 235.92,
      "grad_norm": 7.986395359039307,
      "learning_rate": 2.1361361361361364e-06,
      "loss": 0.1707,
      "step": 78560
    },
    {
      "epoch": 235.95,
      "grad_norm": 7.856678485870361,
      "learning_rate": 2.1351351351351354e-06,
      "loss": 0.1965,
      "step": 78570
    },
    {
      "epoch": 235.98,
      "grad_norm": 8.443798065185547,
      "learning_rate": 2.1341341341341344e-06,
      "loss": 0.1702,
      "step": 78580
    },
    {
      "epoch": 236.0,
      "eval_accuracy": 0.922,
      "eval_loss": 0.34736067056655884,
      "eval_runtime": 30.1621,
      "eval_samples_per_second": 331.542,
      "eval_steps_per_second": 1.326,
      "step": 78588
    },
    {
      "epoch": 236.01,
      "grad_norm": 7.043339252471924,
      "learning_rate": 2.1331331331331334e-06,
      "loss": 0.1863,
      "step": 78590
    },
    {
      "epoch": 236.04,
      "grad_norm": 9.298216819763184,
      "learning_rate": 2.1321321321321325e-06,
      "loss": 0.1859,
      "step": 78600
    },
    {
      "epoch": 236.07,
      "grad_norm": 29.273345947265625,
      "learning_rate": 2.1311311311311315e-06,
      "loss": 0.188,
      "step": 78610
    },
    {
      "epoch": 236.1,
      "grad_norm": 9.741768836975098,
      "learning_rate": 2.13013013013013e-06,
      "loss": 0.2131,
      "step": 78620
    },
    {
      "epoch": 236.13,
      "grad_norm": 8.26407241821289,
      "learning_rate": 2.1291291291291295e-06,
      "loss": 0.2184,
      "step": 78630
    },
    {
      "epoch": 236.16,
      "grad_norm": 6.452900409698486,
      "learning_rate": 2.128128128128128e-06,
      "loss": 0.1601,
      "step": 78640
    },
    {
      "epoch": 236.19,
      "grad_norm": 4.802979946136475,
      "learning_rate": 2.1271271271271275e-06,
      "loss": 0.1949,
      "step": 78650
    },
    {
      "epoch": 236.22,
      "grad_norm": 8.177454948425293,
      "learning_rate": 2.126126126126126e-06,
      "loss": 0.2017,
      "step": 78660
    },
    {
      "epoch": 236.25,
      "grad_norm": 6.947920322418213,
      "learning_rate": 2.125125125125125e-06,
      "loss": 0.1806,
      "step": 78670
    },
    {
      "epoch": 236.28,
      "grad_norm": 7.2344489097595215,
      "learning_rate": 2.124124124124124e-06,
      "loss": 0.2266,
      "step": 78680
    },
    {
      "epoch": 236.31,
      "grad_norm": 7.5352864265441895,
      "learning_rate": 2.123123123123123e-06,
      "loss": 0.1946,
      "step": 78690
    },
    {
      "epoch": 236.34,
      "grad_norm": 10.896170616149902,
      "learning_rate": 2.1221221221221222e-06,
      "loss": 0.1778,
      "step": 78700
    },
    {
      "epoch": 236.37,
      "grad_norm": 7.374946117401123,
      "learning_rate": 2.1211211211211212e-06,
      "loss": 0.2052,
      "step": 78710
    },
    {
      "epoch": 236.4,
      "grad_norm": 7.210081100463867,
      "learning_rate": 2.1201201201201203e-06,
      "loss": 0.2343,
      "step": 78720
    },
    {
      "epoch": 236.43,
      "grad_norm": 11.332025527954102,
      "learning_rate": 2.1191191191191193e-06,
      "loss": 0.2283,
      "step": 78730
    },
    {
      "epoch": 236.46,
      "grad_norm": 6.406346321105957,
      "learning_rate": 2.1181181181181183e-06,
      "loss": 0.1815,
      "step": 78740
    },
    {
      "epoch": 236.49,
      "grad_norm": 5.972035884857178,
      "learning_rate": 2.1171171171171173e-06,
      "loss": 0.1646,
      "step": 78750
    },
    {
      "epoch": 236.52,
      "grad_norm": 14.109481811523438,
      "learning_rate": 2.1161161161161163e-06,
      "loss": 0.2044,
      "step": 78760
    },
    {
      "epoch": 236.55,
      "grad_norm": 11.360723495483398,
      "learning_rate": 2.1151151151151153e-06,
      "loss": 0.2022,
      "step": 78770
    },
    {
      "epoch": 236.58,
      "grad_norm": 6.9900970458984375,
      "learning_rate": 2.1141141141141144e-06,
      "loss": 0.1984,
      "step": 78780
    },
    {
      "epoch": 236.61,
      "grad_norm": 5.4909491539001465,
      "learning_rate": 2.1131131131131134e-06,
      "loss": 0.1922,
      "step": 78790
    },
    {
      "epoch": 236.64,
      "grad_norm": 6.3498215675354,
      "learning_rate": 2.1121121121121124e-06,
      "loss": 0.1662,
      "step": 78800
    },
    {
      "epoch": 236.67,
      "grad_norm": 5.466065883636475,
      "learning_rate": 2.1111111111111114e-06,
      "loss": 0.2012,
      "step": 78810
    },
    {
      "epoch": 236.7,
      "grad_norm": 8.279924392700195,
      "learning_rate": 2.1101101101101104e-06,
      "loss": 0.1847,
      "step": 78820
    },
    {
      "epoch": 236.73,
      "grad_norm": 7.37285041809082,
      "learning_rate": 2.1091091091091095e-06,
      "loss": 0.2158,
      "step": 78830
    },
    {
      "epoch": 236.76,
      "grad_norm": 13.933272361755371,
      "learning_rate": 2.1081081081081085e-06,
      "loss": 0.1981,
      "step": 78840
    },
    {
      "epoch": 236.79,
      "grad_norm": 5.207080841064453,
      "learning_rate": 2.107107107107107e-06,
      "loss": 0.2424,
      "step": 78850
    },
    {
      "epoch": 236.82,
      "grad_norm": 8.660426139831543,
      "learning_rate": 2.1061061061061065e-06,
      "loss": 0.1855,
      "step": 78860
    },
    {
      "epoch": 236.85,
      "grad_norm": 7.759766101837158,
      "learning_rate": 2.105105105105105e-06,
      "loss": 0.2078,
      "step": 78870
    },
    {
      "epoch": 236.88,
      "grad_norm": 6.20126485824585,
      "learning_rate": 2.1041041041041045e-06,
      "loss": 0.158,
      "step": 78880
    },
    {
      "epoch": 236.91,
      "grad_norm": 6.423630714416504,
      "learning_rate": 2.103103103103103e-06,
      "loss": 0.2035,
      "step": 78890
    },
    {
      "epoch": 236.94,
      "grad_norm": 8.128384590148926,
      "learning_rate": 2.102102102102102e-06,
      "loss": 0.1884,
      "step": 78900
    },
    {
      "epoch": 236.97,
      "grad_norm": 6.093522071838379,
      "learning_rate": 2.101101101101101e-06,
      "loss": 0.1816,
      "step": 78910
    },
    {
      "epoch": 237.0,
      "grad_norm": 10.29037094116211,
      "learning_rate": 2.1001001001001e-06,
      "loss": 0.2109,
      "step": 78920
    },
    {
      "epoch": 237.0,
      "eval_accuracy": 0.9221,
      "eval_loss": 0.34987667202949524,
      "eval_runtime": 30.2271,
      "eval_samples_per_second": 330.829,
      "eval_steps_per_second": 1.323,
      "step": 78921
    },
    {
      "epoch": 237.03,
      "grad_norm": 6.273860454559326,
      "learning_rate": 2.0990990990990992e-06,
      "loss": 0.1846,
      "step": 78930
    },
    {
      "epoch": 237.06,
      "grad_norm": 4.031570911407471,
      "learning_rate": 2.0980980980980982e-06,
      "loss": 0.2248,
      "step": 78940
    },
    {
      "epoch": 237.09,
      "grad_norm": 5.685488700866699,
      "learning_rate": 2.0970970970970972e-06,
      "loss": 0.1748,
      "step": 78950
    },
    {
      "epoch": 237.12,
      "grad_norm": 25.554378509521484,
      "learning_rate": 2.0960960960960963e-06,
      "loss": 0.2147,
      "step": 78960
    },
    {
      "epoch": 237.15,
      "grad_norm": 10.339054107666016,
      "learning_rate": 2.0950950950950953e-06,
      "loss": 0.2108,
      "step": 78970
    },
    {
      "epoch": 237.18,
      "grad_norm": 6.170294284820557,
      "learning_rate": 2.0940940940940943e-06,
      "loss": 0.1941,
      "step": 78980
    },
    {
      "epoch": 237.21,
      "grad_norm": 12.41234302520752,
      "learning_rate": 2.0930930930930933e-06,
      "loss": 0.2034,
      "step": 78990
    },
    {
      "epoch": 237.24,
      "grad_norm": 8.521265029907227,
      "learning_rate": 2.0920920920920923e-06,
      "loss": 0.1985,
      "step": 79000
    },
    {
      "epoch": 237.27,
      "grad_norm": 18.846654891967773,
      "learning_rate": 2.0910910910910914e-06,
      "loss": 0.2267,
      "step": 79010
    },
    {
      "epoch": 237.3,
      "grad_norm": 4.700964450836182,
      "learning_rate": 2.0900900900900904e-06,
      "loss": 0.2068,
      "step": 79020
    },
    {
      "epoch": 237.33,
      "grad_norm": 8.67593002319336,
      "learning_rate": 2.089089089089089e-06,
      "loss": 0.2072,
      "step": 79030
    },
    {
      "epoch": 237.36,
      "grad_norm": 10.747883796691895,
      "learning_rate": 2.0880880880880884e-06,
      "loss": 0.1912,
      "step": 79040
    },
    {
      "epoch": 237.39,
      "grad_norm": 8.219335556030273,
      "learning_rate": 2.087087087087087e-06,
      "loss": 0.2048,
      "step": 79050
    },
    {
      "epoch": 237.42,
      "grad_norm": 11.4700345993042,
      "learning_rate": 2.0860860860860865e-06,
      "loss": 0.1832,
      "step": 79060
    },
    {
      "epoch": 237.45,
      "grad_norm": 7.124955654144287,
      "learning_rate": 2.0850850850850855e-06,
      "loss": 0.1802,
      "step": 79070
    },
    {
      "epoch": 237.48,
      "grad_norm": 9.896077156066895,
      "learning_rate": 2.084084084084084e-06,
      "loss": 0.1458,
      "step": 79080
    },
    {
      "epoch": 237.51,
      "grad_norm": 14.089030265808105,
      "learning_rate": 2.0830830830830835e-06,
      "loss": 0.1953,
      "step": 79090
    },
    {
      "epoch": 237.54,
      "grad_norm": 5.760733127593994,
      "learning_rate": 2.082082082082082e-06,
      "loss": 0.1926,
      "step": 79100
    },
    {
      "epoch": 237.57,
      "grad_norm": 9.444108963012695,
      "learning_rate": 2.0810810810810815e-06,
      "loss": 0.1998,
      "step": 79110
    },
    {
      "epoch": 237.6,
      "grad_norm": 10.700116157531738,
      "learning_rate": 2.08008008008008e-06,
      "loss": 0.2077,
      "step": 79120
    },
    {
      "epoch": 237.63,
      "grad_norm": 9.35848617553711,
      "learning_rate": 2.079079079079079e-06,
      "loss": 0.1865,
      "step": 79130
    },
    {
      "epoch": 237.66,
      "grad_norm": 6.834915637969971,
      "learning_rate": 2.078078078078078e-06,
      "loss": 0.2058,
      "step": 79140
    },
    {
      "epoch": 237.69,
      "grad_norm": 6.38962459564209,
      "learning_rate": 2.077077077077077e-06,
      "loss": 0.2099,
      "step": 79150
    },
    {
      "epoch": 237.72,
      "grad_norm": 15.078845977783203,
      "learning_rate": 2.076076076076076e-06,
      "loss": 0.2087,
      "step": 79160
    },
    {
      "epoch": 237.75,
      "grad_norm": 16.132314682006836,
      "learning_rate": 2.0750750750750752e-06,
      "loss": 0.183,
      "step": 79170
    },
    {
      "epoch": 237.78,
      "grad_norm": 5.498672962188721,
      "learning_rate": 2.0740740740740742e-06,
      "loss": 0.1945,
      "step": 79180
    },
    {
      "epoch": 237.81,
      "grad_norm": 8.883649826049805,
      "learning_rate": 2.0730730730730733e-06,
      "loss": 0.1786,
      "step": 79190
    },
    {
      "epoch": 237.84,
      "grad_norm": 6.3171706199646,
      "learning_rate": 2.0720720720720723e-06,
      "loss": 0.1949,
      "step": 79200
    },
    {
      "epoch": 237.87,
      "grad_norm": 10.785465240478516,
      "learning_rate": 2.0710710710710713e-06,
      "loss": 0.1781,
      "step": 79210
    },
    {
      "epoch": 237.9,
      "grad_norm": 9.256567001342773,
      "learning_rate": 2.0700700700700703e-06,
      "loss": 0.2124,
      "step": 79220
    },
    {
      "epoch": 237.93,
      "grad_norm": 7.750977039337158,
      "learning_rate": 2.0690690690690693e-06,
      "loss": 0.2335,
      "step": 79230
    },
    {
      "epoch": 237.96,
      "grad_norm": 9.627055168151855,
      "learning_rate": 2.0680680680680684e-06,
      "loss": 0.221,
      "step": 79240
    },
    {
      "epoch": 237.99,
      "grad_norm": 7.926790237426758,
      "learning_rate": 2.0670670670670674e-06,
      "loss": 0.1768,
      "step": 79250
    },
    {
      "epoch": 238.0,
      "eval_accuracy": 0.9211,
      "eval_loss": 0.345855712890625,
      "eval_runtime": 30.3779,
      "eval_samples_per_second": 329.186,
      "eval_steps_per_second": 1.317,
      "step": 79254
    },
    {
      "epoch": 238.02,
      "grad_norm": 6.106064319610596,
      "learning_rate": 2.066066066066066e-06,
      "loss": 0.1455,
      "step": 79260
    },
    {
      "epoch": 238.05,
      "grad_norm": 5.502146244049072,
      "learning_rate": 2.0650650650650654e-06,
      "loss": 0.1779,
      "step": 79270
    },
    {
      "epoch": 238.08,
      "grad_norm": 5.913329124450684,
      "learning_rate": 2.064064064064064e-06,
      "loss": 0.202,
      "step": 79280
    },
    {
      "epoch": 238.11,
      "grad_norm": 9.881763458251953,
      "learning_rate": 2.0630630630630634e-06,
      "loss": 0.199,
      "step": 79290
    },
    {
      "epoch": 238.14,
      "grad_norm": 6.736797332763672,
      "learning_rate": 2.062062062062062e-06,
      "loss": 0.2208,
      "step": 79300
    },
    {
      "epoch": 238.17,
      "grad_norm": 7.218282222747803,
      "learning_rate": 2.061061061061061e-06,
      "loss": 0.1849,
      "step": 79310
    },
    {
      "epoch": 238.2,
      "grad_norm": 7.004521369934082,
      "learning_rate": 2.06006006006006e-06,
      "loss": 0.1665,
      "step": 79320
    },
    {
      "epoch": 238.23,
      "grad_norm": 10.349236488342285,
      "learning_rate": 2.059059059059059e-06,
      "loss": 0.205,
      "step": 79330
    },
    {
      "epoch": 238.26,
      "grad_norm": 7.454785346984863,
      "learning_rate": 2.0580580580580585e-06,
      "loss": 0.2036,
      "step": 79340
    },
    {
      "epoch": 238.29,
      "grad_norm": 7.259304046630859,
      "learning_rate": 2.057057057057057e-06,
      "loss": 0.1682,
      "step": 79350
    },
    {
      "epoch": 238.32,
      "grad_norm": 6.503686428070068,
      "learning_rate": 2.056056056056056e-06,
      "loss": 0.1947,
      "step": 79360
    },
    {
      "epoch": 238.35,
      "grad_norm": 6.232084274291992,
      "learning_rate": 2.055055055055055e-06,
      "loss": 0.2093,
      "step": 79370
    },
    {
      "epoch": 238.38,
      "grad_norm": 9.637796401977539,
      "learning_rate": 2.054054054054054e-06,
      "loss": 0.2013,
      "step": 79380
    },
    {
      "epoch": 238.41,
      "grad_norm": 8.115212440490723,
      "learning_rate": 2.053053053053053e-06,
      "loss": 0.1701,
      "step": 79390
    },
    {
      "epoch": 238.44,
      "grad_norm": 10.36279010772705,
      "learning_rate": 2.0520520520520522e-06,
      "loss": 0.2463,
      "step": 79400
    },
    {
      "epoch": 238.47,
      "grad_norm": 6.724928855895996,
      "learning_rate": 2.0510510510510512e-06,
      "loss": 0.2123,
      "step": 79410
    },
    {
      "epoch": 238.5,
      "grad_norm": 12.026813507080078,
      "learning_rate": 2.0500500500500503e-06,
      "loss": 0.1961,
      "step": 79420
    },
    {
      "epoch": 238.53,
      "grad_norm": 7.492519855499268,
      "learning_rate": 2.0490490490490493e-06,
      "loss": 0.1751,
      "step": 79430
    },
    {
      "epoch": 238.56,
      "grad_norm": 7.336597442626953,
      "learning_rate": 2.0480480480480483e-06,
      "loss": 0.2019,
      "step": 79440
    },
    {
      "epoch": 238.59,
      "grad_norm": 6.296154499053955,
      "learning_rate": 2.0470470470470473e-06,
      "loss": 0.1673,
      "step": 79450
    },
    {
      "epoch": 238.62,
      "grad_norm": 10.297792434692383,
      "learning_rate": 2.0460460460460463e-06,
      "loss": 0.1896,
      "step": 79460
    },
    {
      "epoch": 238.65,
      "grad_norm": 7.095839023590088,
      "learning_rate": 2.045045045045045e-06,
      "loss": 0.1621,
      "step": 79470
    },
    {
      "epoch": 238.68,
      "grad_norm": 11.630331993103027,
      "learning_rate": 2.0440440440440444e-06,
      "loss": 0.169,
      "step": 79480
    },
    {
      "epoch": 238.71,
      "grad_norm": 10.260640144348145,
      "learning_rate": 2.043043043043043e-06,
      "loss": 0.2225,
      "step": 79490
    },
    {
      "epoch": 238.74,
      "grad_norm": 20.294139862060547,
      "learning_rate": 2.0420420420420424e-06,
      "loss": 0.282,
      "step": 79500
    },
    {
      "epoch": 238.77,
      "grad_norm": 8.06268310546875,
      "learning_rate": 2.041041041041041e-06,
      "loss": 0.1965,
      "step": 79510
    },
    {
      "epoch": 238.8,
      "grad_norm": 12.888213157653809,
      "learning_rate": 2.0400400400400404e-06,
      "loss": 0.2142,
      "step": 79520
    },
    {
      "epoch": 238.83,
      "grad_norm": 7.380039691925049,
      "learning_rate": 2.039039039039039e-06,
      "loss": 0.1954,
      "step": 79530
    },
    {
      "epoch": 238.86,
      "grad_norm": 4.943821907043457,
      "learning_rate": 2.038038038038038e-06,
      "loss": 0.1841,
      "step": 79540
    },
    {
      "epoch": 238.89,
      "grad_norm": 5.902561664581299,
      "learning_rate": 2.037037037037037e-06,
      "loss": 0.2138,
      "step": 79550
    },
    {
      "epoch": 238.92,
      "grad_norm": 7.494679927825928,
      "learning_rate": 2.036036036036036e-06,
      "loss": 0.1813,
      "step": 79560
    },
    {
      "epoch": 238.95,
      "grad_norm": 6.075791835784912,
      "learning_rate": 2.035035035035035e-06,
      "loss": 0.184,
      "step": 79570
    },
    {
      "epoch": 238.98,
      "grad_norm": 6.61543607711792,
      "learning_rate": 2.034034034034034e-06,
      "loss": 0.1731,
      "step": 79580
    },
    {
      "epoch": 239.0,
      "eval_accuracy": 0.9222,
      "eval_loss": 0.34798088669776917,
      "eval_runtime": 30.4269,
      "eval_samples_per_second": 328.656,
      "eval_steps_per_second": 1.315,
      "step": 79587
    },
    {
      "epoch": 239.01,
      "grad_norm": 7.615043640136719,
      "learning_rate": 2.033033033033033e-06,
      "loss": 0.2222,
      "step": 79590
    },
    {
      "epoch": 239.04,
      "grad_norm": 10.073203086853027,
      "learning_rate": 2.032032032032032e-06,
      "loss": 0.2083,
      "step": 79600
    },
    {
      "epoch": 239.07,
      "grad_norm": 15.46485710144043,
      "learning_rate": 2.031031031031031e-06,
      "loss": 0.1941,
      "step": 79610
    },
    {
      "epoch": 239.1,
      "grad_norm": 7.241711139678955,
      "learning_rate": 2.03003003003003e-06,
      "loss": 0.1554,
      "step": 79620
    },
    {
      "epoch": 239.13,
      "grad_norm": 8.407838821411133,
      "learning_rate": 2.0290290290290292e-06,
      "loss": 0.2486,
      "step": 79630
    },
    {
      "epoch": 239.16,
      "grad_norm": 9.458257675170898,
      "learning_rate": 2.0280280280280282e-06,
      "loss": 0.1778,
      "step": 79640
    },
    {
      "epoch": 239.19,
      "grad_norm": 17.67097282409668,
      "learning_rate": 2.0270270270270273e-06,
      "loss": 0.2166,
      "step": 79650
    },
    {
      "epoch": 239.22,
      "grad_norm": 15.756818771362305,
      "learning_rate": 2.0260260260260263e-06,
      "loss": 0.1771,
      "step": 79660
    },
    {
      "epoch": 239.25,
      "grad_norm": 5.676259994506836,
      "learning_rate": 2.0250250250250253e-06,
      "loss": 0.1756,
      "step": 79670
    },
    {
      "epoch": 239.28,
      "grad_norm": 12.100668907165527,
      "learning_rate": 2.0240240240240243e-06,
      "loss": 0.2176,
      "step": 79680
    },
    {
      "epoch": 239.31,
      "grad_norm": 6.945318222045898,
      "learning_rate": 2.0230230230230233e-06,
      "loss": 0.1781,
      "step": 79690
    },
    {
      "epoch": 239.34,
      "grad_norm": 6.00339937210083,
      "learning_rate": 2.022022022022022e-06,
      "loss": 0.2285,
      "step": 79700
    },
    {
      "epoch": 239.37,
      "grad_norm": 13.743782043457031,
      "learning_rate": 2.0210210210210214e-06,
      "loss": 0.222,
      "step": 79710
    },
    {
      "epoch": 239.4,
      "grad_norm": 17.3043212890625,
      "learning_rate": 2.02002002002002e-06,
      "loss": 0.233,
      "step": 79720
    },
    {
      "epoch": 239.43,
      "grad_norm": 11.150293350219727,
      "learning_rate": 2.0190190190190194e-06,
      "loss": 0.1789,
      "step": 79730
    },
    {
      "epoch": 239.46,
      "grad_norm": 9.612808227539062,
      "learning_rate": 2.018018018018018e-06,
      "loss": 0.1968,
      "step": 79740
    },
    {
      "epoch": 239.49,
      "grad_norm": 13.308731079101562,
      "learning_rate": 2.0170170170170174e-06,
      "loss": 0.2162,
      "step": 79750
    },
    {
      "epoch": 239.52,
      "grad_norm": 8.308270454406738,
      "learning_rate": 2.016016016016016e-06,
      "loss": 0.1908,
      "step": 79760
    },
    {
      "epoch": 239.55,
      "grad_norm": 8.905388832092285,
      "learning_rate": 2.015015015015015e-06,
      "loss": 0.2009,
      "step": 79770
    },
    {
      "epoch": 239.58,
      "grad_norm": 10.438644409179688,
      "learning_rate": 2.014014014014014e-06,
      "loss": 0.1754,
      "step": 79780
    },
    {
      "epoch": 239.61,
      "grad_norm": 14.141783714294434,
      "learning_rate": 2.013013013013013e-06,
      "loss": 0.1994,
      "step": 79790
    },
    {
      "epoch": 239.64,
      "grad_norm": 11.073104858398438,
      "learning_rate": 2.012012012012012e-06,
      "loss": 0.2223,
      "step": 79800
    },
    {
      "epoch": 239.67,
      "grad_norm": 14.930084228515625,
      "learning_rate": 2.011011011011011e-06,
      "loss": 0.1581,
      "step": 79810
    },
    {
      "epoch": 239.7,
      "grad_norm": 6.57129430770874,
      "learning_rate": 2.01001001001001e-06,
      "loss": 0.2117,
      "step": 79820
    },
    {
      "epoch": 239.73,
      "grad_norm": 10.752714157104492,
      "learning_rate": 2.009009009009009e-06,
      "loss": 0.1995,
      "step": 79830
    },
    {
      "epoch": 239.76,
      "grad_norm": 11.19072151184082,
      "learning_rate": 2.008008008008008e-06,
      "loss": 0.2001,
      "step": 79840
    },
    {
      "epoch": 239.79,
      "grad_norm": 20.70919418334961,
      "learning_rate": 2.007007007007007e-06,
      "loss": 0.1779,
      "step": 79850
    },
    {
      "epoch": 239.82,
      "grad_norm": 3.7217342853546143,
      "learning_rate": 2.0060060060060062e-06,
      "loss": 0.1663,
      "step": 79860
    },
    {
      "epoch": 239.85,
      "grad_norm": 7.585594654083252,
      "learning_rate": 2.0050050050050052e-06,
      "loss": 0.2065,
      "step": 79870
    },
    {
      "epoch": 239.88,
      "grad_norm": 6.529752731323242,
      "learning_rate": 2.0040040040040043e-06,
      "loss": 0.169,
      "step": 79880
    },
    {
      "epoch": 239.91,
      "grad_norm": 5.585639953613281,
      "learning_rate": 2.0030030030030033e-06,
      "loss": 0.2163,
      "step": 79890
    },
    {
      "epoch": 239.94,
      "grad_norm": 4.716883659362793,
      "learning_rate": 2.0020020020020023e-06,
      "loss": 0.185,
      "step": 79900
    },
    {
      "epoch": 239.97,
      "grad_norm": 6.308490753173828,
      "learning_rate": 2.0010010010010013e-06,
      "loss": 0.2153,
      "step": 79910
    },
    {
      "epoch": 240.0,
      "grad_norm": 0.4902006685733795,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.1834,
      "step": 79920
    },
    {
      "epoch": 240.0,
      "eval_accuracy": 0.9216,
      "eval_loss": 0.3478997051715851,
      "eval_runtime": 30.4603,
      "eval_samples_per_second": 328.296,
      "eval_steps_per_second": 1.313,
      "step": 79920
    },
    {
      "epoch": 240.03,
      "grad_norm": 9.01107406616211,
      "learning_rate": 1.998998998998999e-06,
      "loss": 0.1695,
      "step": 79930
    },
    {
      "epoch": 240.06,
      "grad_norm": 8.933688163757324,
      "learning_rate": 1.9979979979979984e-06,
      "loss": 0.203,
      "step": 79940
    },
    {
      "epoch": 240.09,
      "grad_norm": 9.948750495910645,
      "learning_rate": 1.996996996996997e-06,
      "loss": 0.1904,
      "step": 79950
    },
    {
      "epoch": 240.12,
      "grad_norm": 11.9343843460083,
      "learning_rate": 1.9959959959959964e-06,
      "loss": 0.2208,
      "step": 79960
    },
    {
      "epoch": 240.15,
      "grad_norm": 5.106485843658447,
      "learning_rate": 1.994994994994995e-06,
      "loss": 0.2133,
      "step": 79970
    },
    {
      "epoch": 240.18,
      "grad_norm": 7.129885196685791,
      "learning_rate": 1.9939939939939944e-06,
      "loss": 0.1988,
      "step": 79980
    },
    {
      "epoch": 240.21,
      "grad_norm": 10.71773624420166,
      "learning_rate": 1.992992992992993e-06,
      "loss": 0.1879,
      "step": 79990
    },
    {
      "epoch": 240.24,
      "grad_norm": 7.343792915344238,
      "learning_rate": 1.991991991991992e-06,
      "loss": 0.1652,
      "step": 80000
    },
    {
      "epoch": 240.27,
      "grad_norm": 8.452603340148926,
      "learning_rate": 1.990990990990991e-06,
      "loss": 0.1789,
      "step": 80010
    },
    {
      "epoch": 240.3,
      "grad_norm": 7.41859769821167,
      "learning_rate": 1.98998998998999e-06,
      "loss": 0.1903,
      "step": 80020
    },
    {
      "epoch": 240.33,
      "grad_norm": 9.066368103027344,
      "learning_rate": 1.988988988988989e-06,
      "loss": 0.1957,
      "step": 80030
    },
    {
      "epoch": 240.36,
      "grad_norm": 11.874316215515137,
      "learning_rate": 1.987987987987988e-06,
      "loss": 0.174,
      "step": 80040
    },
    {
      "epoch": 240.39,
      "grad_norm": 8.381061553955078,
      "learning_rate": 1.986986986986987e-06,
      "loss": 0.188,
      "step": 80050
    },
    {
      "epoch": 240.42,
      "grad_norm": 6.0274977684021,
      "learning_rate": 1.985985985985986e-06,
      "loss": 0.1898,
      "step": 80060
    },
    {
      "epoch": 240.45,
      "grad_norm": 10.707037925720215,
      "learning_rate": 1.984984984984985e-06,
      "loss": 0.1965,
      "step": 80070
    },
    {
      "epoch": 240.48,
      "grad_norm": 10.52853775024414,
      "learning_rate": 1.983983983983984e-06,
      "loss": 0.1817,
      "step": 80080
    },
    {
      "epoch": 240.51,
      "grad_norm": 9.512840270996094,
      "learning_rate": 1.9829829829829832e-06,
      "loss": 0.1653,
      "step": 80090
    },
    {
      "epoch": 240.54,
      "grad_norm": 10.488853454589844,
      "learning_rate": 1.9819819819819822e-06,
      "loss": 0.1824,
      "step": 80100
    },
    {
      "epoch": 240.57,
      "grad_norm": 8.174503326416016,
      "learning_rate": 1.980980980980981e-06,
      "loss": 0.2044,
      "step": 80110
    },
    {
      "epoch": 240.6,
      "grad_norm": 10.652790069580078,
      "learning_rate": 1.9799799799799803e-06,
      "loss": 0.1769,
      "step": 80120
    },
    {
      "epoch": 240.63,
      "grad_norm": 9.04178237915039,
      "learning_rate": 1.978978978978979e-06,
      "loss": 0.1752,
      "step": 80130
    },
    {
      "epoch": 240.66,
      "grad_norm": 11.672818183898926,
      "learning_rate": 1.9779779779779783e-06,
      "loss": 0.1866,
      "step": 80140
    },
    {
      "epoch": 240.69,
      "grad_norm": 6.767496585845947,
      "learning_rate": 1.9769769769769773e-06,
      "loss": 0.1967,
      "step": 80150
    },
    {
      "epoch": 240.72,
      "grad_norm": 12.44456672668457,
      "learning_rate": 1.975975975975976e-06,
      "loss": 0.2393,
      "step": 80160
    },
    {
      "epoch": 240.75,
      "grad_norm": 10.524264335632324,
      "learning_rate": 1.9749749749749754e-06,
      "loss": 0.1854,
      "step": 80170
    },
    {
      "epoch": 240.78,
      "grad_norm": 10.895743370056152,
      "learning_rate": 1.973973973973974e-06,
      "loss": 0.2102,
      "step": 80180
    },
    {
      "epoch": 240.81,
      "grad_norm": 9.989991188049316,
      "learning_rate": 1.9729729729729734e-06,
      "loss": 0.1633,
      "step": 80190
    },
    {
      "epoch": 240.84,
      "grad_norm": 56.83787155151367,
      "learning_rate": 1.971971971971972e-06,
      "loss": 0.2079,
      "step": 80200
    },
    {
      "epoch": 240.87,
      "grad_norm": 5.642843723297119,
      "learning_rate": 1.9709709709709714e-06,
      "loss": 0.2047,
      "step": 80210
    },
    {
      "epoch": 240.9,
      "grad_norm": 8.250016212463379,
      "learning_rate": 1.96996996996997e-06,
      "loss": 0.1777,
      "step": 80220
    },
    {
      "epoch": 240.93,
      "grad_norm": 7.632354259490967,
      "learning_rate": 1.968968968968969e-06,
      "loss": 0.1993,
      "step": 80230
    },
    {
      "epoch": 240.96,
      "grad_norm": 33.5927848815918,
      "learning_rate": 1.967967967967968e-06,
      "loss": 0.1638,
      "step": 80240
    },
    {
      "epoch": 240.99,
      "grad_norm": 7.614548206329346,
      "learning_rate": 1.966966966966967e-06,
      "loss": 0.2182,
      "step": 80250
    },
    {
      "epoch": 241.0,
      "eval_accuracy": 0.9218,
      "eval_loss": 0.3483900725841522,
      "eval_runtime": 29.8674,
      "eval_samples_per_second": 334.813,
      "eval_steps_per_second": 1.339,
      "step": 80253
    },
    {
      "epoch": 241.02,
      "grad_norm": 9.144424438476562,
      "learning_rate": 1.965965965965966e-06,
      "loss": 0.1957,
      "step": 80260
    },
    {
      "epoch": 241.05,
      "grad_norm": 22.787656784057617,
      "learning_rate": 1.964964964964965e-06,
      "loss": 0.1933,
      "step": 80270
    },
    {
      "epoch": 241.08,
      "grad_norm": 4.539623260498047,
      "learning_rate": 1.963963963963964e-06,
      "loss": 0.2139,
      "step": 80280
    },
    {
      "epoch": 241.11,
      "grad_norm": 10.329512596130371,
      "learning_rate": 1.962962962962963e-06,
      "loss": 0.1955,
      "step": 80290
    },
    {
      "epoch": 241.14,
      "grad_norm": 9.444132804870605,
      "learning_rate": 1.961961961961962e-06,
      "loss": 0.1958,
      "step": 80300
    },
    {
      "epoch": 241.17,
      "grad_norm": 6.66431188583374,
      "learning_rate": 1.960960960960961e-06,
      "loss": 0.1935,
      "step": 80310
    },
    {
      "epoch": 241.2,
      "grad_norm": 9.24359130859375,
      "learning_rate": 1.9599599599599602e-06,
      "loss": 0.1951,
      "step": 80320
    },
    {
      "epoch": 241.23,
      "grad_norm": 16.12553596496582,
      "learning_rate": 1.9589589589589592e-06,
      "loss": 0.184,
      "step": 80330
    },
    {
      "epoch": 241.26,
      "grad_norm": 5.7978715896606445,
      "learning_rate": 1.957957957957958e-06,
      "loss": 0.1942,
      "step": 80340
    },
    {
      "epoch": 241.29,
      "grad_norm": 27.367225646972656,
      "learning_rate": 1.9569569569569573e-06,
      "loss": 0.1982,
      "step": 80350
    },
    {
      "epoch": 241.32,
      "grad_norm": 7.455659866333008,
      "learning_rate": 1.955955955955956e-06,
      "loss": 0.1671,
      "step": 80360
    },
    {
      "epoch": 241.35,
      "grad_norm": 6.168929576873779,
      "learning_rate": 1.9549549549549553e-06,
      "loss": 0.1828,
      "step": 80370
    },
    {
      "epoch": 241.38,
      "grad_norm": 6.95542573928833,
      "learning_rate": 1.953953953953954e-06,
      "loss": 0.1828,
      "step": 80380
    },
    {
      "epoch": 241.41,
      "grad_norm": 12.507369995117188,
      "learning_rate": 1.952952952952953e-06,
      "loss": 0.1858,
      "step": 80390
    },
    {
      "epoch": 241.44,
      "grad_norm": 6.130943775177002,
      "learning_rate": 1.951951951951952e-06,
      "loss": 0.1696,
      "step": 80400
    },
    {
      "epoch": 241.47,
      "grad_norm": 7.558726787567139,
      "learning_rate": 1.950950950950951e-06,
      "loss": 0.1925,
      "step": 80410
    },
    {
      "epoch": 241.5,
      "grad_norm": 6.965606212615967,
      "learning_rate": 1.9499499499499504e-06,
      "loss": 0.1849,
      "step": 80420
    },
    {
      "epoch": 241.53,
      "grad_norm": 6.944590091705322,
      "learning_rate": 1.948948948948949e-06,
      "loss": 0.1715,
      "step": 80430
    },
    {
      "epoch": 241.56,
      "grad_norm": 8.82198715209961,
      "learning_rate": 1.947947947947948e-06,
      "loss": 0.1732,
      "step": 80440
    },
    {
      "epoch": 241.59,
      "grad_norm": 21.59174156188965,
      "learning_rate": 1.946946946946947e-06,
      "loss": 0.1908,
      "step": 80450
    },
    {
      "epoch": 241.62,
      "grad_norm": 11.463054656982422,
      "learning_rate": 1.945945945945946e-06,
      "loss": 0.1875,
      "step": 80460
    },
    {
      "epoch": 241.65,
      "grad_norm": 6.7346673011779785,
      "learning_rate": 1.944944944944945e-06,
      "loss": 0.1756,
      "step": 80470
    },
    {
      "epoch": 241.68,
      "grad_norm": 8.585199356079102,
      "learning_rate": 1.943943943943944e-06,
      "loss": 0.192,
      "step": 80480
    },
    {
      "epoch": 241.71,
      "grad_norm": 13.282663345336914,
      "learning_rate": 1.942942942942943e-06,
      "loss": 0.1912,
      "step": 80490
    },
    {
      "epoch": 241.74,
      "grad_norm": 8.062803268432617,
      "learning_rate": 1.941941941941942e-06,
      "loss": 0.2265,
      "step": 80500
    },
    {
      "epoch": 241.77,
      "grad_norm": 8.213406562805176,
      "learning_rate": 1.940940940940941e-06,
      "loss": 0.2179,
      "step": 80510
    },
    {
      "epoch": 241.8,
      "grad_norm": 8.45276927947998,
      "learning_rate": 1.93993993993994e-06,
      "loss": 0.1651,
      "step": 80520
    },
    {
      "epoch": 241.83,
      "grad_norm": 7.337998867034912,
      "learning_rate": 1.938938938938939e-06,
      "loss": 0.2035,
      "step": 80530
    },
    {
      "epoch": 241.86,
      "grad_norm": 99.19469451904297,
      "learning_rate": 1.937937937937938e-06,
      "loss": 0.2187,
      "step": 80540
    },
    {
      "epoch": 241.89,
      "grad_norm": 7.189026832580566,
      "learning_rate": 1.9369369369369372e-06,
      "loss": 0.1918,
      "step": 80550
    },
    {
      "epoch": 241.92,
      "grad_norm": 6.732546806335449,
      "learning_rate": 1.9359359359359362e-06,
      "loss": 0.1824,
      "step": 80560
    },
    {
      "epoch": 241.95,
      "grad_norm": 7.350929260253906,
      "learning_rate": 1.934934934934935e-06,
      "loss": 0.2067,
      "step": 80570
    },
    {
      "epoch": 241.98,
      "grad_norm": 7.2217631340026855,
      "learning_rate": 1.9339339339339343e-06,
      "loss": 0.2084,
      "step": 80580
    },
    {
      "epoch": 242.0,
      "eval_accuracy": 0.9222,
      "eval_loss": 0.35150396823883057,
      "eval_runtime": 30.3039,
      "eval_samples_per_second": 329.991,
      "eval_steps_per_second": 1.32,
      "step": 80586
    },
    {
      "epoch": 242.01,
      "grad_norm": 7.4541497230529785,
      "learning_rate": 1.932932932932933e-06,
      "loss": 0.2101,
      "step": 80590
    },
    {
      "epoch": 242.04,
      "grad_norm": 8.98868465423584,
      "learning_rate": 1.9319319319319323e-06,
      "loss": 0.1682,
      "step": 80600
    },
    {
      "epoch": 242.07,
      "grad_norm": 5.549026966094971,
      "learning_rate": 1.930930930930931e-06,
      "loss": 0.1748,
      "step": 80610
    },
    {
      "epoch": 242.1,
      "grad_norm": 8.307046890258789,
      "learning_rate": 1.92992992992993e-06,
      "loss": 0.2078,
      "step": 80620
    },
    {
      "epoch": 242.13,
      "grad_norm": 5.902565956115723,
      "learning_rate": 1.928928928928929e-06,
      "loss": 0.2066,
      "step": 80630
    },
    {
      "epoch": 242.16,
      "grad_norm": 9.077995300292969,
      "learning_rate": 1.927927927927928e-06,
      "loss": 0.1728,
      "step": 80640
    },
    {
      "epoch": 242.19,
      "grad_norm": 6.287807941436768,
      "learning_rate": 1.926926926926927e-06,
      "loss": 0.172,
      "step": 80650
    },
    {
      "epoch": 242.22,
      "grad_norm": 8.406379699707031,
      "learning_rate": 1.925925925925926e-06,
      "loss": 0.2218,
      "step": 80660
    },
    {
      "epoch": 242.25,
      "grad_norm": 9.284985542297363,
      "learning_rate": 1.924924924924925e-06,
      "loss": 0.1891,
      "step": 80670
    },
    {
      "epoch": 242.28,
      "grad_norm": 8.868535041809082,
      "learning_rate": 1.923923923923924e-06,
      "loss": 0.2086,
      "step": 80680
    },
    {
      "epoch": 242.31,
      "grad_norm": 7.43398380279541,
      "learning_rate": 1.922922922922923e-06,
      "loss": 0.2071,
      "step": 80690
    },
    {
      "epoch": 242.34,
      "grad_norm": 9.87989616394043,
      "learning_rate": 1.921921921921922e-06,
      "loss": 0.2098,
      "step": 80700
    },
    {
      "epoch": 242.37,
      "grad_norm": 7.867367744445801,
      "learning_rate": 1.920920920920921e-06,
      "loss": 0.2408,
      "step": 80710
    },
    {
      "epoch": 242.4,
      "grad_norm": 10.339460372924805,
      "learning_rate": 1.91991991991992e-06,
      "loss": 0.1984,
      "step": 80720
    },
    {
      "epoch": 242.43,
      "grad_norm": 9.265618324279785,
      "learning_rate": 1.918918918918919e-06,
      "loss": 0.2016,
      "step": 80730
    },
    {
      "epoch": 242.46,
      "grad_norm": 5.279520034790039,
      "learning_rate": 1.917917917917918e-06,
      "loss": 0.1782,
      "step": 80740
    },
    {
      "epoch": 242.49,
      "grad_norm": 9.794772148132324,
      "learning_rate": 1.916916916916917e-06,
      "loss": 0.181,
      "step": 80750
    },
    {
      "epoch": 242.52,
      "grad_norm": 6.455809593200684,
      "learning_rate": 1.915915915915916e-06,
      "loss": 0.2063,
      "step": 80760
    },
    {
      "epoch": 242.55,
      "grad_norm": 9.485993385314941,
      "learning_rate": 1.914914914914915e-06,
      "loss": 0.1872,
      "step": 80770
    },
    {
      "epoch": 242.58,
      "grad_norm": 9.382390022277832,
      "learning_rate": 1.9139139139139142e-06,
      "loss": 0.2287,
      "step": 80780
    },
    {
      "epoch": 242.61,
      "grad_norm": 9.449283599853516,
      "learning_rate": 1.9129129129129132e-06,
      "loss": 0.1788,
      "step": 80790
    },
    {
      "epoch": 242.64,
      "grad_norm": 8.669525146484375,
      "learning_rate": 1.911911911911912e-06,
      "loss": 0.2541,
      "step": 80800
    },
    {
      "epoch": 242.67,
      "grad_norm": 10.613685607910156,
      "learning_rate": 1.9109109109109113e-06,
      "loss": 0.2067,
      "step": 80810
    },
    {
      "epoch": 242.7,
      "grad_norm": 15.408585548400879,
      "learning_rate": 1.90990990990991e-06,
      "loss": 0.1821,
      "step": 80820
    },
    {
      "epoch": 242.73,
      "grad_norm": 20.41518211364746,
      "learning_rate": 1.9089089089089093e-06,
      "loss": 0.2089,
      "step": 80830
    },
    {
      "epoch": 242.76,
      "grad_norm": 10.093743324279785,
      "learning_rate": 1.907907907907908e-06,
      "loss": 0.1824,
      "step": 80840
    },
    {
      "epoch": 242.79,
      "grad_norm": 11.704672813415527,
      "learning_rate": 1.9069069069069071e-06,
      "loss": 0.2073,
      "step": 80850
    },
    {
      "epoch": 242.82,
      "grad_norm": 6.6625165939331055,
      "learning_rate": 1.905905905905906e-06,
      "loss": 0.1895,
      "step": 80860
    },
    {
      "epoch": 242.85,
      "grad_norm": 13.250521659851074,
      "learning_rate": 1.9049049049049052e-06,
      "loss": 0.1956,
      "step": 80870
    },
    {
      "epoch": 242.88,
      "grad_norm": 6.500404357910156,
      "learning_rate": 1.903903903903904e-06,
      "loss": 0.1866,
      "step": 80880
    },
    {
      "epoch": 242.91,
      "grad_norm": 15.701628684997559,
      "learning_rate": 1.902902902902903e-06,
      "loss": 0.1617,
      "step": 80890
    },
    {
      "epoch": 242.94,
      "grad_norm": 8.276241302490234,
      "learning_rate": 1.901901901901902e-06,
      "loss": 0.2155,
      "step": 80900
    },
    {
      "epoch": 242.97,
      "grad_norm": 7.726240634918213,
      "learning_rate": 1.900900900900901e-06,
      "loss": 0.2006,
      "step": 80910
    },
    {
      "epoch": 243.0,
      "eval_accuracy": 0.9223,
      "eval_loss": 0.34985384345054626,
      "eval_runtime": 30.6089,
      "eval_samples_per_second": 326.702,
      "eval_steps_per_second": 1.307,
      "step": 80919
    },
    {
      "epoch": 243.0,
      "grad_norm": 6.6773295402526855,
      "learning_rate": 1.8998998998998998e-06,
      "loss": 0.1707,
      "step": 80920
    },
    {
      "epoch": 243.03,
      "grad_norm": 7.043368339538574,
      "learning_rate": 1.898898898898899e-06,
      "loss": 0.2469,
      "step": 80930
    },
    {
      "epoch": 243.06,
      "grad_norm": 8.311963081359863,
      "learning_rate": 1.897897897897898e-06,
      "loss": 0.1945,
      "step": 80940
    },
    {
      "epoch": 243.09,
      "grad_norm": 12.459003448486328,
      "learning_rate": 1.896896896896897e-06,
      "loss": 0.1787,
      "step": 80950
    },
    {
      "epoch": 243.12,
      "grad_norm": 10.953930854797363,
      "learning_rate": 1.8958958958958961e-06,
      "loss": 0.1952,
      "step": 80960
    },
    {
      "epoch": 243.15,
      "grad_norm": 5.3934550285339355,
      "learning_rate": 1.894894894894895e-06,
      "loss": 0.2127,
      "step": 80970
    },
    {
      "epoch": 243.18,
      "grad_norm": 9.773137092590332,
      "learning_rate": 1.8938938938938942e-06,
      "loss": 0.1776,
      "step": 80980
    },
    {
      "epoch": 243.21,
      "grad_norm": 9.869041442871094,
      "learning_rate": 1.892892892892893e-06,
      "loss": 0.1935,
      "step": 80990
    },
    {
      "epoch": 243.24,
      "grad_norm": 10.548992156982422,
      "learning_rate": 1.8918918918918922e-06,
      "loss": 0.2016,
      "step": 81000
    },
    {
      "epoch": 243.27,
      "grad_norm": 21.37706756591797,
      "learning_rate": 1.890890890890891e-06,
      "loss": 0.2193,
      "step": 81010
    },
    {
      "epoch": 243.3,
      "grad_norm": 5.590071201324463,
      "learning_rate": 1.8898898898898902e-06,
      "loss": 0.1688,
      "step": 81020
    },
    {
      "epoch": 243.33,
      "grad_norm": 6.1485419273376465,
      "learning_rate": 1.888888888888889e-06,
      "loss": 0.1502,
      "step": 81030
    },
    {
      "epoch": 243.36,
      "grad_norm": 8.387727737426758,
      "learning_rate": 1.887887887887888e-06,
      "loss": 0.1922,
      "step": 81040
    },
    {
      "epoch": 243.39,
      "grad_norm": 7.218709945678711,
      "learning_rate": 1.886886886886887e-06,
      "loss": 0.185,
      "step": 81050
    },
    {
      "epoch": 243.42,
      "grad_norm": 14.886263847351074,
      "learning_rate": 1.885885885885886e-06,
      "loss": 0.196,
      "step": 81060
    },
    {
      "epoch": 243.45,
      "grad_norm": 7.264105319976807,
      "learning_rate": 1.884884884884885e-06,
      "loss": 0.1944,
      "step": 81070
    },
    {
      "epoch": 243.48,
      "grad_norm": 5.530055046081543,
      "learning_rate": 1.8838838838838841e-06,
      "loss": 0.2031,
      "step": 81080
    },
    {
      "epoch": 243.51,
      "grad_norm": 9.746648788452148,
      "learning_rate": 1.882882882882883e-06,
      "loss": 0.1961,
      "step": 81090
    },
    {
      "epoch": 243.54,
      "grad_norm": 9.778495788574219,
      "learning_rate": 1.8818818818818822e-06,
      "loss": 0.2175,
      "step": 81100
    },
    {
      "epoch": 243.57,
      "grad_norm": 6.119612693786621,
      "learning_rate": 1.880880880880881e-06,
      "loss": 0.1967,
      "step": 81110
    },
    {
      "epoch": 243.6,
      "grad_norm": 10.222867012023926,
      "learning_rate": 1.87987987987988e-06,
      "loss": 0.1503,
      "step": 81120
    },
    {
      "epoch": 243.63,
      "grad_norm": 7.553834438323975,
      "learning_rate": 1.878878878878879e-06,
      "loss": 0.2072,
      "step": 81130
    },
    {
      "epoch": 243.66,
      "grad_norm": 7.55220890045166,
      "learning_rate": 1.877877877877878e-06,
      "loss": 0.1365,
      "step": 81140
    },
    {
      "epoch": 243.69,
      "grad_norm": 13.188750267028809,
      "learning_rate": 1.8768768768768768e-06,
      "loss": 0.208,
      "step": 81150
    },
    {
      "epoch": 243.72,
      "grad_norm": 7.451780796051025,
      "learning_rate": 1.875875875875876e-06,
      "loss": 0.2347,
      "step": 81160
    },
    {
      "epoch": 243.75,
      "grad_norm": 10.676434516906738,
      "learning_rate": 1.8748748748748749e-06,
      "loss": 0.215,
      "step": 81170
    },
    {
      "epoch": 243.78,
      "grad_norm": 8.939967155456543,
      "learning_rate": 1.873873873873874e-06,
      "loss": 0.197,
      "step": 81180
    },
    {
      "epoch": 243.81,
      "grad_norm": 9.767823219299316,
      "learning_rate": 1.872872872872873e-06,
      "loss": 0.196,
      "step": 81190
    },
    {
      "epoch": 243.84,
      "grad_norm": 8.432971954345703,
      "learning_rate": 1.871871871871872e-06,
      "loss": 0.1959,
      "step": 81200
    },
    {
      "epoch": 243.87,
      "grad_norm": 3.9382998943328857,
      "learning_rate": 1.8708708708708712e-06,
      "loss": 0.1888,
      "step": 81210
    },
    {
      "epoch": 243.9,
      "grad_norm": 12.998598098754883,
      "learning_rate": 1.86986986986987e-06,
      "loss": 0.2198,
      "step": 81220
    },
    {
      "epoch": 243.93,
      "grad_norm": 15.163290023803711,
      "learning_rate": 1.8688688688688692e-06,
      "loss": 0.1878,
      "step": 81230
    },
    {
      "epoch": 243.96,
      "grad_norm": 7.05983829498291,
      "learning_rate": 1.867867867867868e-06,
      "loss": 0.1747,
      "step": 81240
    },
    {
      "epoch": 243.99,
      "grad_norm": 9.408228874206543,
      "learning_rate": 1.8668668668668672e-06,
      "loss": 0.221,
      "step": 81250
    },
    {
      "epoch": 244.0,
      "eval_accuracy": 0.9223,
      "eval_loss": 0.3502088785171509,
      "eval_runtime": 30.3071,
      "eval_samples_per_second": 329.956,
      "eval_steps_per_second": 1.32,
      "step": 81252
    },
    {
      "epoch": 244.02,
      "grad_norm": 7.395842552185059,
      "learning_rate": 1.865865865865866e-06,
      "loss": 0.1729,
      "step": 81260
    },
    {
      "epoch": 244.05,
      "grad_norm": 12.488890647888184,
      "learning_rate": 1.864864864864865e-06,
      "loss": 0.1919,
      "step": 81270
    },
    {
      "epoch": 244.08,
      "grad_norm": 8.765775680541992,
      "learning_rate": 1.863863863863864e-06,
      "loss": 0.1949,
      "step": 81280
    },
    {
      "epoch": 244.11,
      "grad_norm": 7.7351603507995605,
      "learning_rate": 1.862862862862863e-06,
      "loss": 0.191,
      "step": 81290
    },
    {
      "epoch": 244.14,
      "grad_norm": 42.35386657714844,
      "learning_rate": 1.861861861861862e-06,
      "loss": 0.1553,
      "step": 81300
    },
    {
      "epoch": 244.17,
      "grad_norm": 9.823132514953613,
      "learning_rate": 1.8608608608608611e-06,
      "loss": 0.2064,
      "step": 81310
    },
    {
      "epoch": 244.2,
      "grad_norm": 6.402418613433838,
      "learning_rate": 1.85985985985986e-06,
      "loss": 0.2086,
      "step": 81320
    },
    {
      "epoch": 244.23,
      "grad_norm": 9.069867134094238,
      "learning_rate": 1.8588588588588592e-06,
      "loss": 0.2004,
      "step": 81330
    },
    {
      "epoch": 244.26,
      "grad_norm": 10.044825553894043,
      "learning_rate": 1.857857857857858e-06,
      "loss": 0.1776,
      "step": 81340
    },
    {
      "epoch": 244.29,
      "grad_norm": 9.58326530456543,
      "learning_rate": 1.856856856856857e-06,
      "loss": 0.1799,
      "step": 81350
    },
    {
      "epoch": 244.32,
      "grad_norm": 11.311469078063965,
      "learning_rate": 1.855855855855856e-06,
      "loss": 0.1931,
      "step": 81360
    },
    {
      "epoch": 244.35,
      "grad_norm": 6.825306415557861,
      "learning_rate": 1.854854854854855e-06,
      "loss": 0.1913,
      "step": 81370
    },
    {
      "epoch": 244.38,
      "grad_norm": 10.696843147277832,
      "learning_rate": 1.8538538538538538e-06,
      "loss": 0.2061,
      "step": 81380
    },
    {
      "epoch": 244.41,
      "grad_norm": 7.799194812774658,
      "learning_rate": 1.852852852852853e-06,
      "loss": 0.2141,
      "step": 81390
    },
    {
      "epoch": 244.44,
      "grad_norm": 7.540597438812256,
      "learning_rate": 1.8518518518518519e-06,
      "loss": 0.2174,
      "step": 81400
    },
    {
      "epoch": 244.47,
      "grad_norm": 5.775651454925537,
      "learning_rate": 1.850850850850851e-06,
      "loss": 0.1976,
      "step": 81410
    },
    {
      "epoch": 244.5,
      "grad_norm": 10.367962837219238,
      "learning_rate": 1.84984984984985e-06,
      "loss": 0.1991,
      "step": 81420
    },
    {
      "epoch": 244.53,
      "grad_norm": 9.032089233398438,
      "learning_rate": 1.848848848848849e-06,
      "loss": 0.1453,
      "step": 81430
    },
    {
      "epoch": 244.56,
      "grad_norm": 7.541386604309082,
      "learning_rate": 1.847847847847848e-06,
      "loss": 0.1812,
      "step": 81440
    },
    {
      "epoch": 244.59,
      "grad_norm": 10.58566951751709,
      "learning_rate": 1.846846846846847e-06,
      "loss": 0.2016,
      "step": 81450
    },
    {
      "epoch": 244.62,
      "grad_norm": 9.284521102905273,
      "learning_rate": 1.8458458458458458e-06,
      "loss": 0.1824,
      "step": 81460
    },
    {
      "epoch": 244.65,
      "grad_norm": 6.162008762359619,
      "learning_rate": 1.844844844844845e-06,
      "loss": 0.1985,
      "step": 81470
    },
    {
      "epoch": 244.68,
      "grad_norm": 7.373632907867432,
      "learning_rate": 1.8438438438438442e-06,
      "loss": 0.1958,
      "step": 81480
    },
    {
      "epoch": 244.71,
      "grad_norm": 7.610928535461426,
      "learning_rate": 1.842842842842843e-06,
      "loss": 0.1886,
      "step": 81490
    },
    {
      "epoch": 244.74,
      "grad_norm": 11.813370704650879,
      "learning_rate": 1.841841841841842e-06,
      "loss": 0.2048,
      "step": 81500
    },
    {
      "epoch": 244.77,
      "grad_norm": 28.065319061279297,
      "learning_rate": 1.840840840840841e-06,
      "loss": 0.1861,
      "step": 81510
    },
    {
      "epoch": 244.8,
      "grad_norm": 23.484426498413086,
      "learning_rate": 1.83983983983984e-06,
      "loss": 0.2065,
      "step": 81520
    },
    {
      "epoch": 244.83,
      "grad_norm": 10.81051254272461,
      "learning_rate": 1.838838838838839e-06,
      "loss": 0.2149,
      "step": 81530
    },
    {
      "epoch": 244.86,
      "grad_norm": 10.764946937561035,
      "learning_rate": 1.8378378378378381e-06,
      "loss": 0.1964,
      "step": 81540
    },
    {
      "epoch": 244.89,
      "grad_norm": 8.3318510055542,
      "learning_rate": 1.836836836836837e-06,
      "loss": 0.1821,
      "step": 81550
    },
    {
      "epoch": 244.92,
      "grad_norm": 4.065110206604004,
      "learning_rate": 1.8358358358358362e-06,
      "loss": 0.1625,
      "step": 81560
    },
    {
      "epoch": 244.95,
      "grad_norm": 6.834167957305908,
      "learning_rate": 1.834834834834835e-06,
      "loss": 0.1478,
      "step": 81570
    },
    {
      "epoch": 244.98,
      "grad_norm": 12.001399040222168,
      "learning_rate": 1.833833833833834e-06,
      "loss": 0.1835,
      "step": 81580
    },
    {
      "epoch": 245.0,
      "eval_accuracy": 0.9212,
      "eval_loss": 0.35256287455558777,
      "eval_runtime": 29.7254,
      "eval_samples_per_second": 336.412,
      "eval_steps_per_second": 1.346,
      "step": 81585
    },
    {
      "epoch": 245.02,
      "grad_norm": 6.807450771331787,
      "learning_rate": 1.832832832832833e-06,
      "loss": 0.2188,
      "step": 81590
    },
    {
      "epoch": 245.05,
      "grad_norm": 8.643406867980957,
      "learning_rate": 1.831831831831832e-06,
      "loss": 0.1745,
      "step": 81600
    },
    {
      "epoch": 245.08,
      "grad_norm": 4.670201778411865,
      "learning_rate": 1.8308308308308308e-06,
      "loss": 0.1841,
      "step": 81610
    },
    {
      "epoch": 245.11,
      "grad_norm": 6.1473894119262695,
      "learning_rate": 1.82982982982983e-06,
      "loss": 0.2042,
      "step": 81620
    },
    {
      "epoch": 245.14,
      "grad_norm": 10.868907928466797,
      "learning_rate": 1.8288288288288289e-06,
      "loss": 0.2198,
      "step": 81630
    },
    {
      "epoch": 245.17,
      "grad_norm": 8.870309829711914,
      "learning_rate": 1.827827827827828e-06,
      "loss": 0.179,
      "step": 81640
    },
    {
      "epoch": 245.2,
      "grad_norm": 7.887211799621582,
      "learning_rate": 1.826826826826827e-06,
      "loss": 0.1914,
      "step": 81650
    },
    {
      "epoch": 245.23,
      "grad_norm": 15.313419342041016,
      "learning_rate": 1.825825825825826e-06,
      "loss": 0.1556,
      "step": 81660
    },
    {
      "epoch": 245.26,
      "grad_norm": 7.164863109588623,
      "learning_rate": 1.824824824824825e-06,
      "loss": 0.1465,
      "step": 81670
    },
    {
      "epoch": 245.29,
      "grad_norm": 10.705216407775879,
      "learning_rate": 1.823823823823824e-06,
      "loss": 0.18,
      "step": 81680
    },
    {
      "epoch": 245.32,
      "grad_norm": 5.923380374908447,
      "learning_rate": 1.8228228228228228e-06,
      "loss": 0.2041,
      "step": 81690
    },
    {
      "epoch": 245.35,
      "grad_norm": 13.973617553710938,
      "learning_rate": 1.821821821821822e-06,
      "loss": 0.2187,
      "step": 81700
    },
    {
      "epoch": 245.38,
      "grad_norm": 7.7829694747924805,
      "learning_rate": 1.8208208208208208e-06,
      "loss": 0.244,
      "step": 81710
    },
    {
      "epoch": 245.41,
      "grad_norm": 14.379873275756836,
      "learning_rate": 1.81981981981982e-06,
      "loss": 0.1949,
      "step": 81720
    },
    {
      "epoch": 245.44,
      "grad_norm": 6.469885349273682,
      "learning_rate": 1.8188188188188188e-06,
      "loss": 0.1868,
      "step": 81730
    },
    {
      "epoch": 245.47,
      "grad_norm": 7.877725601196289,
      "learning_rate": 1.817817817817818e-06,
      "loss": 0.2127,
      "step": 81740
    },
    {
      "epoch": 245.5,
      "grad_norm": 8.92708683013916,
      "learning_rate": 1.816816816816817e-06,
      "loss": 0.1519,
      "step": 81750
    },
    {
      "epoch": 245.53,
      "grad_norm": 21.06083869934082,
      "learning_rate": 1.815815815815816e-06,
      "loss": 0.1901,
      "step": 81760
    },
    {
      "epoch": 245.56,
      "grad_norm": 9.618289947509766,
      "learning_rate": 1.8148148148148151e-06,
      "loss": 0.1992,
      "step": 81770
    },
    {
      "epoch": 245.59,
      "grad_norm": 7.9456658363342285,
      "learning_rate": 1.813813813813814e-06,
      "loss": 0.1919,
      "step": 81780
    },
    {
      "epoch": 245.62,
      "grad_norm": 11.614968299865723,
      "learning_rate": 1.8128128128128132e-06,
      "loss": 0.1882,
      "step": 81790
    },
    {
      "epoch": 245.65,
      "grad_norm": 7.037623882293701,
      "learning_rate": 1.811811811811812e-06,
      "loss": 0.1563,
      "step": 81800
    },
    {
      "epoch": 245.68,
      "grad_norm": 8.680950164794922,
      "learning_rate": 1.810810810810811e-06,
      "loss": 0.1541,
      "step": 81810
    },
    {
      "epoch": 245.71,
      "grad_norm": 6.6445512771606445,
      "learning_rate": 1.80980980980981e-06,
      "loss": 0.1834,
      "step": 81820
    },
    {
      "epoch": 245.74,
      "grad_norm": 14.939486503601074,
      "learning_rate": 1.808808808808809e-06,
      "loss": 0.1904,
      "step": 81830
    },
    {
      "epoch": 245.77,
      "grad_norm": 10.652714729309082,
      "learning_rate": 1.8078078078078078e-06,
      "loss": 0.1865,
      "step": 81840
    },
    {
      "epoch": 245.8,
      "grad_norm": 10.957291603088379,
      "learning_rate": 1.806806806806807e-06,
      "loss": 0.1903,
      "step": 81850
    },
    {
      "epoch": 245.83,
      "grad_norm": 5.049372673034668,
      "learning_rate": 1.8058058058058059e-06,
      "loss": 0.1626,
      "step": 81860
    },
    {
      "epoch": 245.86,
      "grad_norm": 6.235400676727295,
      "learning_rate": 1.804804804804805e-06,
      "loss": 0.2171,
      "step": 81870
    },
    {
      "epoch": 245.89,
      "grad_norm": 7.803573131561279,
      "learning_rate": 1.803803803803804e-06,
      "loss": 0.187,
      "step": 81880
    },
    {
      "epoch": 245.92,
      "grad_norm": 6.949105262756348,
      "learning_rate": 1.802802802802803e-06,
      "loss": 0.1875,
      "step": 81890
    },
    {
      "epoch": 245.95,
      "grad_norm": 4.648994445800781,
      "learning_rate": 1.801801801801802e-06,
      "loss": 0.1984,
      "step": 81900
    },
    {
      "epoch": 245.98,
      "grad_norm": 11.811627388000488,
      "learning_rate": 1.800800800800801e-06,
      "loss": 0.2469,
      "step": 81910
    },
    {
      "epoch": 246.0,
      "eval_accuracy": 0.9215,
      "eval_loss": 0.3473447263240814,
      "eval_runtime": 30.3189,
      "eval_samples_per_second": 329.827,
      "eval_steps_per_second": 1.319,
      "step": 81918
    },
    {
      "epoch": 246.01,
      "grad_norm": 11.487900733947754,
      "learning_rate": 1.7997997997997998e-06,
      "loss": 0.1831,
      "step": 81920
    },
    {
      "epoch": 246.04,
      "grad_norm": 8.590137481689453,
      "learning_rate": 1.798798798798799e-06,
      "loss": 0.1499,
      "step": 81930
    },
    {
      "epoch": 246.07,
      "grad_norm": 2.64656662940979,
      "learning_rate": 1.7977977977977978e-06,
      "loss": 0.2015,
      "step": 81940
    },
    {
      "epoch": 246.1,
      "grad_norm": 7.567853927612305,
      "learning_rate": 1.796796796796797e-06,
      "loss": 0.201,
      "step": 81950
    },
    {
      "epoch": 246.13,
      "grad_norm": 4.190827369689941,
      "learning_rate": 1.7957957957957958e-06,
      "loss": 0.1876,
      "step": 81960
    },
    {
      "epoch": 246.16,
      "grad_norm": 10.198352813720703,
      "learning_rate": 1.7947947947947949e-06,
      "loss": 0.2,
      "step": 81970
    },
    {
      "epoch": 246.19,
      "grad_norm": 8.431456565856934,
      "learning_rate": 1.7937937937937939e-06,
      "loss": 0.1884,
      "step": 81980
    },
    {
      "epoch": 246.22,
      "grad_norm": 11.77056884765625,
      "learning_rate": 1.7927927927927929e-06,
      "loss": 0.1655,
      "step": 81990
    },
    {
      "epoch": 246.25,
      "grad_norm": 10.530085563659668,
      "learning_rate": 1.7917917917917917e-06,
      "loss": 0.1917,
      "step": 82000
    },
    {
      "epoch": 246.28,
      "grad_norm": 7.093303680419922,
      "learning_rate": 1.790790790790791e-06,
      "loss": 0.1693,
      "step": 82010
    },
    {
      "epoch": 246.31,
      "grad_norm": 8.447652816772461,
      "learning_rate": 1.7897897897897902e-06,
      "loss": 0.1914,
      "step": 82020
    },
    {
      "epoch": 246.34,
      "grad_norm": 10.54896068572998,
      "learning_rate": 1.788788788788789e-06,
      "loss": 0.1924,
      "step": 82030
    },
    {
      "epoch": 246.37,
      "grad_norm": 8.842206954956055,
      "learning_rate": 1.787787787787788e-06,
      "loss": 0.1674,
      "step": 82040
    },
    {
      "epoch": 246.4,
      "grad_norm": 7.691274642944336,
      "learning_rate": 1.786786786786787e-06,
      "loss": 0.2116,
      "step": 82050
    },
    {
      "epoch": 246.43,
      "grad_norm": 4.772304058074951,
      "learning_rate": 1.785785785785786e-06,
      "loss": 0.1747,
      "step": 82060
    },
    {
      "epoch": 246.46,
      "grad_norm": 9.3211030960083,
      "learning_rate": 1.7847847847847848e-06,
      "loss": 0.1925,
      "step": 82070
    },
    {
      "epoch": 246.49,
      "grad_norm": 6.52603006362915,
      "learning_rate": 1.783783783783784e-06,
      "loss": 0.1673,
      "step": 82080
    },
    {
      "epoch": 246.52,
      "grad_norm": 6.716470718383789,
      "learning_rate": 1.7827827827827829e-06,
      "loss": 0.2214,
      "step": 82090
    },
    {
      "epoch": 246.55,
      "grad_norm": 11.4483642578125,
      "learning_rate": 1.781781781781782e-06,
      "loss": 0.2122,
      "step": 82100
    },
    {
      "epoch": 246.58,
      "grad_norm": 12.311321258544922,
      "learning_rate": 1.780780780780781e-06,
      "loss": 0.1694,
      "step": 82110
    },
    {
      "epoch": 246.61,
      "grad_norm": 5.814987659454346,
      "learning_rate": 1.77977977977978e-06,
      "loss": 0.2145,
      "step": 82120
    },
    {
      "epoch": 246.64,
      "grad_norm": 5.882806301116943,
      "learning_rate": 1.778778778778779e-06,
      "loss": 0.1844,
      "step": 82130
    },
    {
      "epoch": 246.67,
      "grad_norm": 11.198575019836426,
      "learning_rate": 1.777777777777778e-06,
      "loss": 0.2408,
      "step": 82140
    },
    {
      "epoch": 246.7,
      "grad_norm": 11.964673042297363,
      "learning_rate": 1.7767767767767768e-06,
      "loss": 0.1699,
      "step": 82150
    },
    {
      "epoch": 246.73,
      "grad_norm": 7.000456809997559,
      "learning_rate": 1.775775775775776e-06,
      "loss": 0.2284,
      "step": 82160
    },
    {
      "epoch": 246.76,
      "grad_norm": 6.472318649291992,
      "learning_rate": 1.7747747747747748e-06,
      "loss": 0.1818,
      "step": 82170
    },
    {
      "epoch": 246.79,
      "grad_norm": 11.842066764831543,
      "learning_rate": 1.773773773773774e-06,
      "loss": 0.2027,
      "step": 82180
    },
    {
      "epoch": 246.82,
      "grad_norm": 6.202572345733643,
      "learning_rate": 1.7727727727727728e-06,
      "loss": 0.1892,
      "step": 82190
    },
    {
      "epoch": 246.85,
      "grad_norm": 6.512704849243164,
      "learning_rate": 1.7717717717717719e-06,
      "loss": 0.1785,
      "step": 82200
    },
    {
      "epoch": 246.88,
      "grad_norm": 9.116239547729492,
      "learning_rate": 1.7707707707707709e-06,
      "loss": 0.1949,
      "step": 82210
    },
    {
      "epoch": 246.91,
      "grad_norm": 9.982625961303711,
      "learning_rate": 1.7697697697697699e-06,
      "loss": 0.1809,
      "step": 82220
    },
    {
      "epoch": 246.94,
      "grad_norm": 9.559281349182129,
      "learning_rate": 1.7687687687687687e-06,
      "loss": 0.1747,
      "step": 82230
    },
    {
      "epoch": 246.97,
      "grad_norm": 8.356477737426758,
      "learning_rate": 1.767767767767768e-06,
      "loss": 0.204,
      "step": 82240
    },
    {
      "epoch": 247.0,
      "grad_norm": 11.921745300292969,
      "learning_rate": 1.7667667667667667e-06,
      "loss": 0.1844,
      "step": 82250
    },
    {
      "epoch": 247.0,
      "eval_accuracy": 0.9228,
      "eval_loss": 0.3472795784473419,
      "eval_runtime": 30.3391,
      "eval_samples_per_second": 329.608,
      "eval_steps_per_second": 1.318,
      "step": 82251
    },
    {
      "epoch": 247.03,
      "grad_norm": 14.630331993103027,
      "learning_rate": 1.765765765765766e-06,
      "loss": 0.462,
      "step": 82260
    },
    {
      "epoch": 247.06,
      "grad_norm": 55.01515197753906,
      "learning_rate": 1.7647647647647648e-06,
      "loss": 0.1909,
      "step": 82270
    },
    {
      "epoch": 247.09,
      "grad_norm": 9.004843711853027,
      "learning_rate": 1.763763763763764e-06,
      "loss": 0.2275,
      "step": 82280
    },
    {
      "epoch": 247.12,
      "grad_norm": 8.229368209838867,
      "learning_rate": 1.762762762762763e-06,
      "loss": 0.1934,
      "step": 82290
    },
    {
      "epoch": 247.15,
      "grad_norm": 7.453774452209473,
      "learning_rate": 1.7617617617617618e-06,
      "loss": 0.184,
      "step": 82300
    },
    {
      "epoch": 247.18,
      "grad_norm": 6.0209479331970215,
      "learning_rate": 1.760760760760761e-06,
      "loss": 0.1808,
      "step": 82310
    },
    {
      "epoch": 247.21,
      "grad_norm": 20.38810157775879,
      "learning_rate": 1.7597597597597599e-06,
      "loss": 0.1916,
      "step": 82320
    },
    {
      "epoch": 247.24,
      "grad_norm": 7.868497848510742,
      "learning_rate": 1.758758758758759e-06,
      "loss": 0.1965,
      "step": 82330
    },
    {
      "epoch": 247.27,
      "grad_norm": 7.831243515014648,
      "learning_rate": 1.757757757757758e-06,
      "loss": 0.1962,
      "step": 82340
    },
    {
      "epoch": 247.3,
      "grad_norm": 8.840407371520996,
      "learning_rate": 1.756756756756757e-06,
      "loss": 0.1568,
      "step": 82350
    },
    {
      "epoch": 247.33,
      "grad_norm": 7.31623649597168,
      "learning_rate": 1.755755755755756e-06,
      "loss": 0.1816,
      "step": 82360
    },
    {
      "epoch": 247.36,
      "grad_norm": 6.744307041168213,
      "learning_rate": 1.754754754754755e-06,
      "loss": 0.1826,
      "step": 82370
    },
    {
      "epoch": 247.39,
      "grad_norm": 8.584980964660645,
      "learning_rate": 1.7537537537537538e-06,
      "loss": 0.19,
      "step": 82380
    },
    {
      "epoch": 247.42,
      "grad_norm": 6.131076812744141,
      "learning_rate": 1.752752752752753e-06,
      "loss": 0.1776,
      "step": 82390
    },
    {
      "epoch": 247.45,
      "grad_norm": 7.938048839569092,
      "learning_rate": 1.7517517517517518e-06,
      "loss": 0.2248,
      "step": 82400
    },
    {
      "epoch": 247.48,
      "grad_norm": 12.422223091125488,
      "learning_rate": 1.750750750750751e-06,
      "loss": 0.1821,
      "step": 82410
    },
    {
      "epoch": 247.51,
      "grad_norm": 10.041374206542969,
      "learning_rate": 1.7497497497497498e-06,
      "loss": 0.187,
      "step": 82420
    },
    {
      "epoch": 247.54,
      "grad_norm": 5.3734588623046875,
      "learning_rate": 1.7487487487487489e-06,
      "loss": 0.1682,
      "step": 82430
    },
    {
      "epoch": 247.57,
      "grad_norm": 7.93886661529541,
      "learning_rate": 1.7477477477477479e-06,
      "loss": 0.1812,
      "step": 82440
    },
    {
      "epoch": 247.6,
      "grad_norm": 5.559296131134033,
      "learning_rate": 1.7467467467467469e-06,
      "loss": 0.1941,
      "step": 82450
    },
    {
      "epoch": 247.63,
      "grad_norm": 12.253649711608887,
      "learning_rate": 1.7457457457457457e-06,
      "loss": 0.195,
      "step": 82460
    },
    {
      "epoch": 247.66,
      "grad_norm": 6.433213233947754,
      "learning_rate": 1.744744744744745e-06,
      "loss": 0.1903,
      "step": 82470
    },
    {
      "epoch": 247.69,
      "grad_norm": 9.42849063873291,
      "learning_rate": 1.7437437437437437e-06,
      "loss": 0.2136,
      "step": 82480
    },
    {
      "epoch": 247.72,
      "grad_norm": 11.565710067749023,
      "learning_rate": 1.742742742742743e-06,
      "loss": 0.1976,
      "step": 82490
    },
    {
      "epoch": 247.75,
      "grad_norm": 9.55038833618164,
      "learning_rate": 1.7417417417417418e-06,
      "loss": 0.1729,
      "step": 82500
    },
    {
      "epoch": 247.78,
      "grad_norm": 8.88013744354248,
      "learning_rate": 1.740740740740741e-06,
      "loss": 0.2324,
      "step": 82510
    },
    {
      "epoch": 247.81,
      "grad_norm": 6.600421905517578,
      "learning_rate": 1.7397397397397398e-06,
      "loss": 0.175,
      "step": 82520
    },
    {
      "epoch": 247.84,
      "grad_norm": 10.077155113220215,
      "learning_rate": 1.7387387387387388e-06,
      "loss": 0.2288,
      "step": 82530
    },
    {
      "epoch": 247.87,
      "grad_norm": 24.199609756469727,
      "learning_rate": 1.737737737737738e-06,
      "loss": 0.1848,
      "step": 82540
    },
    {
      "epoch": 247.9,
      "grad_norm": 6.78046178817749,
      "learning_rate": 1.7367367367367369e-06,
      "loss": 0.1834,
      "step": 82550
    },
    {
      "epoch": 247.93,
      "grad_norm": 9.775473594665527,
      "learning_rate": 1.735735735735736e-06,
      "loss": 0.1744,
      "step": 82560
    },
    {
      "epoch": 247.96,
      "grad_norm": 5.0841875076293945,
      "learning_rate": 1.734734734734735e-06,
      "loss": 0.1757,
      "step": 82570
    },
    {
      "epoch": 247.99,
      "grad_norm": 7.3065972328186035,
      "learning_rate": 1.733733733733734e-06,
      "loss": 0.1972,
      "step": 82580
    },
    {
      "epoch": 248.0,
      "eval_accuracy": 0.9213,
      "eval_loss": 0.3493152856826782,
      "eval_runtime": 30.2499,
      "eval_samples_per_second": 330.58,
      "eval_steps_per_second": 1.322,
      "step": 82584
    },
    {
      "epoch": 248.02,
      "grad_norm": 11.192063331604004,
      "learning_rate": 1.732732732732733e-06,
      "loss": 0.1701,
      "step": 82590
    },
    {
      "epoch": 248.05,
      "grad_norm": 6.71538782119751,
      "learning_rate": 1.731731731731732e-06,
      "loss": 0.2065,
      "step": 82600
    },
    {
      "epoch": 248.08,
      "grad_norm": 4.317703723907471,
      "learning_rate": 1.7307307307307308e-06,
      "loss": 0.1818,
      "step": 82610
    },
    {
      "epoch": 248.11,
      "grad_norm": 6.366308689117432,
      "learning_rate": 1.72972972972973e-06,
      "loss": 0.1761,
      "step": 82620
    },
    {
      "epoch": 248.14,
      "grad_norm": 7.912144184112549,
      "learning_rate": 1.7287287287287288e-06,
      "loss": 0.205,
      "step": 82630
    },
    {
      "epoch": 248.17,
      "grad_norm": 11.038451194763184,
      "learning_rate": 1.727727727727728e-06,
      "loss": 0.2288,
      "step": 82640
    },
    {
      "epoch": 248.2,
      "grad_norm": 8.040534973144531,
      "learning_rate": 1.7267267267267268e-06,
      "loss": 0.2112,
      "step": 82650
    },
    {
      "epoch": 248.23,
      "grad_norm": 8.974067687988281,
      "learning_rate": 1.7257257257257259e-06,
      "loss": 0.1893,
      "step": 82660
    },
    {
      "epoch": 248.26,
      "grad_norm": 7.508707523345947,
      "learning_rate": 1.7247247247247249e-06,
      "loss": 0.2135,
      "step": 82670
    },
    {
      "epoch": 248.29,
      "grad_norm": 12.552742958068848,
      "learning_rate": 1.7237237237237239e-06,
      "loss": 0.2094,
      "step": 82680
    },
    {
      "epoch": 248.32,
      "grad_norm": 4.0663275718688965,
      "learning_rate": 1.7227227227227227e-06,
      "loss": 0.1943,
      "step": 82690
    },
    {
      "epoch": 248.35,
      "grad_norm": 7.627383232116699,
      "learning_rate": 1.721721721721722e-06,
      "loss": 0.1599,
      "step": 82700
    },
    {
      "epoch": 248.38,
      "grad_norm": 4.953036785125732,
      "learning_rate": 1.7207207207207207e-06,
      "loss": 0.2194,
      "step": 82710
    },
    {
      "epoch": 248.41,
      "grad_norm": 6.6075568199157715,
      "learning_rate": 1.71971971971972e-06,
      "loss": 0.1696,
      "step": 82720
    },
    {
      "epoch": 248.44,
      "grad_norm": 12.699784278869629,
      "learning_rate": 1.7187187187187188e-06,
      "loss": 0.1792,
      "step": 82730
    },
    {
      "epoch": 248.47,
      "grad_norm": 3.9411630630493164,
      "learning_rate": 1.717717717717718e-06,
      "loss": 0.1871,
      "step": 82740
    },
    {
      "epoch": 248.5,
      "grad_norm": 5.924014091491699,
      "learning_rate": 1.7167167167167168e-06,
      "loss": 0.2181,
      "step": 82750
    },
    {
      "epoch": 248.53,
      "grad_norm": 7.455838203430176,
      "learning_rate": 1.7157157157157158e-06,
      "loss": 0.2069,
      "step": 82760
    },
    {
      "epoch": 248.56,
      "grad_norm": 9.53943157196045,
      "learning_rate": 1.7147147147147146e-06,
      "loss": 0.1787,
      "step": 82770
    },
    {
      "epoch": 248.59,
      "grad_norm": 14.007837295532227,
      "learning_rate": 1.7137137137137139e-06,
      "loss": 0.1617,
      "step": 82780
    },
    {
      "epoch": 248.62,
      "grad_norm": 13.697224617004395,
      "learning_rate": 1.7127127127127127e-06,
      "loss": 0.2322,
      "step": 82790
    },
    {
      "epoch": 248.65,
      "grad_norm": 6.481261253356934,
      "learning_rate": 1.711711711711712e-06,
      "loss": 0.2086,
      "step": 82800
    },
    {
      "epoch": 248.68,
      "grad_norm": 9.432978630065918,
      "learning_rate": 1.710710710710711e-06,
      "loss": 0.1716,
      "step": 82810
    },
    {
      "epoch": 248.71,
      "grad_norm": 9.390044212341309,
      "learning_rate": 1.70970970970971e-06,
      "loss": 0.1649,
      "step": 82820
    },
    {
      "epoch": 248.74,
      "grad_norm": 20.68706703186035,
      "learning_rate": 1.708708708708709e-06,
      "loss": 0.209,
      "step": 82830
    },
    {
      "epoch": 248.77,
      "grad_norm": 14.985505104064941,
      "learning_rate": 1.7077077077077078e-06,
      "loss": 0.2051,
      "step": 82840
    },
    {
      "epoch": 248.8,
      "grad_norm": 7.688532829284668,
      "learning_rate": 1.706706706706707e-06,
      "loss": 0.1494,
      "step": 82850
    },
    {
      "epoch": 248.83,
      "grad_norm": 11.922987937927246,
      "learning_rate": 1.7057057057057058e-06,
      "loss": 0.2092,
      "step": 82860
    },
    {
      "epoch": 248.86,
      "grad_norm": 10.049397468566895,
      "learning_rate": 1.704704704704705e-06,
      "loss": 0.2214,
      "step": 82870
    },
    {
      "epoch": 248.89,
      "grad_norm": 9.013925552368164,
      "learning_rate": 1.7037037037037038e-06,
      "loss": 0.2126,
      "step": 82880
    },
    {
      "epoch": 248.92,
      "grad_norm": 7.4640421867370605,
      "learning_rate": 1.7027027027027028e-06,
      "loss": 0.2063,
      "step": 82890
    },
    {
      "epoch": 248.95,
      "grad_norm": 7.834887504577637,
      "learning_rate": 1.7017017017017019e-06,
      "loss": 0.1803,
      "step": 82900
    },
    {
      "epoch": 248.98,
      "grad_norm": 11.28691577911377,
      "learning_rate": 1.7007007007007009e-06,
      "loss": 0.1821,
      "step": 82910
    },
    {
      "epoch": 249.0,
      "eval_accuracy": 0.9212,
      "eval_loss": 0.3503081202507019,
      "eval_runtime": 30.2094,
      "eval_samples_per_second": 331.023,
      "eval_steps_per_second": 1.324,
      "step": 82917
    },
    {
      "epoch": 249.01,
      "grad_norm": 15.634957313537598,
      "learning_rate": 1.6996996996996997e-06,
      "loss": 0.1841,
      "step": 82920
    },
    {
      "epoch": 249.04,
      "grad_norm": 10.059416770935059,
      "learning_rate": 1.698698698698699e-06,
      "loss": 0.1934,
      "step": 82930
    },
    {
      "epoch": 249.07,
      "grad_norm": 6.692573070526123,
      "learning_rate": 1.6976976976976977e-06,
      "loss": 0.1669,
      "step": 82940
    },
    {
      "epoch": 249.1,
      "grad_norm": 8.994446754455566,
      "learning_rate": 1.696696696696697e-06,
      "loss": 0.1646,
      "step": 82950
    },
    {
      "epoch": 249.13,
      "grad_norm": 6.401300430297852,
      "learning_rate": 1.6956956956956958e-06,
      "loss": 0.2159,
      "step": 82960
    },
    {
      "epoch": 249.16,
      "grad_norm": 8.857094764709473,
      "learning_rate": 1.6946946946946948e-06,
      "loss": 0.1754,
      "step": 82970
    },
    {
      "epoch": 249.19,
      "grad_norm": 6.035069465637207,
      "learning_rate": 1.6936936936936938e-06,
      "loss": 0.1814,
      "step": 82980
    },
    {
      "epoch": 249.22,
      "grad_norm": 13.959473609924316,
      "learning_rate": 1.6926926926926928e-06,
      "loss": 0.1918,
      "step": 82990
    },
    {
      "epoch": 249.25,
      "grad_norm": 10.818880081176758,
      "learning_rate": 1.6916916916916916e-06,
      "loss": 0.2123,
      "step": 83000
    },
    {
      "epoch": 249.28,
      "grad_norm": 9.364628791809082,
      "learning_rate": 1.6906906906906909e-06,
      "loss": 0.1659,
      "step": 83010
    },
    {
      "epoch": 249.31,
      "grad_norm": 6.963338375091553,
      "learning_rate": 1.6896896896896897e-06,
      "loss": 0.1568,
      "step": 83020
    },
    {
      "epoch": 249.34,
      "grad_norm": 9.53274154663086,
      "learning_rate": 1.6886886886886889e-06,
      "loss": 0.2274,
      "step": 83030
    },
    {
      "epoch": 249.37,
      "grad_norm": 8.57834529876709,
      "learning_rate": 1.6876876876876877e-06,
      "loss": 0.2002,
      "step": 83040
    },
    {
      "epoch": 249.4,
      "grad_norm": 7.677379131317139,
      "learning_rate": 1.686686686686687e-06,
      "loss": 0.2217,
      "step": 83050
    },
    {
      "epoch": 249.43,
      "grad_norm": 6.710238456726074,
      "learning_rate": 1.6856856856856857e-06,
      "loss": 0.2246,
      "step": 83060
    },
    {
      "epoch": 249.46,
      "grad_norm": 7.678417205810547,
      "learning_rate": 1.6846846846846848e-06,
      "loss": 0.175,
      "step": 83070
    },
    {
      "epoch": 249.49,
      "grad_norm": 9.287692070007324,
      "learning_rate": 1.683683683683684e-06,
      "loss": 0.2006,
      "step": 83080
    },
    {
      "epoch": 249.52,
      "grad_norm": 7.984561920166016,
      "learning_rate": 1.6826826826826828e-06,
      "loss": 0.2075,
      "step": 83090
    },
    {
      "epoch": 249.55,
      "grad_norm": 8.192875862121582,
      "learning_rate": 1.681681681681682e-06,
      "loss": 0.2044,
      "step": 83100
    },
    {
      "epoch": 249.58,
      "grad_norm": 12.270228385925293,
      "learning_rate": 1.6806806806806808e-06,
      "loss": 0.1827,
      "step": 83110
    },
    {
      "epoch": 249.61,
      "grad_norm": 7.075067043304443,
      "learning_rate": 1.6796796796796798e-06,
      "loss": 0.1629,
      "step": 83120
    },
    {
      "epoch": 249.64,
      "grad_norm": 10.782404899597168,
      "learning_rate": 1.6786786786786789e-06,
      "loss": 0.1881,
      "step": 83130
    },
    {
      "epoch": 249.67,
      "grad_norm": 7.539949417114258,
      "learning_rate": 1.6776776776776779e-06,
      "loss": 0.1953,
      "step": 83140
    },
    {
      "epoch": 249.7,
      "grad_norm": 51.986572265625,
      "learning_rate": 1.6766766766766767e-06,
      "loss": 0.1928,
      "step": 83150
    },
    {
      "epoch": 249.73,
      "grad_norm": 5.117084503173828,
      "learning_rate": 1.675675675675676e-06,
      "loss": 0.2182,
      "step": 83160
    },
    {
      "epoch": 249.76,
      "grad_norm": 8.288114547729492,
      "learning_rate": 1.6746746746746747e-06,
      "loss": 0.2307,
      "step": 83170
    },
    {
      "epoch": 249.79,
      "grad_norm": 4.563942909240723,
      "learning_rate": 1.673673673673674e-06,
      "loss": 0.1714,
      "step": 83180
    },
    {
      "epoch": 249.82,
      "grad_norm": 8.681012153625488,
      "learning_rate": 1.6726726726726728e-06,
      "loss": 0.1601,
      "step": 83190
    },
    {
      "epoch": 249.85,
      "grad_norm": 8.255584716796875,
      "learning_rate": 1.6716716716716718e-06,
      "loss": 0.2304,
      "step": 83200
    },
    {
      "epoch": 249.88,
      "grad_norm": 9.307951927185059,
      "learning_rate": 1.6706706706706708e-06,
      "loss": 0.1852,
      "step": 83210
    },
    {
      "epoch": 249.91,
      "grad_norm": 5.491052150726318,
      "learning_rate": 1.6696696696696698e-06,
      "loss": 0.1809,
      "step": 83220
    },
    {
      "epoch": 249.94,
      "grad_norm": 10.044771194458008,
      "learning_rate": 1.6686686686686686e-06,
      "loss": 0.1957,
      "step": 83230
    },
    {
      "epoch": 249.97,
      "grad_norm": 10.411616325378418,
      "learning_rate": 1.6676676676676679e-06,
      "loss": 0.2064,
      "step": 83240
    },
    {
      "epoch": 250.0,
      "grad_norm": 0.00879343505948782,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 0.2,
      "step": 83250
    },
    {
      "epoch": 250.0,
      "eval_accuracy": 0.9213,
      "eval_loss": 0.351757675409317,
      "eval_runtime": 30.3439,
      "eval_samples_per_second": 329.555,
      "eval_steps_per_second": 1.318,
      "step": 83250
    },
    {
      "epoch": 250.03,
      "grad_norm": 10.794776916503906,
      "learning_rate": 1.6656656656656659e-06,
      "loss": 0.1755,
      "step": 83260
    },
    {
      "epoch": 250.06,
      "grad_norm": 10.45780086517334,
      "learning_rate": 1.6646646646646647e-06,
      "loss": 0.1873,
      "step": 83270
    },
    {
      "epoch": 250.09,
      "grad_norm": 8.762728691101074,
      "learning_rate": 1.663663663663664e-06,
      "loss": 0.198,
      "step": 83280
    },
    {
      "epoch": 250.12,
      "grad_norm": 7.063131332397461,
      "learning_rate": 1.6626626626626627e-06,
      "loss": 0.2109,
      "step": 83290
    },
    {
      "epoch": 250.15,
      "grad_norm": 6.513935565948486,
      "learning_rate": 1.6616616616616618e-06,
      "loss": 0.1768,
      "step": 83300
    },
    {
      "epoch": 250.18,
      "grad_norm": 8.813648223876953,
      "learning_rate": 1.6606606606606608e-06,
      "loss": 0.1841,
      "step": 83310
    },
    {
      "epoch": 250.21,
      "grad_norm": 7.861964702606201,
      "learning_rate": 1.6596596596596598e-06,
      "loss": 0.2045,
      "step": 83320
    },
    {
      "epoch": 250.24,
      "grad_norm": 11.378218650817871,
      "learning_rate": 1.6586586586586586e-06,
      "loss": 0.2238,
      "step": 83330
    },
    {
      "epoch": 250.27,
      "grad_norm": 8.671664237976074,
      "learning_rate": 1.6576576576576578e-06,
      "loss": 0.1924,
      "step": 83340
    },
    {
      "epoch": 250.3,
      "grad_norm": 28.415851593017578,
      "learning_rate": 1.6566566566566568e-06,
      "loss": 0.2038,
      "step": 83350
    },
    {
      "epoch": 250.33,
      "grad_norm": 11.25166130065918,
      "learning_rate": 1.6556556556556559e-06,
      "loss": 0.1655,
      "step": 83360
    },
    {
      "epoch": 250.36,
      "grad_norm": 8.424314498901367,
      "learning_rate": 1.6546546546546549e-06,
      "loss": 0.1963,
      "step": 83370
    },
    {
      "epoch": 250.39,
      "grad_norm": 9.694987297058105,
      "learning_rate": 1.6536536536536537e-06,
      "loss": 0.2005,
      "step": 83380
    },
    {
      "epoch": 250.42,
      "grad_norm": 10.84288215637207,
      "learning_rate": 1.652652652652653e-06,
      "loss": 0.1841,
      "step": 83390
    },
    {
      "epoch": 250.45,
      "grad_norm": 10.561574935913086,
      "learning_rate": 1.6516516516516517e-06,
      "loss": 0.2153,
      "step": 83400
    },
    {
      "epoch": 250.48,
      "grad_norm": 6.654036998748779,
      "learning_rate": 1.650650650650651e-06,
      "loss": 0.2011,
      "step": 83410
    },
    {
      "epoch": 250.51,
      "grad_norm": 13.62534236907959,
      "learning_rate": 1.6496496496496498e-06,
      "loss": 0.1781,
      "step": 83420
    },
    {
      "epoch": 250.54,
      "grad_norm": 6.13314962387085,
      "learning_rate": 1.6486486486486488e-06,
      "loss": 0.2207,
      "step": 83430
    },
    {
      "epoch": 250.57,
      "grad_norm": 10.895158767700195,
      "learning_rate": 1.6476476476476478e-06,
      "loss": 0.1925,
      "step": 83440
    },
    {
      "epoch": 250.6,
      "grad_norm": 9.450677871704102,
      "learning_rate": 1.6466466466466468e-06,
      "loss": 0.2035,
      "step": 83450
    },
    {
      "epoch": 250.63,
      "grad_norm": 10.317159652709961,
      "learning_rate": 1.6456456456456456e-06,
      "loss": 0.1885,
      "step": 83460
    },
    {
      "epoch": 250.66,
      "grad_norm": 7.747138500213623,
      "learning_rate": 1.6446446446446449e-06,
      "loss": 0.1988,
      "step": 83470
    },
    {
      "epoch": 250.69,
      "grad_norm": 6.049726486206055,
      "learning_rate": 1.6436436436436437e-06,
      "loss": 0.203,
      "step": 83480
    },
    {
      "epoch": 250.72,
      "grad_norm": 11.186992645263672,
      "learning_rate": 1.6426426426426429e-06,
      "loss": 0.2076,
      "step": 83490
    },
    {
      "epoch": 250.75,
      "grad_norm": 5.211151123046875,
      "learning_rate": 1.6416416416416417e-06,
      "loss": 0.1903,
      "step": 83500
    },
    {
      "epoch": 250.78,
      "grad_norm": 6.41312313079834,
      "learning_rate": 1.640640640640641e-06,
      "loss": 0.158,
      "step": 83510
    },
    {
      "epoch": 250.81,
      "grad_norm": 125.48281860351562,
      "learning_rate": 1.6396396396396397e-06,
      "loss": 0.153,
      "step": 83520
    },
    {
      "epoch": 250.84,
      "grad_norm": 4.708923816680908,
      "learning_rate": 1.6386386386386388e-06,
      "loss": 0.188,
      "step": 83530
    },
    {
      "epoch": 250.87,
      "grad_norm": 5.9217729568481445,
      "learning_rate": 1.6376376376376378e-06,
      "loss": 0.1947,
      "step": 83540
    },
    {
      "epoch": 250.9,
      "grad_norm": 7.820193290710449,
      "learning_rate": 1.6366366366366368e-06,
      "loss": 0.1996,
      "step": 83550
    },
    {
      "epoch": 250.93,
      "grad_norm": 9.93327808380127,
      "learning_rate": 1.6356356356356356e-06,
      "loss": 0.2072,
      "step": 83560
    },
    {
      "epoch": 250.96,
      "grad_norm": 7.829597473144531,
      "learning_rate": 1.6346346346346348e-06,
      "loss": 0.2176,
      "step": 83570
    },
    {
      "epoch": 250.99,
      "grad_norm": 7.138813495635986,
      "learning_rate": 1.6336336336336336e-06,
      "loss": 0.1888,
      "step": 83580
    },
    {
      "epoch": 251.0,
      "eval_accuracy": 0.9219,
      "eval_loss": 0.35089170932769775,
      "eval_runtime": 30.0869,
      "eval_samples_per_second": 332.371,
      "eval_steps_per_second": 1.329,
      "step": 83583
    },
    {
      "epoch": 251.02,
      "grad_norm": 9.018195152282715,
      "learning_rate": 1.6326326326326329e-06,
      "loss": 0.1701,
      "step": 83590
    },
    {
      "epoch": 251.05,
      "grad_norm": 20.73291778564453,
      "learning_rate": 1.6316316316316317e-06,
      "loss": 0.1622,
      "step": 83600
    },
    {
      "epoch": 251.08,
      "grad_norm": 5.560314178466797,
      "learning_rate": 1.6306306306306307e-06,
      "loss": 0.1899,
      "step": 83610
    },
    {
      "epoch": 251.11,
      "grad_norm": 9.696847915649414,
      "learning_rate": 1.62962962962963e-06,
      "loss": 0.2126,
      "step": 83620
    },
    {
      "epoch": 251.14,
      "grad_norm": 9.058853149414062,
      "learning_rate": 1.6286286286286287e-06,
      "loss": 0.1798,
      "step": 83630
    },
    {
      "epoch": 251.17,
      "grad_norm": 13.52623462677002,
      "learning_rate": 1.627627627627628e-06,
      "loss": 0.1736,
      "step": 83640
    },
    {
      "epoch": 251.2,
      "grad_norm": 10.900666236877441,
      "learning_rate": 1.6266266266266268e-06,
      "loss": 0.1906,
      "step": 83650
    },
    {
      "epoch": 251.23,
      "grad_norm": 10.697422981262207,
      "learning_rate": 1.6256256256256258e-06,
      "loss": 0.1915,
      "step": 83660
    },
    {
      "epoch": 251.26,
      "grad_norm": 9.612259864807129,
      "learning_rate": 1.6246246246246248e-06,
      "loss": 0.2065,
      "step": 83670
    },
    {
      "epoch": 251.29,
      "grad_norm": 6.621509075164795,
      "learning_rate": 1.6236236236236238e-06,
      "loss": 0.1617,
      "step": 83680
    },
    {
      "epoch": 251.32,
      "grad_norm": 12.840551376342773,
      "learning_rate": 1.6226226226226226e-06,
      "loss": 0.1552,
      "step": 83690
    },
    {
      "epoch": 251.35,
      "grad_norm": 52.70194625854492,
      "learning_rate": 1.6216216216216219e-06,
      "loss": 0.1466,
      "step": 83700
    },
    {
      "epoch": 251.38,
      "grad_norm": 13.823783874511719,
      "learning_rate": 1.6206206206206207e-06,
      "loss": 0.2075,
      "step": 83710
    },
    {
      "epoch": 251.41,
      "grad_norm": 10.1600341796875,
      "learning_rate": 1.6196196196196199e-06,
      "loss": 0.2006,
      "step": 83720
    },
    {
      "epoch": 251.44,
      "grad_norm": 6.789018154144287,
      "learning_rate": 1.6186186186186187e-06,
      "loss": 0.1797,
      "step": 83730
    },
    {
      "epoch": 251.47,
      "grad_norm": 5.59959077835083,
      "learning_rate": 1.617617617617618e-06,
      "loss": 0.1488,
      "step": 83740
    },
    {
      "epoch": 251.5,
      "grad_norm": 9.087600708007812,
      "learning_rate": 1.6166166166166167e-06,
      "loss": 0.1768,
      "step": 83750
    },
    {
      "epoch": 251.53,
      "grad_norm": 14.305049896240234,
      "learning_rate": 1.6156156156156157e-06,
      "loss": 0.1923,
      "step": 83760
    },
    {
      "epoch": 251.56,
      "grad_norm": 8.383808135986328,
      "learning_rate": 1.6146146146146146e-06,
      "loss": 0.1882,
      "step": 83770
    },
    {
      "epoch": 251.59,
      "grad_norm": 6.739475727081299,
      "learning_rate": 1.6136136136136138e-06,
      "loss": 0.1943,
      "step": 83780
    },
    {
      "epoch": 251.62,
      "grad_norm": 23.598520278930664,
      "learning_rate": 1.6126126126126126e-06,
      "loss": 0.2008,
      "step": 83790
    },
    {
      "epoch": 251.65,
      "grad_norm": 12.7157564163208,
      "learning_rate": 1.6116116116116118e-06,
      "loss": 0.1967,
      "step": 83800
    },
    {
      "epoch": 251.68,
      "grad_norm": 10.075831413269043,
      "learning_rate": 1.6106106106106106e-06,
      "loss": 0.2182,
      "step": 83810
    },
    {
      "epoch": 251.71,
      "grad_norm": 5.4692301750183105,
      "learning_rate": 1.6096096096096099e-06,
      "loss": 0.1873,
      "step": 83820
    },
    {
      "epoch": 251.74,
      "grad_norm": 11.832146644592285,
      "learning_rate": 1.6086086086086087e-06,
      "loss": 0.1653,
      "step": 83830
    },
    {
      "epoch": 251.77,
      "grad_norm": 6.496932029724121,
      "learning_rate": 1.6076076076076077e-06,
      "loss": 0.1845,
      "step": 83840
    },
    {
      "epoch": 251.8,
      "grad_norm": 10.087106704711914,
      "learning_rate": 1.6066066066066067e-06,
      "loss": 0.173,
      "step": 83850
    },
    {
      "epoch": 251.83,
      "grad_norm": 12.621432304382324,
      "learning_rate": 1.6056056056056057e-06,
      "loss": 0.1976,
      "step": 83860
    },
    {
      "epoch": 251.86,
      "grad_norm": 14.162644386291504,
      "learning_rate": 1.6046046046046045e-06,
      "loss": 0.2128,
      "step": 83870
    },
    {
      "epoch": 251.89,
      "grad_norm": 6.390702247619629,
      "learning_rate": 1.6036036036036038e-06,
      "loss": 0.1987,
      "step": 83880
    },
    {
      "epoch": 251.92,
      "grad_norm": 9.23155403137207,
      "learning_rate": 1.6026026026026028e-06,
      "loss": 0.1949,
      "step": 83890
    },
    {
      "epoch": 251.95,
      "grad_norm": 8.36522102355957,
      "learning_rate": 1.6016016016016018e-06,
      "loss": 0.1831,
      "step": 83900
    },
    {
      "epoch": 251.98,
      "grad_norm": 11.57300090789795,
      "learning_rate": 1.6006006006006008e-06,
      "loss": 0.2034,
      "step": 83910
    },
    {
      "epoch": 252.0,
      "eval_accuracy": 0.9207,
      "eval_loss": 0.3488443195819855,
      "eval_runtime": 29.777,
      "eval_samples_per_second": 335.83,
      "eval_steps_per_second": 1.343,
      "step": 83916
    },
    {
      "epoch": 252.01,
      "grad_norm": 9.973349571228027,
      "learning_rate": 1.5995995995995996e-06,
      "loss": 0.1799,
      "step": 83920
    },
    {
      "epoch": 252.04,
      "grad_norm": 14.604726791381836,
      "learning_rate": 1.5985985985985988e-06,
      "loss": 0.1763,
      "step": 83930
    },
    {
      "epoch": 252.07,
      "grad_norm": 7.50577974319458,
      "learning_rate": 1.5975975975975977e-06,
      "loss": 0.2141,
      "step": 83940
    },
    {
      "epoch": 252.1,
      "grad_norm": 8.15921401977539,
      "learning_rate": 1.5965965965965969e-06,
      "loss": 0.1894,
      "step": 83950
    },
    {
      "epoch": 252.13,
      "grad_norm": 5.557632923126221,
      "learning_rate": 1.5955955955955957e-06,
      "loss": 0.1664,
      "step": 83960
    },
    {
      "epoch": 252.16,
      "grad_norm": 7.298821449279785,
      "learning_rate": 1.5945945945945947e-06,
      "loss": 0.1875,
      "step": 83970
    },
    {
      "epoch": 252.19,
      "grad_norm": 6.991804599761963,
      "learning_rate": 1.5935935935935937e-06,
      "loss": 0.1702,
      "step": 83980
    },
    {
      "epoch": 252.22,
      "grad_norm": 9.439970016479492,
      "learning_rate": 1.5925925925925927e-06,
      "loss": 0.2392,
      "step": 83990
    },
    {
      "epoch": 252.25,
      "grad_norm": 4.867175102233887,
      "learning_rate": 1.5915915915915916e-06,
      "loss": 0.1431,
      "step": 84000
    },
    {
      "epoch": 252.28,
      "grad_norm": 9.074277877807617,
      "learning_rate": 1.5905905905905908e-06,
      "loss": 0.2042,
      "step": 84010
    },
    {
      "epoch": 252.31,
      "grad_norm": 15.082208633422852,
      "learning_rate": 1.5895895895895896e-06,
      "loss": 0.1941,
      "step": 84020
    },
    {
      "epoch": 252.34,
      "grad_norm": 5.623330116271973,
      "learning_rate": 1.5885885885885888e-06,
      "loss": 0.1901,
      "step": 84030
    },
    {
      "epoch": 252.37,
      "grad_norm": 5.122713088989258,
      "learning_rate": 1.5875875875875876e-06,
      "loss": 0.1849,
      "step": 84040
    },
    {
      "epoch": 252.4,
      "grad_norm": 7.240313529968262,
      "learning_rate": 1.5865865865865869e-06,
      "loss": 0.2071,
      "step": 84050
    },
    {
      "epoch": 252.43,
      "grad_norm": 6.952949523925781,
      "learning_rate": 1.5855855855855857e-06,
      "loss": 0.1833,
      "step": 84060
    },
    {
      "epoch": 252.46,
      "grad_norm": 10.746049880981445,
      "learning_rate": 1.5845845845845847e-06,
      "loss": 0.1816,
      "step": 84070
    },
    {
      "epoch": 252.49,
      "grad_norm": 7.284366607666016,
      "learning_rate": 1.5835835835835837e-06,
      "loss": 0.1931,
      "step": 84080
    },
    {
      "epoch": 252.52,
      "grad_norm": 9.014360427856445,
      "learning_rate": 1.5825825825825827e-06,
      "loss": 0.1639,
      "step": 84090
    },
    {
      "epoch": 252.55,
      "grad_norm": 8.819592475891113,
      "learning_rate": 1.5815815815815815e-06,
      "loss": 0.2105,
      "step": 84100
    },
    {
      "epoch": 252.58,
      "grad_norm": 9.230751991271973,
      "learning_rate": 1.5805805805805808e-06,
      "loss": 0.2132,
      "step": 84110
    },
    {
      "epoch": 252.61,
      "grad_norm": 8.656737327575684,
      "learning_rate": 1.5795795795795796e-06,
      "loss": 0.1928,
      "step": 84120
    },
    {
      "epoch": 252.64,
      "grad_norm": 10.193033218383789,
      "learning_rate": 1.5785785785785788e-06,
      "loss": 0.1793,
      "step": 84130
    },
    {
      "epoch": 252.67,
      "grad_norm": 6.419241428375244,
      "learning_rate": 1.5775775775775776e-06,
      "loss": 0.1943,
      "step": 84140
    },
    {
      "epoch": 252.7,
      "grad_norm": 10.893065452575684,
      "learning_rate": 1.5765765765765766e-06,
      "loss": 0.1911,
      "step": 84150
    },
    {
      "epoch": 252.73,
      "grad_norm": 8.113825798034668,
      "learning_rate": 1.5755755755755758e-06,
      "loss": 0.1674,
      "step": 84160
    },
    {
      "epoch": 252.76,
      "grad_norm": 6.306522846221924,
      "learning_rate": 1.5745745745745747e-06,
      "loss": 0.1551,
      "step": 84170
    },
    {
      "epoch": 252.79,
      "grad_norm": 5.48300838470459,
      "learning_rate": 1.5735735735735739e-06,
      "loss": 0.1645,
      "step": 84180
    },
    {
      "epoch": 252.82,
      "grad_norm": 9.34778881072998,
      "learning_rate": 1.5725725725725727e-06,
      "loss": 0.174,
      "step": 84190
    },
    {
      "epoch": 252.85,
      "grad_norm": 11.933847427368164,
      "learning_rate": 1.5715715715715717e-06,
      "loss": 0.2129,
      "step": 84200
    },
    {
      "epoch": 252.88,
      "grad_norm": 6.29820442199707,
      "learning_rate": 1.5705705705705707e-06,
      "loss": 0.1898,
      "step": 84210
    },
    {
      "epoch": 252.91,
      "grad_norm": 5.9054036140441895,
      "learning_rate": 1.5695695695695697e-06,
      "loss": 0.1556,
      "step": 84220
    },
    {
      "epoch": 252.94,
      "grad_norm": 10.038996696472168,
      "learning_rate": 1.5685685685685686e-06,
      "loss": 0.1754,
      "step": 84230
    },
    {
      "epoch": 252.97,
      "grad_norm": 12.041454315185547,
      "learning_rate": 1.5675675675675678e-06,
      "loss": 0.2062,
      "step": 84240
    },
    {
      "epoch": 253.0,
      "eval_accuracy": 0.9217,
      "eval_loss": 0.3463594317436218,
      "eval_runtime": 30.0961,
      "eval_samples_per_second": 332.268,
      "eval_steps_per_second": 1.329,
      "step": 84249
    },
    {
      "epoch": 253.0,
      "grad_norm": 8.440600395202637,
      "learning_rate": 1.5665665665665666e-06,
      "loss": 0.1773,
      "step": 84250
    },
    {
      "epoch": 253.03,
      "grad_norm": 5.997506141662598,
      "learning_rate": 1.5655655655655658e-06,
      "loss": 0.1842,
      "step": 84260
    },
    {
      "epoch": 253.06,
      "grad_norm": 4.635735034942627,
      "learning_rate": 1.5645645645645646e-06,
      "loss": 0.2025,
      "step": 84270
    },
    {
      "epoch": 253.09,
      "grad_norm": 9.375762939453125,
      "learning_rate": 1.5635635635635639e-06,
      "loss": 0.187,
      "step": 84280
    },
    {
      "epoch": 253.12,
      "grad_norm": 40.768558502197266,
      "learning_rate": 1.5625625625625627e-06,
      "loss": 0.21,
      "step": 84290
    },
    {
      "epoch": 253.15,
      "grad_norm": 5.650484085083008,
      "learning_rate": 1.5615615615615617e-06,
      "loss": 0.1998,
      "step": 84300
    },
    {
      "epoch": 253.18,
      "grad_norm": 13.966201782226562,
      "learning_rate": 1.5605605605605607e-06,
      "loss": 0.1756,
      "step": 84310
    },
    {
      "epoch": 253.21,
      "grad_norm": 6.89492654800415,
      "learning_rate": 1.5595595595595597e-06,
      "loss": 0.205,
      "step": 84320
    },
    {
      "epoch": 253.24,
      "grad_norm": 7.538625717163086,
      "learning_rate": 1.5585585585585585e-06,
      "loss": 0.2537,
      "step": 84330
    },
    {
      "epoch": 253.27,
      "grad_norm": 19.087453842163086,
      "learning_rate": 1.5575575575575578e-06,
      "loss": 0.1821,
      "step": 84340
    },
    {
      "epoch": 253.3,
      "grad_norm": 15.082911491394043,
      "learning_rate": 1.5565565565565566e-06,
      "loss": 0.2179,
      "step": 84350
    },
    {
      "epoch": 253.33,
      "grad_norm": 10.164932250976562,
      "learning_rate": 1.5555555555555558e-06,
      "loss": 0.2199,
      "step": 84360
    },
    {
      "epoch": 253.36,
      "grad_norm": 10.499350547790527,
      "learning_rate": 1.5545545545545546e-06,
      "loss": 0.1848,
      "step": 84370
    },
    {
      "epoch": 253.39,
      "grad_norm": 12.79383659362793,
      "learning_rate": 1.5535535535535536e-06,
      "loss": 0.1937,
      "step": 84380
    },
    {
      "epoch": 253.42,
      "grad_norm": 8.8218412399292,
      "learning_rate": 1.5525525525525526e-06,
      "loss": 0.2196,
      "step": 84390
    },
    {
      "epoch": 253.45,
      "grad_norm": 25.5424861907959,
      "learning_rate": 1.5515515515515517e-06,
      "loss": 0.2187,
      "step": 84400
    },
    {
      "epoch": 253.48,
      "grad_norm": 6.663982391357422,
      "learning_rate": 1.5505505505505509e-06,
      "loss": 0.1718,
      "step": 84410
    },
    {
      "epoch": 253.51,
      "grad_norm": 7.582066059112549,
      "learning_rate": 1.5495495495495497e-06,
      "loss": 0.1986,
      "step": 84420
    },
    {
      "epoch": 253.54,
      "grad_norm": 9.242899894714355,
      "learning_rate": 1.5485485485485487e-06,
      "loss": 0.2676,
      "step": 84430
    },
    {
      "epoch": 253.57,
      "grad_norm": 11.762062072753906,
      "learning_rate": 1.5475475475475477e-06,
      "loss": 0.2555,
      "step": 84440
    },
    {
      "epoch": 253.6,
      "grad_norm": 19.258106231689453,
      "learning_rate": 1.5465465465465467e-06,
      "loss": 0.2025,
      "step": 84450
    },
    {
      "epoch": 253.63,
      "grad_norm": 10.920361518859863,
      "learning_rate": 1.5455455455455456e-06,
      "loss": 0.1706,
      "step": 84460
    },
    {
      "epoch": 253.66,
      "grad_norm": 10.302874565124512,
      "learning_rate": 1.5445445445445448e-06,
      "loss": 0.2056,
      "step": 84470
    },
    {
      "epoch": 253.69,
      "grad_norm": 8.208734512329102,
      "learning_rate": 1.5435435435435436e-06,
      "loss": 0.1863,
      "step": 84480
    },
    {
      "epoch": 253.72,
      "grad_norm": 6.946019649505615,
      "learning_rate": 1.5425425425425428e-06,
      "loss": 0.1793,
      "step": 84490
    },
    {
      "epoch": 253.75,
      "grad_norm": 5.847348690032959,
      "learning_rate": 1.5415415415415416e-06,
      "loss": 0.1962,
      "step": 84500
    },
    {
      "epoch": 253.78,
      "grad_norm": 6.867198944091797,
      "learning_rate": 1.5405405405405409e-06,
      "loss": 0.1979,
      "step": 84510
    },
    {
      "epoch": 253.81,
      "grad_norm": 6.343681812286377,
      "learning_rate": 1.5395395395395397e-06,
      "loss": 0.159,
      "step": 84520
    },
    {
      "epoch": 253.84,
      "grad_norm": 3.756673812866211,
      "learning_rate": 1.5385385385385387e-06,
      "loss": 0.2166,
      "step": 84530
    },
    {
      "epoch": 253.87,
      "grad_norm": 9.11819839477539,
      "learning_rate": 1.5375375375375377e-06,
      "loss": 0.1757,
      "step": 84540
    },
    {
      "epoch": 253.9,
      "grad_norm": 14.766769409179688,
      "learning_rate": 1.5365365365365367e-06,
      "loss": 0.2174,
      "step": 84550
    },
    {
      "epoch": 253.93,
      "grad_norm": 13.795622825622559,
      "learning_rate": 1.5355355355355355e-06,
      "loss": 0.1565,
      "step": 84560
    },
    {
      "epoch": 253.96,
      "grad_norm": 7.059438228607178,
      "learning_rate": 1.5345345345345348e-06,
      "loss": 0.1781,
      "step": 84570
    },
    {
      "epoch": 253.99,
      "grad_norm": 9.605006217956543,
      "learning_rate": 1.5335335335335336e-06,
      "loss": 0.1906,
      "step": 84580
    },
    {
      "epoch": 254.0,
      "eval_accuracy": 0.9224,
      "eval_loss": 0.3479554057121277,
      "eval_runtime": 30.2379,
      "eval_samples_per_second": 330.711,
      "eval_steps_per_second": 1.323,
      "step": 84582
    },
    {
      "epoch": 254.02,
      "grad_norm": 6.219067573547363,
      "learning_rate": 1.5325325325325328e-06,
      "loss": 0.2006,
      "step": 84590
    },
    {
      "epoch": 254.05,
      "grad_norm": 11.411277770996094,
      "learning_rate": 1.5315315315315316e-06,
      "loss": 0.2029,
      "step": 84600
    },
    {
      "epoch": 254.08,
      "grad_norm": 9.270541191101074,
      "learning_rate": 1.5305305305305306e-06,
      "loss": 0.1572,
      "step": 84610
    },
    {
      "epoch": 254.11,
      "grad_norm": 9.961223602294922,
      "learning_rate": 1.5295295295295296e-06,
      "loss": 0.1726,
      "step": 84620
    },
    {
      "epoch": 254.14,
      "grad_norm": 12.920795440673828,
      "learning_rate": 1.5285285285285287e-06,
      "loss": 0.2023,
      "step": 84630
    },
    {
      "epoch": 254.17,
      "grad_norm": 12.385236740112305,
      "learning_rate": 1.5275275275275275e-06,
      "loss": 0.1213,
      "step": 84640
    },
    {
      "epoch": 254.2,
      "grad_norm": 9.512368202209473,
      "learning_rate": 1.5265265265265267e-06,
      "loss": 0.1947,
      "step": 84650
    },
    {
      "epoch": 254.23,
      "grad_norm": 3.4474141597747803,
      "learning_rate": 1.5255255255255255e-06,
      "loss": 0.1976,
      "step": 84660
    },
    {
      "epoch": 254.26,
      "grad_norm": 12.85355281829834,
      "learning_rate": 1.5245245245245247e-06,
      "loss": 0.2065,
      "step": 84670
    },
    {
      "epoch": 254.29,
      "grad_norm": 15.37999439239502,
      "learning_rate": 1.5235235235235237e-06,
      "loss": 0.1732,
      "step": 84680
    },
    {
      "epoch": 254.32,
      "grad_norm": 11.213558197021484,
      "learning_rate": 1.5225225225225225e-06,
      "loss": 0.2014,
      "step": 84690
    },
    {
      "epoch": 254.35,
      "grad_norm": 36.02573013305664,
      "learning_rate": 1.5215215215215218e-06,
      "loss": 0.2223,
      "step": 84700
    },
    {
      "epoch": 254.38,
      "grad_norm": 7.8706440925598145,
      "learning_rate": 1.5205205205205206e-06,
      "loss": 0.1984,
      "step": 84710
    },
    {
      "epoch": 254.41,
      "grad_norm": 8.846600532531738,
      "learning_rate": 1.5195195195195198e-06,
      "loss": 0.1869,
      "step": 84720
    },
    {
      "epoch": 254.44,
      "grad_norm": 7.854784965515137,
      "learning_rate": 1.5185185185185186e-06,
      "loss": 0.1659,
      "step": 84730
    },
    {
      "epoch": 254.47,
      "grad_norm": 12.509539604187012,
      "learning_rate": 1.5175175175175176e-06,
      "loss": 0.144,
      "step": 84740
    },
    {
      "epoch": 254.5,
      "grad_norm": 10.894037246704102,
      "learning_rate": 1.5165165165165167e-06,
      "loss": 0.1824,
      "step": 84750
    },
    {
      "epoch": 254.53,
      "grad_norm": 12.091239929199219,
      "learning_rate": 1.5155155155155157e-06,
      "loss": 0.2182,
      "step": 84760
    },
    {
      "epoch": 254.56,
      "grad_norm": 6.806854248046875,
      "learning_rate": 1.5145145145145145e-06,
      "loss": 0.1781,
      "step": 84770
    },
    {
      "epoch": 254.59,
      "grad_norm": 6.2557597160339355,
      "learning_rate": 1.5135135135135137e-06,
      "loss": 0.1958,
      "step": 84780
    },
    {
      "epoch": 254.62,
      "grad_norm": 8.194812774658203,
      "learning_rate": 1.5125125125125125e-06,
      "loss": 0.1613,
      "step": 84790
    },
    {
      "epoch": 254.65,
      "grad_norm": 8.89258861541748,
      "learning_rate": 1.5115115115115118e-06,
      "loss": 0.1847,
      "step": 84800
    },
    {
      "epoch": 254.68,
      "grad_norm": 12.996636390686035,
      "learning_rate": 1.5105105105105106e-06,
      "loss": 0.1788,
      "step": 84810
    },
    {
      "epoch": 254.71,
      "grad_norm": 9.71920108795166,
      "learning_rate": 1.5095095095095098e-06,
      "loss": 0.2025,
      "step": 84820
    },
    {
      "epoch": 254.74,
      "grad_norm": 12.199868202209473,
      "learning_rate": 1.5085085085085086e-06,
      "loss": 0.2157,
      "step": 84830
    },
    {
      "epoch": 254.77,
      "grad_norm": 11.93258285522461,
      "learning_rate": 1.5075075075075076e-06,
      "loss": 0.1773,
      "step": 84840
    },
    {
      "epoch": 254.8,
      "grad_norm": 10.124260902404785,
      "learning_rate": 1.5065065065065066e-06,
      "loss": 0.2179,
      "step": 84850
    },
    {
      "epoch": 254.83,
      "grad_norm": 5.199721336364746,
      "learning_rate": 1.5055055055055056e-06,
      "loss": 0.1859,
      "step": 84860
    },
    {
      "epoch": 254.86,
      "grad_norm": 10.023232460021973,
      "learning_rate": 1.5045045045045045e-06,
      "loss": 0.2162,
      "step": 84870
    },
    {
      "epoch": 254.89,
      "grad_norm": 7.644779682159424,
      "learning_rate": 1.5035035035035037e-06,
      "loss": 0.1891,
      "step": 84880
    },
    {
      "epoch": 254.92,
      "grad_norm": 5.8513264656066895,
      "learning_rate": 1.5025025025025025e-06,
      "loss": 0.1825,
      "step": 84890
    },
    {
      "epoch": 254.95,
      "grad_norm": 6.81651496887207,
      "learning_rate": 1.5015015015015017e-06,
      "loss": 0.1704,
      "step": 84900
    },
    {
      "epoch": 254.98,
      "grad_norm": 22.217727661132812,
      "learning_rate": 1.5005005005005005e-06,
      "loss": 0.1996,
      "step": 84910
    },
    {
      "epoch": 255.0,
      "eval_accuracy": 0.9219,
      "eval_loss": 0.34810134768486023,
      "eval_runtime": 29.8983,
      "eval_samples_per_second": 334.467,
      "eval_steps_per_second": 1.338,
      "step": 84915
    },
    {
      "epoch": 255.02,
      "grad_norm": 7.188487529754639,
      "learning_rate": 1.4994994994994995e-06,
      "loss": 0.2521,
      "step": 84920
    },
    {
      "epoch": 255.05,
      "grad_norm": 11.705162048339844,
      "learning_rate": 1.4984984984984986e-06,
      "loss": 0.2042,
      "step": 84930
    },
    {
      "epoch": 255.08,
      "grad_norm": 4.870003700256348,
      "learning_rate": 1.4974974974974976e-06,
      "loss": 0.1625,
      "step": 84940
    },
    {
      "epoch": 255.11,
      "grad_norm": 16.507972717285156,
      "learning_rate": 1.4964964964964968e-06,
      "loss": 0.225,
      "step": 84950
    },
    {
      "epoch": 255.14,
      "grad_norm": 6.937869071960449,
      "learning_rate": 1.4954954954954956e-06,
      "loss": 0.2294,
      "step": 84960
    },
    {
      "epoch": 255.17,
      "grad_norm": 28.14149284362793,
      "learning_rate": 1.4944944944944946e-06,
      "loss": 0.1826,
      "step": 84970
    },
    {
      "epoch": 255.2,
      "grad_norm": 4.894869327545166,
      "learning_rate": 1.4934934934934937e-06,
      "loss": 0.1542,
      "step": 84980
    },
    {
      "epoch": 255.23,
      "grad_norm": 14.234757423400879,
      "learning_rate": 1.4924924924924927e-06,
      "loss": 0.2066,
      "step": 84990
    },
    {
      "epoch": 255.26,
      "grad_norm": 11.596181869506836,
      "learning_rate": 1.4914914914914915e-06,
      "loss": 0.1768,
      "step": 85000
    },
    {
      "epoch": 255.29,
      "grad_norm": 6.5646748542785645,
      "learning_rate": 1.4904904904904907e-06,
      "loss": 0.1889,
      "step": 85010
    },
    {
      "epoch": 255.32,
      "grad_norm": 14.641605377197266,
      "learning_rate": 1.4894894894894895e-06,
      "loss": 0.2023,
      "step": 85020
    },
    {
      "epoch": 255.35,
      "grad_norm": 7.9419264793396,
      "learning_rate": 1.4884884884884887e-06,
      "loss": 0.1853,
      "step": 85030
    },
    {
      "epoch": 255.38,
      "grad_norm": 13.468498229980469,
      "learning_rate": 1.4874874874874876e-06,
      "loss": 0.1993,
      "step": 85040
    },
    {
      "epoch": 255.41,
      "grad_norm": 6.743235111236572,
      "learning_rate": 1.4864864864864868e-06,
      "loss": 0.179,
      "step": 85050
    },
    {
      "epoch": 255.44,
      "grad_norm": 15.95434856414795,
      "learning_rate": 1.4854854854854856e-06,
      "loss": 0.1942,
      "step": 85060
    },
    {
      "epoch": 255.47,
      "grad_norm": 6.251765727996826,
      "learning_rate": 1.4844844844844846e-06,
      "loss": 0.2159,
      "step": 85070
    },
    {
      "epoch": 255.5,
      "grad_norm": 9.780799865722656,
      "learning_rate": 1.4834834834834836e-06,
      "loss": 0.1874,
      "step": 85080
    },
    {
      "epoch": 255.53,
      "grad_norm": 6.645637512207031,
      "learning_rate": 1.4824824824824826e-06,
      "loss": 0.1927,
      "step": 85090
    },
    {
      "epoch": 255.56,
      "grad_norm": 23.570375442504883,
      "learning_rate": 1.4814814814814815e-06,
      "loss": 0.1602,
      "step": 85100
    },
    {
      "epoch": 255.59,
      "grad_norm": 9.964300155639648,
      "learning_rate": 1.4804804804804807e-06,
      "loss": 0.2117,
      "step": 85110
    },
    {
      "epoch": 255.62,
      "grad_norm": 8.119893074035645,
      "learning_rate": 1.4794794794794795e-06,
      "loss": 0.201,
      "step": 85120
    },
    {
      "epoch": 255.65,
      "grad_norm": 14.222092628479004,
      "learning_rate": 1.4784784784784787e-06,
      "loss": 0.188,
      "step": 85130
    },
    {
      "epoch": 255.68,
      "grad_norm": 8.058115005493164,
      "learning_rate": 1.4774774774774775e-06,
      "loss": 0.1742,
      "step": 85140
    },
    {
      "epoch": 255.71,
      "grad_norm": 9.349413871765137,
      "learning_rate": 1.4764764764764765e-06,
      "loss": 0.2116,
      "step": 85150
    },
    {
      "epoch": 255.74,
      "grad_norm": 7.031284332275391,
      "learning_rate": 1.4754754754754756e-06,
      "loss": 0.1742,
      "step": 85160
    },
    {
      "epoch": 255.77,
      "grad_norm": 6.378910064697266,
      "learning_rate": 1.4744744744744746e-06,
      "loss": 0.1996,
      "step": 85170
    },
    {
      "epoch": 255.8,
      "grad_norm": 16.11748504638672,
      "learning_rate": 1.4734734734734734e-06,
      "loss": 0.2084,
      "step": 85180
    },
    {
      "epoch": 255.83,
      "grad_norm": 6.35882568359375,
      "learning_rate": 1.4724724724724726e-06,
      "loss": 0.1709,
      "step": 85190
    },
    {
      "epoch": 255.86,
      "grad_norm": 19.991010665893555,
      "learning_rate": 1.4714714714714714e-06,
      "loss": 0.227,
      "step": 85200
    },
    {
      "epoch": 255.89,
      "grad_norm": 6.595259666442871,
      "learning_rate": 1.4704704704704707e-06,
      "loss": 0.2189,
      "step": 85210
    },
    {
      "epoch": 255.92,
      "grad_norm": 15.530816078186035,
      "learning_rate": 1.4694694694694697e-06,
      "loss": 0.1904,
      "step": 85220
    },
    {
      "epoch": 255.95,
      "grad_norm": 6.775717258453369,
      "learning_rate": 1.4684684684684685e-06,
      "loss": 0.1834,
      "step": 85230
    },
    {
      "epoch": 255.98,
      "grad_norm": 8.963126182556152,
      "learning_rate": 1.4674674674674677e-06,
      "loss": 0.2447,
      "step": 85240
    },
    {
      "epoch": 256.0,
      "eval_accuracy": 0.9217,
      "eval_loss": 0.34845760464668274,
      "eval_runtime": 30.0796,
      "eval_samples_per_second": 332.451,
      "eval_steps_per_second": 1.33,
      "step": 85248
    },
    {
      "epoch": 256.01,
      "grad_norm": 7.669626235961914,
      "learning_rate": 1.4664664664664665e-06,
      "loss": 0.2023,
      "step": 85250
    },
    {
      "epoch": 256.04,
      "grad_norm": 6.88312292098999,
      "learning_rate": 1.4654654654654657e-06,
      "loss": 0.1683,
      "step": 85260
    },
    {
      "epoch": 256.07,
      "grad_norm": 15.465850830078125,
      "learning_rate": 1.4644644644644646e-06,
      "loss": 0.1946,
      "step": 85270
    },
    {
      "epoch": 256.1,
      "grad_norm": 7.2835373878479,
      "learning_rate": 1.4634634634634638e-06,
      "loss": 0.192,
      "step": 85280
    },
    {
      "epoch": 256.13,
      "grad_norm": 7.033202171325684,
      "learning_rate": 1.4624624624624626e-06,
      "loss": 0.1975,
      "step": 85290
    },
    {
      "epoch": 256.16,
      "grad_norm": 6.180278301239014,
      "learning_rate": 1.4614614614614616e-06,
      "loss": 0.1826,
      "step": 85300
    },
    {
      "epoch": 256.19,
      "grad_norm": 14.589160919189453,
      "learning_rate": 1.4604604604604606e-06,
      "loss": 0.1977,
      "step": 85310
    },
    {
      "epoch": 256.22,
      "grad_norm": 14.76817798614502,
      "learning_rate": 1.4594594594594596e-06,
      "loss": 0.2159,
      "step": 85320
    },
    {
      "epoch": 256.25,
      "grad_norm": 13.509636878967285,
      "learning_rate": 1.4584584584584585e-06,
      "loss": 0.1933,
      "step": 85330
    },
    {
      "epoch": 256.28,
      "grad_norm": 11.038846969604492,
      "learning_rate": 1.4574574574574577e-06,
      "loss": 0.1806,
      "step": 85340
    },
    {
      "epoch": 256.31,
      "grad_norm": 9.421637535095215,
      "learning_rate": 1.4564564564564565e-06,
      "loss": 0.2019,
      "step": 85350
    },
    {
      "epoch": 256.34,
      "grad_norm": 9.085336685180664,
      "learning_rate": 1.4554554554554557e-06,
      "loss": 0.1925,
      "step": 85360
    },
    {
      "epoch": 256.37,
      "grad_norm": 7.961391448974609,
      "learning_rate": 1.4544544544544545e-06,
      "loss": 0.1712,
      "step": 85370
    },
    {
      "epoch": 256.4,
      "grad_norm": 8.49344253540039,
      "learning_rate": 1.4534534534534535e-06,
      "loss": 0.177,
      "step": 85380
    },
    {
      "epoch": 256.43,
      "grad_norm": 5.981365203857422,
      "learning_rate": 1.4524524524524526e-06,
      "loss": 0.1894,
      "step": 85390
    },
    {
      "epoch": 256.46,
      "grad_norm": 5.877206802368164,
      "learning_rate": 1.4514514514514516e-06,
      "loss": 0.1841,
      "step": 85400
    },
    {
      "epoch": 256.49,
      "grad_norm": 6.2272772789001465,
      "learning_rate": 1.4504504504504504e-06,
      "loss": 0.2011,
      "step": 85410
    },
    {
      "epoch": 256.52,
      "grad_norm": 7.824004173278809,
      "learning_rate": 1.4494494494494496e-06,
      "loss": 0.1916,
      "step": 85420
    },
    {
      "epoch": 256.55,
      "grad_norm": 7.4414167404174805,
      "learning_rate": 1.4484484484484484e-06,
      "loss": 0.2172,
      "step": 85430
    },
    {
      "epoch": 256.58,
      "grad_norm": 11.22990608215332,
      "learning_rate": 1.4474474474474477e-06,
      "loss": 0.2271,
      "step": 85440
    },
    {
      "epoch": 256.61,
      "grad_norm": 6.846374988555908,
      "learning_rate": 1.4464464464464465e-06,
      "loss": 0.1685,
      "step": 85450
    },
    {
      "epoch": 256.64,
      "grad_norm": 18.15487289428711,
      "learning_rate": 1.4454454454454455e-06,
      "loss": 0.1871,
      "step": 85460
    },
    {
      "epoch": 256.67,
      "grad_norm": 9.422703742980957,
      "learning_rate": 1.4444444444444445e-06,
      "loss": 0.2157,
      "step": 85470
    },
    {
      "epoch": 256.7,
      "grad_norm": 6.05608606338501,
      "learning_rate": 1.4434434434434435e-06,
      "loss": 0.2094,
      "step": 85480
    },
    {
      "epoch": 256.73,
      "grad_norm": 9.45822811126709,
      "learning_rate": 1.4424424424424427e-06,
      "loss": 0.2198,
      "step": 85490
    },
    {
      "epoch": 256.76,
      "grad_norm": 9.619877815246582,
      "learning_rate": 1.4414414414414416e-06,
      "loss": 0.2055,
      "step": 85500
    },
    {
      "epoch": 256.79,
      "grad_norm": 13.199268341064453,
      "learning_rate": 1.4404404404404408e-06,
      "loss": 0.2056,
      "step": 85510
    },
    {
      "epoch": 256.82,
      "grad_norm": 6.681459903717041,
      "learning_rate": 1.4394394394394396e-06,
      "loss": 0.1878,
      "step": 85520
    },
    {
      "epoch": 256.85,
      "grad_norm": 7.777061939239502,
      "learning_rate": 1.4384384384384386e-06,
      "loss": 0.1862,
      "step": 85530
    },
    {
      "epoch": 256.88,
      "grad_norm": 11.829968452453613,
      "learning_rate": 1.4374374374374376e-06,
      "loss": 0.1929,
      "step": 85540
    },
    {
      "epoch": 256.91,
      "grad_norm": 10.027358055114746,
      "learning_rate": 1.4364364364364366e-06,
      "loss": 0.2309,
      "step": 85550
    },
    {
      "epoch": 256.94,
      "grad_norm": 8.998444557189941,
      "learning_rate": 1.4354354354354354e-06,
      "loss": 0.1634,
      "step": 85560
    },
    {
      "epoch": 256.97,
      "grad_norm": 21.71193504333496,
      "learning_rate": 1.4344344344344347e-06,
      "loss": 0.2024,
      "step": 85570
    },
    {
      "epoch": 257.0,
      "grad_norm": 10.483360290527344,
      "learning_rate": 1.4334334334334335e-06,
      "loss": 0.1975,
      "step": 85580
    },
    {
      "epoch": 257.0,
      "eval_accuracy": 0.9215,
      "eval_loss": 0.35088658332824707,
      "eval_runtime": 29.842,
      "eval_samples_per_second": 335.099,
      "eval_steps_per_second": 1.34,
      "step": 85581
    },
    {
      "epoch": 257.03,
      "grad_norm": 12.70344066619873,
      "learning_rate": 1.4324324324324327e-06,
      "loss": 0.1486,
      "step": 85590
    },
    {
      "epoch": 257.06,
      "grad_norm": 16.770376205444336,
      "learning_rate": 1.4314314314314315e-06,
      "loss": 0.1681,
      "step": 85600
    },
    {
      "epoch": 257.09,
      "grad_norm": 6.173611164093018,
      "learning_rate": 1.4304304304304305e-06,
      "loss": 0.1891,
      "step": 85610
    },
    {
      "epoch": 257.12,
      "grad_norm": 28.7573299407959,
      "learning_rate": 1.4294294294294296e-06,
      "loss": 0.2037,
      "step": 85620
    },
    {
      "epoch": 257.15,
      "grad_norm": 10.991841316223145,
      "learning_rate": 1.4284284284284286e-06,
      "loss": 0.229,
      "step": 85630
    },
    {
      "epoch": 257.18,
      "grad_norm": 14.98281192779541,
      "learning_rate": 1.4274274274274274e-06,
      "loss": 0.1713,
      "step": 85640
    },
    {
      "epoch": 257.21,
      "grad_norm": 4.453299522399902,
      "learning_rate": 1.4264264264264266e-06,
      "loss": 0.1749,
      "step": 85650
    },
    {
      "epoch": 257.24,
      "grad_norm": 4.231456756591797,
      "learning_rate": 1.4254254254254254e-06,
      "loss": 0.1841,
      "step": 85660
    },
    {
      "epoch": 257.27,
      "grad_norm": 7.980093955993652,
      "learning_rate": 1.4244244244244247e-06,
      "loss": 0.176,
      "step": 85670
    },
    {
      "epoch": 257.3,
      "grad_norm": 5.927448272705078,
      "learning_rate": 1.4234234234234235e-06,
      "loss": 0.2244,
      "step": 85680
    },
    {
      "epoch": 257.33,
      "grad_norm": 8.416011810302734,
      "learning_rate": 1.4224224224224225e-06,
      "loss": 0.1906,
      "step": 85690
    },
    {
      "epoch": 257.36,
      "grad_norm": 7.969036102294922,
      "learning_rate": 1.4214214214214215e-06,
      "loss": 0.2016,
      "step": 85700
    },
    {
      "epoch": 257.39,
      "grad_norm": 7.667527198791504,
      "learning_rate": 1.4204204204204205e-06,
      "loss": 0.2108,
      "step": 85710
    },
    {
      "epoch": 257.42,
      "grad_norm": 15.430827140808105,
      "learning_rate": 1.4194194194194193e-06,
      "loss": 0.2067,
      "step": 85720
    },
    {
      "epoch": 257.45,
      "grad_norm": 12.713275909423828,
      "learning_rate": 1.4184184184184185e-06,
      "loss": 0.1904,
      "step": 85730
    },
    {
      "epoch": 257.48,
      "grad_norm": 5.295680522918701,
      "learning_rate": 1.4174174174174174e-06,
      "loss": 0.1496,
      "step": 85740
    },
    {
      "epoch": 257.51,
      "grad_norm": 4.210613250732422,
      "learning_rate": 1.4164164164164166e-06,
      "loss": 0.1799,
      "step": 85750
    },
    {
      "epoch": 257.54,
      "grad_norm": 6.023449420928955,
      "learning_rate": 1.4154154154154156e-06,
      "loss": 0.1991,
      "step": 85760
    },
    {
      "epoch": 257.57,
      "grad_norm": 10.513994216918945,
      "learning_rate": 1.4144144144144144e-06,
      "loss": 0.1912,
      "step": 85770
    },
    {
      "epoch": 257.6,
      "grad_norm": 9.003552436828613,
      "learning_rate": 1.4134134134134136e-06,
      "loss": 0.1948,
      "step": 85780
    },
    {
      "epoch": 257.63,
      "grad_norm": 9.759734153747559,
      "learning_rate": 1.4124124124124124e-06,
      "loss": 0.1942,
      "step": 85790
    },
    {
      "epoch": 257.66,
      "grad_norm": 5.5993828773498535,
      "learning_rate": 1.4114114114114117e-06,
      "loss": 0.1811,
      "step": 85800
    },
    {
      "epoch": 257.69,
      "grad_norm": 5.843337535858154,
      "learning_rate": 1.4104104104104105e-06,
      "loss": 0.2051,
      "step": 85810
    },
    {
      "epoch": 257.72,
      "grad_norm": 10.354949951171875,
      "learning_rate": 1.4094094094094097e-06,
      "loss": 0.2297,
      "step": 85820
    },
    {
      "epoch": 257.75,
      "grad_norm": 7.541104316711426,
      "learning_rate": 1.4084084084084085e-06,
      "loss": 0.1765,
      "step": 85830
    },
    {
      "epoch": 257.78,
      "grad_norm": 14.106831550598145,
      "learning_rate": 1.4074074074074075e-06,
      "loss": 0.185,
      "step": 85840
    },
    {
      "epoch": 257.81,
      "grad_norm": 8.711864471435547,
      "learning_rate": 1.4064064064064066e-06,
      "loss": 0.1554,
      "step": 85850
    },
    {
      "epoch": 257.84,
      "grad_norm": 12.349267959594727,
      "learning_rate": 1.4054054054054056e-06,
      "loss": 0.2152,
      "step": 85860
    },
    {
      "epoch": 257.87,
      "grad_norm": 11.269804954528809,
      "learning_rate": 1.4044044044044044e-06,
      "loss": 0.1717,
      "step": 85870
    },
    {
      "epoch": 257.9,
      "grad_norm": 7.189770221710205,
      "learning_rate": 1.4034034034034036e-06,
      "loss": 0.2049,
      "step": 85880
    },
    {
      "epoch": 257.93,
      "grad_norm": 7.7067999839782715,
      "learning_rate": 1.4024024024024024e-06,
      "loss": 0.178,
      "step": 85890
    },
    {
      "epoch": 257.96,
      "grad_norm": 4.738809108734131,
      "learning_rate": 1.4014014014014016e-06,
      "loss": 0.1658,
      "step": 85900
    },
    {
      "epoch": 257.99,
      "grad_norm": 44.575199127197266,
      "learning_rate": 1.4004004004004005e-06,
      "loss": 0.1787,
      "step": 85910
    },
    {
      "epoch": 258.0,
      "eval_accuracy": 0.9203,
      "eval_loss": 0.34971100091934204,
      "eval_runtime": 30.3732,
      "eval_samples_per_second": 329.237,
      "eval_steps_per_second": 1.317,
      "step": 85914
    },
    {
      "epoch": 258.02,
      "grad_norm": 7.202541828155518,
      "learning_rate": 1.3993993993993995e-06,
      "loss": 0.1276,
      "step": 85920
    },
    {
      "epoch": 258.05,
      "grad_norm": 4.3788299560546875,
      "learning_rate": 1.3983983983983985e-06,
      "loss": 0.2187,
      "step": 85930
    },
    {
      "epoch": 258.08,
      "grad_norm": 10.237161636352539,
      "learning_rate": 1.3973973973973975e-06,
      "loss": 0.2032,
      "step": 85940
    },
    {
      "epoch": 258.11,
      "grad_norm": 4.894404411315918,
      "learning_rate": 1.3963963963963963e-06,
      "loss": 0.1656,
      "step": 85950
    },
    {
      "epoch": 258.14,
      "grad_norm": 15.913201332092285,
      "learning_rate": 1.3953953953953955e-06,
      "loss": 0.2111,
      "step": 85960
    },
    {
      "epoch": 258.17,
      "grad_norm": 6.558067798614502,
      "learning_rate": 1.3943943943943944e-06,
      "loss": 0.2318,
      "step": 85970
    },
    {
      "epoch": 258.2,
      "grad_norm": 11.790576934814453,
      "learning_rate": 1.3933933933933936e-06,
      "loss": 0.2087,
      "step": 85980
    },
    {
      "epoch": 258.23,
      "grad_norm": 6.555428504943848,
      "learning_rate": 1.3923923923923924e-06,
      "loss": 0.169,
      "step": 85990
    },
    {
      "epoch": 258.26,
      "grad_norm": 10.55886459350586,
      "learning_rate": 1.3913913913913914e-06,
      "loss": 0.243,
      "step": 86000
    },
    {
      "epoch": 258.29,
      "grad_norm": 14.066753387451172,
      "learning_rate": 1.3903903903903904e-06,
      "loss": 0.2246,
      "step": 86010
    },
    {
      "epoch": 258.32,
      "grad_norm": 7.860253810882568,
      "learning_rate": 1.3893893893893894e-06,
      "loss": 0.1964,
      "step": 86020
    },
    {
      "epoch": 258.35,
      "grad_norm": 5.095664978027344,
      "learning_rate": 1.3883883883883887e-06,
      "loss": 0.1975,
      "step": 86030
    },
    {
      "epoch": 258.38,
      "grad_norm": 8.543028831481934,
      "learning_rate": 1.3873873873873875e-06,
      "loss": 0.1833,
      "step": 86040
    },
    {
      "epoch": 258.41,
      "grad_norm": 6.84279727935791,
      "learning_rate": 1.3863863863863867e-06,
      "loss": 0.2257,
      "step": 86050
    },
    {
      "epoch": 258.44,
      "grad_norm": 12.435588836669922,
      "learning_rate": 1.3853853853853855e-06,
      "loss": 0.1827,
      "step": 86060
    },
    {
      "epoch": 258.47,
      "grad_norm": 8.298110008239746,
      "learning_rate": 1.3843843843843845e-06,
      "loss": 0.1823,
      "step": 86070
    },
    {
      "epoch": 258.5,
      "grad_norm": 8.762423515319824,
      "learning_rate": 1.3833833833833836e-06,
      "loss": 0.1816,
      "step": 86080
    },
    {
      "epoch": 258.53,
      "grad_norm": 8.900404930114746,
      "learning_rate": 1.3823823823823826e-06,
      "loss": 0.1783,
      "step": 86090
    },
    {
      "epoch": 258.56,
      "grad_norm": 17.4864444732666,
      "learning_rate": 1.3813813813813814e-06,
      "loss": 0.2063,
      "step": 86100
    },
    {
      "epoch": 258.59,
      "grad_norm": 9.39680290222168,
      "learning_rate": 1.3803803803803806e-06,
      "loss": 0.1717,
      "step": 86110
    },
    {
      "epoch": 258.62,
      "grad_norm": 6.834825038909912,
      "learning_rate": 1.3793793793793794e-06,
      "loss": 0.1774,
      "step": 86120
    },
    {
      "epoch": 258.65,
      "grad_norm": 7.838572025299072,
      "learning_rate": 1.3783783783783786e-06,
      "loss": 0.1921,
      "step": 86130
    },
    {
      "epoch": 258.68,
      "grad_norm": 7.798161506652832,
      "learning_rate": 1.3773773773773775e-06,
      "loss": 0.2327,
      "step": 86140
    },
    {
      "epoch": 258.71,
      "grad_norm": 6.061465263366699,
      "learning_rate": 1.3763763763763765e-06,
      "loss": 0.1552,
      "step": 86150
    },
    {
      "epoch": 258.74,
      "grad_norm": 29.39352035522461,
      "learning_rate": 1.3753753753753755e-06,
      "loss": 0.1896,
      "step": 86160
    },
    {
      "epoch": 258.77,
      "grad_norm": 7.439678192138672,
      "learning_rate": 1.3743743743743745e-06,
      "loss": 0.2054,
      "step": 86170
    },
    {
      "epoch": 258.8,
      "grad_norm": 5.902533054351807,
      "learning_rate": 1.3733733733733733e-06,
      "loss": 0.2156,
      "step": 86180
    },
    {
      "epoch": 258.83,
      "grad_norm": 7.626037120819092,
      "learning_rate": 1.3723723723723725e-06,
      "loss": 0.2065,
      "step": 86190
    },
    {
      "epoch": 258.86,
      "grad_norm": 7.087588787078857,
      "learning_rate": 1.3713713713713714e-06,
      "loss": 0.1678,
      "step": 86200
    },
    {
      "epoch": 258.89,
      "grad_norm": 10.597246170043945,
      "learning_rate": 1.3703703703703706e-06,
      "loss": 0.1791,
      "step": 86210
    },
    {
      "epoch": 258.92,
      "grad_norm": 11.490535736083984,
      "learning_rate": 1.3693693693693694e-06,
      "loss": 0.2117,
      "step": 86220
    },
    {
      "epoch": 258.95,
      "grad_norm": 12.363826751708984,
      "learning_rate": 1.3683683683683684e-06,
      "loss": 0.1994,
      "step": 86230
    },
    {
      "epoch": 258.98,
      "grad_norm": 6.386078357696533,
      "learning_rate": 1.3673673673673674e-06,
      "loss": 0.1599,
      "step": 86240
    },
    {
      "epoch": 259.0,
      "eval_accuracy": 0.9208,
      "eval_loss": 0.35299789905548096,
      "eval_runtime": 30.3733,
      "eval_samples_per_second": 329.237,
      "eval_steps_per_second": 1.317,
      "step": 86247
    },
    {
      "epoch": 259.01,
      "grad_norm": 7.6022443771362305,
      "learning_rate": 1.3663663663663664e-06,
      "loss": 0.1588,
      "step": 86250
    },
    {
      "epoch": 259.04,
      "grad_norm": 12.167149543762207,
      "learning_rate": 1.3653653653653653e-06,
      "loss": 0.1914,
      "step": 86260
    },
    {
      "epoch": 259.07,
      "grad_norm": 17.485782623291016,
      "learning_rate": 1.3643643643643645e-06,
      "loss": 0.1876,
      "step": 86270
    },
    {
      "epoch": 259.1,
      "grad_norm": 7.604406356811523,
      "learning_rate": 1.3633633633633637e-06,
      "loss": 0.1883,
      "step": 86280
    },
    {
      "epoch": 259.13,
      "grad_norm": 7.886700630187988,
      "learning_rate": 1.3623623623623625e-06,
      "loss": 0.208,
      "step": 86290
    },
    {
      "epoch": 259.16,
      "grad_norm": 9.948296546936035,
      "learning_rate": 1.3613613613613615e-06,
      "loss": 0.2073,
      "step": 86300
    },
    {
      "epoch": 259.19,
      "grad_norm": 7.321438312530518,
      "learning_rate": 1.3603603603603606e-06,
      "loss": 0.1709,
      "step": 86310
    },
    {
      "epoch": 259.22,
      "grad_norm": 9.707128524780273,
      "learning_rate": 1.3593593593593596e-06,
      "loss": 0.1802,
      "step": 86320
    },
    {
      "epoch": 259.25,
      "grad_norm": 6.3317999839782715,
      "learning_rate": 1.3583583583583584e-06,
      "loss": 0.1999,
      "step": 86330
    },
    {
      "epoch": 259.28,
      "grad_norm": 3.123000383377075,
      "learning_rate": 1.3573573573573576e-06,
      "loss": 0.1983,
      "step": 86340
    },
    {
      "epoch": 259.31,
      "grad_norm": 10.582231521606445,
      "learning_rate": 1.3563563563563564e-06,
      "loss": 0.1768,
      "step": 86350
    },
    {
      "epoch": 259.34,
      "grad_norm": 8.16207218170166,
      "learning_rate": 1.3553553553553556e-06,
      "loss": 0.2042,
      "step": 86360
    },
    {
      "epoch": 259.37,
      "grad_norm": 6.210763454437256,
      "learning_rate": 1.3543543543543545e-06,
      "loss": 0.1411,
      "step": 86370
    },
    {
      "epoch": 259.4,
      "grad_norm": 9.150815963745117,
      "learning_rate": 1.3533533533533535e-06,
      "loss": 0.176,
      "step": 86380
    },
    {
      "epoch": 259.43,
      "grad_norm": 16.620712280273438,
      "learning_rate": 1.3523523523523525e-06,
      "loss": 0.2256,
      "step": 86390
    },
    {
      "epoch": 259.46,
      "grad_norm": 6.8983659744262695,
      "learning_rate": 1.3513513513513515e-06,
      "loss": 0.1978,
      "step": 86400
    },
    {
      "epoch": 259.49,
      "grad_norm": 8.530314445495605,
      "learning_rate": 1.3503503503503503e-06,
      "loss": 0.1961,
      "step": 86410
    },
    {
      "epoch": 259.52,
      "grad_norm": 10.918275833129883,
      "learning_rate": 1.3493493493493495e-06,
      "loss": 0.1963,
      "step": 86420
    },
    {
      "epoch": 259.55,
      "grad_norm": 9.83089542388916,
      "learning_rate": 1.3483483483483484e-06,
      "loss": 0.1879,
      "step": 86430
    },
    {
      "epoch": 259.58,
      "grad_norm": 10.315208435058594,
      "learning_rate": 1.3473473473473476e-06,
      "loss": 0.1901,
      "step": 86440
    },
    {
      "epoch": 259.61,
      "grad_norm": 6.406423091888428,
      "learning_rate": 1.3463463463463464e-06,
      "loss": 0.195,
      "step": 86450
    },
    {
      "epoch": 259.64,
      "grad_norm": 5.129600524902344,
      "learning_rate": 1.3453453453453454e-06,
      "loss": 0.1916,
      "step": 86460
    },
    {
      "epoch": 259.67,
      "grad_norm": 10.63400936126709,
      "learning_rate": 1.3443443443443444e-06,
      "loss": 0.192,
      "step": 86470
    },
    {
      "epoch": 259.7,
      "grad_norm": 8.510932922363281,
      "learning_rate": 1.3433433433433434e-06,
      "loss": 0.1925,
      "step": 86480
    },
    {
      "epoch": 259.73,
      "grad_norm": 25.825260162353516,
      "learning_rate": 1.3423423423423422e-06,
      "loss": 0.1928,
      "step": 86490
    },
    {
      "epoch": 259.76,
      "grad_norm": 10.60590648651123,
      "learning_rate": 1.3413413413413415e-06,
      "loss": 0.2013,
      "step": 86500
    },
    {
      "epoch": 259.79,
      "grad_norm": 8.945733070373535,
      "learning_rate": 1.3403403403403403e-06,
      "loss": 0.1931,
      "step": 86510
    },
    {
      "epoch": 259.82,
      "grad_norm": 12.120131492614746,
      "learning_rate": 1.3393393393393395e-06,
      "loss": 0.1505,
      "step": 86520
    },
    {
      "epoch": 259.85,
      "grad_norm": 14.895244598388672,
      "learning_rate": 1.3383383383383383e-06,
      "loss": 0.172,
      "step": 86530
    },
    {
      "epoch": 259.88,
      "grad_norm": 7.067532062530518,
      "learning_rate": 1.3373373373373376e-06,
      "loss": 0.1612,
      "step": 86540
    },
    {
      "epoch": 259.91,
      "grad_norm": 7.113483905792236,
      "learning_rate": 1.3363363363363366e-06,
      "loss": 0.1809,
      "step": 86550
    },
    {
      "epoch": 259.94,
      "grad_norm": 9.33173656463623,
      "learning_rate": 1.3353353353353354e-06,
      "loss": 0.2155,
      "step": 86560
    },
    {
      "epoch": 259.97,
      "grad_norm": 6.294746398925781,
      "learning_rate": 1.3343343343343346e-06,
      "loss": 0.1814,
      "step": 86570
    },
    {
      "epoch": 260.0,
      "grad_norm": 58.41077423095703,
      "learning_rate": 1.3333333333333334e-06,
      "loss": 0.2455,
      "step": 86580
    },
    {
      "epoch": 260.0,
      "eval_accuracy": 0.9206,
      "eval_loss": 0.3507091999053955,
      "eval_runtime": 30.5192,
      "eval_samples_per_second": 327.663,
      "eval_steps_per_second": 1.311,
      "step": 86580
    },
    {
      "epoch": 260.03,
      "grad_norm": 13.102090835571289,
      "learning_rate": 1.3323323323323326e-06,
      "loss": 0.1696,
      "step": 86590
    },
    {
      "epoch": 260.06,
      "grad_norm": 9.358597755432129,
      "learning_rate": 1.3313313313313315e-06,
      "loss": 0.1605,
      "step": 86600
    },
    {
      "epoch": 260.09,
      "grad_norm": 8.082695960998535,
      "learning_rate": 1.3303303303303305e-06,
      "loss": 0.1622,
      "step": 86610
    },
    {
      "epoch": 260.12,
      "grad_norm": 8.858349800109863,
      "learning_rate": 1.3293293293293295e-06,
      "loss": 0.1604,
      "step": 86620
    },
    {
      "epoch": 260.15,
      "grad_norm": 7.153818130493164,
      "learning_rate": 1.3283283283283285e-06,
      "loss": 0.186,
      "step": 86630
    },
    {
      "epoch": 260.18,
      "grad_norm": 10.017526626586914,
      "learning_rate": 1.3273273273273273e-06,
      "loss": 0.2047,
      "step": 86640
    },
    {
      "epoch": 260.21,
      "grad_norm": 13.12317943572998,
      "learning_rate": 1.3263263263263265e-06,
      "loss": 0.197,
      "step": 86650
    },
    {
      "epoch": 260.24,
      "grad_norm": 9.063514709472656,
      "learning_rate": 1.3253253253253253e-06,
      "loss": 0.1484,
      "step": 86660
    },
    {
      "epoch": 260.27,
      "grad_norm": 7.464595317840576,
      "learning_rate": 1.3243243243243246e-06,
      "loss": 0.1525,
      "step": 86670
    },
    {
      "epoch": 260.3,
      "grad_norm": 6.013031005859375,
      "learning_rate": 1.3233233233233234e-06,
      "loss": 0.1684,
      "step": 86680
    },
    {
      "epoch": 260.33,
      "grad_norm": 4.126685619354248,
      "learning_rate": 1.3223223223223224e-06,
      "loss": 0.1514,
      "step": 86690
    },
    {
      "epoch": 260.36,
      "grad_norm": 6.589600563049316,
      "learning_rate": 1.3213213213213214e-06,
      "loss": 0.176,
      "step": 86700
    },
    {
      "epoch": 260.39,
      "grad_norm": 10.44698429107666,
      "learning_rate": 1.3203203203203204e-06,
      "loss": 0.2296,
      "step": 86710
    },
    {
      "epoch": 260.42,
      "grad_norm": 5.38411283493042,
      "learning_rate": 1.3193193193193192e-06,
      "loss": 0.1989,
      "step": 86720
    },
    {
      "epoch": 260.45,
      "grad_norm": 13.676097869873047,
      "learning_rate": 1.3183183183183185e-06,
      "loss": 0.1805,
      "step": 86730
    },
    {
      "epoch": 260.48,
      "grad_norm": 7.307704925537109,
      "learning_rate": 1.3173173173173173e-06,
      "loss": 0.1932,
      "step": 86740
    },
    {
      "epoch": 260.51,
      "grad_norm": 9.191420555114746,
      "learning_rate": 1.3163163163163165e-06,
      "loss": 0.2289,
      "step": 86750
    },
    {
      "epoch": 260.54,
      "grad_norm": 9.891104698181152,
      "learning_rate": 1.3153153153153153e-06,
      "loss": 0.2185,
      "step": 86760
    },
    {
      "epoch": 260.57,
      "grad_norm": 8.165251731872559,
      "learning_rate": 1.3143143143143143e-06,
      "loss": 0.212,
      "step": 86770
    },
    {
      "epoch": 260.6,
      "grad_norm": 7.970928192138672,
      "learning_rate": 1.3133133133133134e-06,
      "loss": 0.2087,
      "step": 86780
    },
    {
      "epoch": 260.63,
      "grad_norm": 11.913784980773926,
      "learning_rate": 1.3123123123123124e-06,
      "loss": 0.214,
      "step": 86790
    },
    {
      "epoch": 260.66,
      "grad_norm": 8.810667037963867,
      "learning_rate": 1.3113113113113112e-06,
      "loss": 0.1864,
      "step": 86800
    },
    {
      "epoch": 260.69,
      "grad_norm": 15.070138931274414,
      "learning_rate": 1.3103103103103104e-06,
      "loss": 0.182,
      "step": 86810
    },
    {
      "epoch": 260.72,
      "grad_norm": 8.279069900512695,
      "learning_rate": 1.3093093093093096e-06,
      "loss": 0.2031,
      "step": 86820
    },
    {
      "epoch": 260.75,
      "grad_norm": 6.313838005065918,
      "learning_rate": 1.3083083083083084e-06,
      "loss": 0.149,
      "step": 86830
    },
    {
      "epoch": 260.78,
      "grad_norm": 6.434418678283691,
      "learning_rate": 1.3073073073073075e-06,
      "loss": 0.1935,
      "step": 86840
    },
    {
      "epoch": 260.81,
      "grad_norm": 9.396056175231934,
      "learning_rate": 1.3063063063063065e-06,
      "loss": 0.18,
      "step": 86850
    },
    {
      "epoch": 260.84,
      "grad_norm": 9.642794609069824,
      "learning_rate": 1.3053053053053055e-06,
      "loss": 0.1835,
      "step": 86860
    },
    {
      "epoch": 260.87,
      "grad_norm": 10.643091201782227,
      "learning_rate": 1.3043043043043043e-06,
      "loss": 0.2227,
      "step": 86870
    },
    {
      "epoch": 260.9,
      "grad_norm": 6.99910306930542,
      "learning_rate": 1.3033033033033035e-06,
      "loss": 0.1742,
      "step": 86880
    },
    {
      "epoch": 260.93,
      "grad_norm": 10.86156177520752,
      "learning_rate": 1.3023023023023023e-06,
      "loss": 0.2155,
      "step": 86890
    },
    {
      "epoch": 260.96,
      "grad_norm": 9.934700012207031,
      "learning_rate": 1.3013013013013016e-06,
      "loss": 0.2206,
      "step": 86900
    },
    {
      "epoch": 260.99,
      "grad_norm": 15.605317115783691,
      "learning_rate": 1.3003003003003004e-06,
      "loss": 0.2159,
      "step": 86910
    },
    {
      "epoch": 261.0,
      "eval_accuracy": 0.9224,
      "eval_loss": 0.3509752154350281,
      "eval_runtime": 30.518,
      "eval_samples_per_second": 327.675,
      "eval_steps_per_second": 1.311,
      "step": 86913
    },
    {
      "epoch": 261.02,
      "grad_norm": 12.005228042602539,
      "learning_rate": 1.2992992992992994e-06,
      "loss": 0.197,
      "step": 86920
    },
    {
      "epoch": 261.05,
      "grad_norm": 9.42856216430664,
      "learning_rate": 1.2982982982982984e-06,
      "loss": 0.1817,
      "step": 86930
    },
    {
      "epoch": 261.08,
      "grad_norm": 14.269763946533203,
      "learning_rate": 1.2972972972972974e-06,
      "loss": 0.1648,
      "step": 86940
    },
    {
      "epoch": 261.11,
      "grad_norm": 8.400077819824219,
      "learning_rate": 1.2962962962962962e-06,
      "loss": 0.1817,
      "step": 86950
    },
    {
      "epoch": 261.14,
      "grad_norm": 6.29348087310791,
      "learning_rate": 1.2952952952952955e-06,
      "loss": 0.1809,
      "step": 86960
    },
    {
      "epoch": 261.17,
      "grad_norm": 9.137596130371094,
      "learning_rate": 1.2942942942942943e-06,
      "loss": 0.2049,
      "step": 86970
    },
    {
      "epoch": 261.2,
      "grad_norm": 16.597759246826172,
      "learning_rate": 1.2932932932932935e-06,
      "loss": 0.2003,
      "step": 86980
    },
    {
      "epoch": 261.23,
      "grad_norm": 9.205568313598633,
      "learning_rate": 1.2922922922922923e-06,
      "loss": 0.1879,
      "step": 86990
    },
    {
      "epoch": 261.26,
      "grad_norm": 10.825709342956543,
      "learning_rate": 1.2912912912912913e-06,
      "loss": 0.2025,
      "step": 87000
    },
    {
      "epoch": 261.29,
      "grad_norm": 5.23825740814209,
      "learning_rate": 1.2902902902902904e-06,
      "loss": 0.1745,
      "step": 87010
    },
    {
      "epoch": 261.32,
      "grad_norm": 11.605137825012207,
      "learning_rate": 1.2892892892892894e-06,
      "loss": 0.1454,
      "step": 87020
    },
    {
      "epoch": 261.35,
      "grad_norm": 9.723823547363281,
      "learning_rate": 1.2882882882882882e-06,
      "loss": 0.2071,
      "step": 87030
    },
    {
      "epoch": 261.38,
      "grad_norm": 8.525981903076172,
      "learning_rate": 1.2872872872872874e-06,
      "loss": 0.1963,
      "step": 87040
    },
    {
      "epoch": 261.41,
      "grad_norm": 5.417150974273682,
      "learning_rate": 1.2862862862862862e-06,
      "loss": 0.2083,
      "step": 87050
    },
    {
      "epoch": 261.44,
      "grad_norm": 11.19099235534668,
      "learning_rate": 1.2852852852852854e-06,
      "loss": 0.1935,
      "step": 87060
    },
    {
      "epoch": 261.47,
      "grad_norm": 16.607933044433594,
      "learning_rate": 1.2842842842842843e-06,
      "loss": 0.207,
      "step": 87070
    },
    {
      "epoch": 261.5,
      "grad_norm": 7.416262626647949,
      "learning_rate": 1.2832832832832835e-06,
      "loss": 0.1698,
      "step": 87080
    },
    {
      "epoch": 261.53,
      "grad_norm": 32.015960693359375,
      "learning_rate": 1.2822822822822825e-06,
      "loss": 0.1606,
      "step": 87090
    },
    {
      "epoch": 261.56,
      "grad_norm": 8.5110502243042,
      "learning_rate": 1.2812812812812813e-06,
      "loss": 0.158,
      "step": 87100
    },
    {
      "epoch": 261.59,
      "grad_norm": 7.608427047729492,
      "learning_rate": 1.2802802802802805e-06,
      "loss": 0.1866,
      "step": 87110
    },
    {
      "epoch": 261.62,
      "grad_norm": 10.826437950134277,
      "learning_rate": 1.2792792792792793e-06,
      "loss": 0.2467,
      "step": 87120
    },
    {
      "epoch": 261.65,
      "grad_norm": 13.416484832763672,
      "learning_rate": 1.2782782782782786e-06,
      "loss": 0.1893,
      "step": 87130
    },
    {
      "epoch": 261.68,
      "grad_norm": 6.2040863037109375,
      "learning_rate": 1.2772772772772774e-06,
      "loss": 0.2175,
      "step": 87140
    },
    {
      "epoch": 261.71,
      "grad_norm": 10.095404624938965,
      "learning_rate": 1.2762762762762764e-06,
      "loss": 0.1819,
      "step": 87150
    },
    {
      "epoch": 261.74,
      "grad_norm": 9.410802841186523,
      "learning_rate": 1.2752752752752754e-06,
      "loss": 0.1832,
      "step": 87160
    },
    {
      "epoch": 261.77,
      "grad_norm": 6.399357318878174,
      "learning_rate": 1.2742742742742744e-06,
      "loss": 0.2011,
      "step": 87170
    },
    {
      "epoch": 261.8,
      "grad_norm": 6.508025646209717,
      "learning_rate": 1.2732732732732732e-06,
      "loss": 0.2,
      "step": 87180
    },
    {
      "epoch": 261.83,
      "grad_norm": 10.548603057861328,
      "learning_rate": 1.2722722722722725e-06,
      "loss": 0.212,
      "step": 87190
    },
    {
      "epoch": 261.86,
      "grad_norm": 9.975257873535156,
      "learning_rate": 1.2712712712712713e-06,
      "loss": 0.2052,
      "step": 87200
    },
    {
      "epoch": 261.89,
      "grad_norm": 5.850875377655029,
      "learning_rate": 1.2702702702702705e-06,
      "loss": 0.1655,
      "step": 87210
    },
    {
      "epoch": 261.92,
      "grad_norm": 12.088890075683594,
      "learning_rate": 1.2692692692692693e-06,
      "loss": 0.2096,
      "step": 87220
    },
    {
      "epoch": 261.95,
      "grad_norm": 7.122886657714844,
      "learning_rate": 1.2682682682682683e-06,
      "loss": 0.214,
      "step": 87230
    },
    {
      "epoch": 261.98,
      "grad_norm": 12.097441673278809,
      "learning_rate": 1.2672672672672674e-06,
      "loss": 0.2032,
      "step": 87240
    },
    {
      "epoch": 262.0,
      "eval_accuracy": 0.9215,
      "eval_loss": 0.3502107262611389,
      "eval_runtime": 29.873,
      "eval_samples_per_second": 334.75,
      "eval_steps_per_second": 1.339,
      "step": 87246
    },
    {
      "epoch": 262.01,
      "grad_norm": 5.921419143676758,
      "learning_rate": 1.2662662662662664e-06,
      "loss": 0.1203,
      "step": 87250
    },
    {
      "epoch": 262.04,
      "grad_norm": 10.00297737121582,
      "learning_rate": 1.2652652652652652e-06,
      "loss": 0.213,
      "step": 87260
    },
    {
      "epoch": 262.07,
      "grad_norm": 4.69972562789917,
      "learning_rate": 1.2642642642642644e-06,
      "loss": 0.1786,
      "step": 87270
    },
    {
      "epoch": 262.1,
      "grad_norm": 5.844058036804199,
      "learning_rate": 1.2632632632632632e-06,
      "loss": 0.1818,
      "step": 87280
    },
    {
      "epoch": 262.13,
      "grad_norm": 11.169988632202148,
      "learning_rate": 1.2622622622622624e-06,
      "loss": 0.2201,
      "step": 87290
    },
    {
      "epoch": 262.16,
      "grad_norm": 12.559307098388672,
      "learning_rate": 1.2612612612612613e-06,
      "loss": 0.1937,
      "step": 87300
    },
    {
      "epoch": 262.19,
      "grad_norm": 6.218175411224365,
      "learning_rate": 1.2602602602602605e-06,
      "loss": 0.1619,
      "step": 87310
    },
    {
      "epoch": 262.22,
      "grad_norm": 15.785118103027344,
      "learning_rate": 1.2592592592592593e-06,
      "loss": 0.1753,
      "step": 87320
    },
    {
      "epoch": 262.25,
      "grad_norm": 8.364099502563477,
      "learning_rate": 1.2582582582582583e-06,
      "loss": 0.202,
      "step": 87330
    },
    {
      "epoch": 262.28,
      "grad_norm": 7.594366550445557,
      "learning_rate": 1.2572572572572573e-06,
      "loss": 0.1986,
      "step": 87340
    },
    {
      "epoch": 262.31,
      "grad_norm": 8.322405815124512,
      "learning_rate": 1.2562562562562563e-06,
      "loss": 0.1737,
      "step": 87350
    },
    {
      "epoch": 262.34,
      "grad_norm": 5.311370849609375,
      "learning_rate": 1.2552552552552556e-06,
      "loss": 0.1553,
      "step": 87360
    },
    {
      "epoch": 262.37,
      "grad_norm": 10.21589469909668,
      "learning_rate": 1.2542542542542544e-06,
      "loss": 0.2175,
      "step": 87370
    },
    {
      "epoch": 262.4,
      "grad_norm": 9.362808227539062,
      "learning_rate": 1.2532532532532534e-06,
      "loss": 0.2082,
      "step": 87380
    },
    {
      "epoch": 262.43,
      "grad_norm": 11.36048698425293,
      "learning_rate": 1.2522522522522524e-06,
      "loss": 0.1781,
      "step": 87390
    },
    {
      "epoch": 262.46,
      "grad_norm": 15.735811233520508,
      "learning_rate": 1.2512512512512514e-06,
      "loss": 0.2067,
      "step": 87400
    },
    {
      "epoch": 262.49,
      "grad_norm": 4.522373199462891,
      "learning_rate": 1.2502502502502502e-06,
      "loss": 0.1798,
      "step": 87410
    },
    {
      "epoch": 262.52,
      "grad_norm": 13.891327857971191,
      "learning_rate": 1.2492492492492493e-06,
      "loss": 0.1809,
      "step": 87420
    },
    {
      "epoch": 262.55,
      "grad_norm": 9.970108032226562,
      "learning_rate": 1.2482482482482483e-06,
      "loss": 0.1623,
      "step": 87430
    },
    {
      "epoch": 262.58,
      "grad_norm": 6.586647987365723,
      "learning_rate": 1.2472472472472473e-06,
      "loss": 0.2033,
      "step": 87440
    },
    {
      "epoch": 262.61,
      "grad_norm": 8.965142250061035,
      "learning_rate": 1.2462462462462463e-06,
      "loss": 0.1558,
      "step": 87450
    },
    {
      "epoch": 262.64,
      "grad_norm": 10.067773818969727,
      "learning_rate": 1.2452452452452453e-06,
      "loss": 0.1886,
      "step": 87460
    },
    {
      "epoch": 262.67,
      "grad_norm": 7.841918468475342,
      "learning_rate": 1.2442442442442444e-06,
      "loss": 0.2055,
      "step": 87470
    },
    {
      "epoch": 262.7,
      "grad_norm": 7.875239372253418,
      "learning_rate": 1.2432432432432434e-06,
      "loss": 0.1775,
      "step": 87480
    },
    {
      "epoch": 262.73,
      "grad_norm": 7.245392799377441,
      "learning_rate": 1.2422422422422424e-06,
      "loss": 0.216,
      "step": 87490
    },
    {
      "epoch": 262.76,
      "grad_norm": 5.624518871307373,
      "learning_rate": 1.2412412412412414e-06,
      "loss": 0.1924,
      "step": 87500
    },
    {
      "epoch": 262.79,
      "grad_norm": 9.525177955627441,
      "learning_rate": 1.2402402402402404e-06,
      "loss": 0.1948,
      "step": 87510
    },
    {
      "epoch": 262.82,
      "grad_norm": 7.662585258483887,
      "learning_rate": 1.2392392392392394e-06,
      "loss": 0.183,
      "step": 87520
    },
    {
      "epoch": 262.85,
      "grad_norm": 7.151435375213623,
      "learning_rate": 1.2382382382382385e-06,
      "loss": 0.1775,
      "step": 87530
    },
    {
      "epoch": 262.88,
      "grad_norm": 11.87088394165039,
      "learning_rate": 1.2372372372372375e-06,
      "loss": 0.1695,
      "step": 87540
    },
    {
      "epoch": 262.91,
      "grad_norm": 9.37271499633789,
      "learning_rate": 1.2362362362362363e-06,
      "loss": 0.188,
      "step": 87550
    },
    {
      "epoch": 262.94,
      "grad_norm": 11.052803993225098,
      "learning_rate": 1.2352352352352353e-06,
      "loss": 0.2105,
      "step": 87560
    },
    {
      "epoch": 262.97,
      "grad_norm": 7.413532733917236,
      "learning_rate": 1.2342342342342343e-06,
      "loss": 0.1453,
      "step": 87570
    },
    {
      "epoch": 263.0,
      "eval_accuracy": 0.9225,
      "eval_loss": 0.35011398792266846,
      "eval_runtime": 30.2795,
      "eval_samples_per_second": 330.256,
      "eval_steps_per_second": 1.321,
      "step": 87579
    },
    {
      "epoch": 263.0,
      "grad_norm": 8.52251148223877,
      "learning_rate": 1.2332332332332333e-06,
      "loss": 0.1661,
      "step": 87580
    },
    {
      "epoch": 263.03,
      "grad_norm": 12.211222648620605,
      "learning_rate": 1.2322322322322324e-06,
      "loss": 0.169,
      "step": 87590
    },
    {
      "epoch": 263.06,
      "grad_norm": 11.529379844665527,
      "learning_rate": 1.2312312312312314e-06,
      "loss": 0.1718,
      "step": 87600
    },
    {
      "epoch": 263.09,
      "grad_norm": 9.70719051361084,
      "learning_rate": 1.2302302302302304e-06,
      "loss": 0.1931,
      "step": 87610
    },
    {
      "epoch": 263.12,
      "grad_norm": 6.746806621551514,
      "learning_rate": 1.2292292292292294e-06,
      "loss": 0.1751,
      "step": 87620
    },
    {
      "epoch": 263.15,
      "grad_norm": 7.364534854888916,
      "learning_rate": 1.2282282282282282e-06,
      "loss": 0.1849,
      "step": 87630
    },
    {
      "epoch": 263.18,
      "grad_norm": 11.46689510345459,
      "learning_rate": 1.2272272272272272e-06,
      "loss": 0.1663,
      "step": 87640
    },
    {
      "epoch": 263.21,
      "grad_norm": 10.42052936553955,
      "learning_rate": 1.2262262262262263e-06,
      "loss": 0.1886,
      "step": 87650
    },
    {
      "epoch": 263.24,
      "grad_norm": 8.532315254211426,
      "learning_rate": 1.2252252252252253e-06,
      "loss": 0.2171,
      "step": 87660
    },
    {
      "epoch": 263.27,
      "grad_norm": 7.573886394500732,
      "learning_rate": 1.2242242242242243e-06,
      "loss": 0.1422,
      "step": 87670
    },
    {
      "epoch": 263.3,
      "grad_norm": 14.422628402709961,
      "learning_rate": 1.2232232232232233e-06,
      "loss": 0.1923,
      "step": 87680
    },
    {
      "epoch": 263.33,
      "grad_norm": 10.853494644165039,
      "learning_rate": 1.2222222222222223e-06,
      "loss": 0.2357,
      "step": 87690
    },
    {
      "epoch": 263.36,
      "grad_norm": 8.683931350708008,
      "learning_rate": 1.2212212212212213e-06,
      "loss": 0.1744,
      "step": 87700
    },
    {
      "epoch": 263.39,
      "grad_norm": 8.250410079956055,
      "learning_rate": 1.2202202202202204e-06,
      "loss": 0.1917,
      "step": 87710
    },
    {
      "epoch": 263.42,
      "grad_norm": 8.349910736083984,
      "learning_rate": 1.2192192192192192e-06,
      "loss": 0.1507,
      "step": 87720
    },
    {
      "epoch": 263.45,
      "grad_norm": 8.414097785949707,
      "learning_rate": 1.2182182182182182e-06,
      "loss": 0.1704,
      "step": 87730
    },
    {
      "epoch": 263.48,
      "grad_norm": 5.883696556091309,
      "learning_rate": 1.2172172172172172e-06,
      "loss": 0.2174,
      "step": 87740
    },
    {
      "epoch": 263.51,
      "grad_norm": 8.280817031860352,
      "learning_rate": 1.2162162162162164e-06,
      "loss": 0.1563,
      "step": 87750
    },
    {
      "epoch": 263.54,
      "grad_norm": 7.610733509063721,
      "learning_rate": 1.2152152152152155e-06,
      "loss": 0.1681,
      "step": 87760
    },
    {
      "epoch": 263.57,
      "grad_norm": 4.330366134643555,
      "learning_rate": 1.2142142142142143e-06,
      "loss": 0.1635,
      "step": 87770
    },
    {
      "epoch": 263.6,
      "grad_norm": 7.145688533782959,
      "learning_rate": 1.2132132132132133e-06,
      "loss": 0.1702,
      "step": 87780
    },
    {
      "epoch": 263.63,
      "grad_norm": 6.271562099456787,
      "learning_rate": 1.2122122122122123e-06,
      "loss": 0.1655,
      "step": 87790
    },
    {
      "epoch": 263.66,
      "grad_norm": 8.450474739074707,
      "learning_rate": 1.2112112112112113e-06,
      "loss": 0.1693,
      "step": 87800
    },
    {
      "epoch": 263.69,
      "grad_norm": 8.695162773132324,
      "learning_rate": 1.2102102102102103e-06,
      "loss": 0.1932,
      "step": 87810
    },
    {
      "epoch": 263.72,
      "grad_norm": 10.97397232055664,
      "learning_rate": 1.2092092092092094e-06,
      "loss": 0.2122,
      "step": 87820
    },
    {
      "epoch": 263.75,
      "grad_norm": 8.756152153015137,
      "learning_rate": 1.2082082082082084e-06,
      "loss": 0.188,
      "step": 87830
    },
    {
      "epoch": 263.78,
      "grad_norm": 6.786104679107666,
      "learning_rate": 1.2072072072072074e-06,
      "loss": 0.2096,
      "step": 87840
    },
    {
      "epoch": 263.81,
      "grad_norm": 7.975718975067139,
      "learning_rate": 1.2062062062062064e-06,
      "loss": 0.1662,
      "step": 87850
    },
    {
      "epoch": 263.84,
      "grad_norm": 9.599562644958496,
      "learning_rate": 1.2052052052052052e-06,
      "loss": 0.2164,
      "step": 87860
    },
    {
      "epoch": 263.87,
      "grad_norm": 8.685479164123535,
      "learning_rate": 1.2042042042042042e-06,
      "loss": 0.1899,
      "step": 87870
    },
    {
      "epoch": 263.9,
      "grad_norm": 7.091196060180664,
      "learning_rate": 1.2032032032032033e-06,
      "loss": 0.1748,
      "step": 87880
    },
    {
      "epoch": 263.93,
      "grad_norm": 7.660480499267578,
      "learning_rate": 1.2022022022022023e-06,
      "loss": 0.2158,
      "step": 87890
    },
    {
      "epoch": 263.96,
      "grad_norm": 11.4668550491333,
      "learning_rate": 1.2012012012012013e-06,
      "loss": 0.1694,
      "step": 87900
    },
    {
      "epoch": 263.99,
      "grad_norm": 7.652865409851074,
      "learning_rate": 1.2002002002002003e-06,
      "loss": 0.1922,
      "step": 87910
    },
    {
      "epoch": 264.0,
      "eval_accuracy": 0.9235,
      "eval_loss": 0.3494115471839905,
      "eval_runtime": 30.3414,
      "eval_samples_per_second": 329.583,
      "eval_steps_per_second": 1.318,
      "step": 87912
    },
    {
      "epoch": 264.02,
      "grad_norm": 9.797829627990723,
      "learning_rate": 1.1991991991991993e-06,
      "loss": 0.1758,
      "step": 87920
    },
    {
      "epoch": 264.05,
      "grad_norm": 12.817207336425781,
      "learning_rate": 1.1981981981981983e-06,
      "loss": 0.1742,
      "step": 87930
    },
    {
      "epoch": 264.08,
      "grad_norm": 7.6134772300720215,
      "learning_rate": 1.1971971971971974e-06,
      "loss": 0.2009,
      "step": 87940
    },
    {
      "epoch": 264.11,
      "grad_norm": 15.030085563659668,
      "learning_rate": 1.1961961961961962e-06,
      "loss": 0.1746,
      "step": 87950
    },
    {
      "epoch": 264.14,
      "grad_norm": 4.278116226196289,
      "learning_rate": 1.1951951951951952e-06,
      "loss": 0.1712,
      "step": 87960
    },
    {
      "epoch": 264.17,
      "grad_norm": 5.93629789352417,
      "learning_rate": 1.1941941941941942e-06,
      "loss": 0.1759,
      "step": 87970
    },
    {
      "epoch": 264.2,
      "grad_norm": 9.395746231079102,
      "learning_rate": 1.1931931931931932e-06,
      "loss": 0.2076,
      "step": 87980
    },
    {
      "epoch": 264.23,
      "grad_norm": 8.906595230102539,
      "learning_rate": 1.1921921921921922e-06,
      "loss": 0.1747,
      "step": 87990
    },
    {
      "epoch": 264.26,
      "grad_norm": 8.331314086914062,
      "learning_rate": 1.1911911911911913e-06,
      "loss": 0.1871,
      "step": 88000
    },
    {
      "epoch": 264.29,
      "grad_norm": 5.739186763763428,
      "learning_rate": 1.1901901901901903e-06,
      "loss": 0.1922,
      "step": 88010
    },
    {
      "epoch": 264.32,
      "grad_norm": 8.588458061218262,
      "learning_rate": 1.1891891891891893e-06,
      "loss": 0.1677,
      "step": 88020
    },
    {
      "epoch": 264.35,
      "grad_norm": 5.164900302886963,
      "learning_rate": 1.1881881881881883e-06,
      "loss": 0.2092,
      "step": 88030
    },
    {
      "epoch": 264.38,
      "grad_norm": 13.131478309631348,
      "learning_rate": 1.1871871871871873e-06,
      "loss": 0.1909,
      "step": 88040
    },
    {
      "epoch": 264.41,
      "grad_norm": 4.863766670227051,
      "learning_rate": 1.1861861861861864e-06,
      "loss": 0.198,
      "step": 88050
    },
    {
      "epoch": 264.44,
      "grad_norm": 7.452154159545898,
      "learning_rate": 1.1851851851851854e-06,
      "loss": 0.1918,
      "step": 88060
    },
    {
      "epoch": 264.47,
      "grad_norm": 7.267922878265381,
      "learning_rate": 1.1841841841841844e-06,
      "loss": 0.21,
      "step": 88070
    },
    {
      "epoch": 264.5,
      "grad_norm": 13.468724250793457,
      "learning_rate": 1.1831831831831834e-06,
      "loss": 0.1791,
      "step": 88080
    },
    {
      "epoch": 264.53,
      "grad_norm": 9.94389820098877,
      "learning_rate": 1.1821821821821822e-06,
      "loss": 0.2213,
      "step": 88090
    },
    {
      "epoch": 264.56,
      "grad_norm": 7.967116832733154,
      "learning_rate": 1.1811811811811812e-06,
      "loss": 0.1689,
      "step": 88100
    },
    {
      "epoch": 264.59,
      "grad_norm": 5.384822368621826,
      "learning_rate": 1.1801801801801803e-06,
      "loss": 0.1933,
      "step": 88110
    },
    {
      "epoch": 264.62,
      "grad_norm": 10.267915725708008,
      "learning_rate": 1.1791791791791793e-06,
      "loss": 0.222,
      "step": 88120
    },
    {
      "epoch": 264.65,
      "grad_norm": 13.364748001098633,
      "learning_rate": 1.1781781781781783e-06,
      "loss": 0.2046,
      "step": 88130
    },
    {
      "epoch": 264.68,
      "grad_norm": 8.720773696899414,
      "learning_rate": 1.1771771771771773e-06,
      "loss": 0.1797,
      "step": 88140
    },
    {
      "epoch": 264.71,
      "grad_norm": 7.089114665985107,
      "learning_rate": 1.1761761761761763e-06,
      "loss": 0.2467,
      "step": 88150
    },
    {
      "epoch": 264.74,
      "grad_norm": 26.587535858154297,
      "learning_rate": 1.1751751751751753e-06,
      "loss": 0.1645,
      "step": 88160
    },
    {
      "epoch": 264.77,
      "grad_norm": 9.832258224487305,
      "learning_rate": 1.1741741741741742e-06,
      "loss": 0.1766,
      "step": 88170
    },
    {
      "epoch": 264.8,
      "grad_norm": 13.741827964782715,
      "learning_rate": 1.1731731731731732e-06,
      "loss": 0.1929,
      "step": 88180
    },
    {
      "epoch": 264.83,
      "grad_norm": 9.594232559204102,
      "learning_rate": 1.1721721721721722e-06,
      "loss": 0.1803,
      "step": 88190
    },
    {
      "epoch": 264.86,
      "grad_norm": 6.149837017059326,
      "learning_rate": 1.1711711711711712e-06,
      "loss": 0.1771,
      "step": 88200
    },
    {
      "epoch": 264.89,
      "grad_norm": 5.523835182189941,
      "learning_rate": 1.1701701701701702e-06,
      "loss": 0.171,
      "step": 88210
    },
    {
      "epoch": 264.92,
      "grad_norm": 5.417267799377441,
      "learning_rate": 1.1691691691691692e-06,
      "loss": 0.152,
      "step": 88220
    },
    {
      "epoch": 264.95,
      "grad_norm": 8.108104705810547,
      "learning_rate": 1.1681681681681683e-06,
      "loss": 0.1736,
      "step": 88230
    },
    {
      "epoch": 264.98,
      "grad_norm": 6.968517303466797,
      "learning_rate": 1.1671671671671673e-06,
      "loss": 0.2038,
      "step": 88240
    },
    {
      "epoch": 265.0,
      "eval_accuracy": 0.9229,
      "eval_loss": 0.348117858171463,
      "eval_runtime": 30.2873,
      "eval_samples_per_second": 330.171,
      "eval_steps_per_second": 1.321,
      "step": 88245
    },
    {
      "epoch": 265.02,
      "grad_norm": 12.232633590698242,
      "learning_rate": 1.1661661661661663e-06,
      "loss": 0.2429,
      "step": 88250
    },
    {
      "epoch": 265.05,
      "grad_norm": 8.29024600982666,
      "learning_rate": 1.1651651651651651e-06,
      "loss": 0.1714,
      "step": 88260
    },
    {
      "epoch": 265.08,
      "grad_norm": 12.637201309204102,
      "learning_rate": 1.1641641641641641e-06,
      "loss": 0.209,
      "step": 88270
    },
    {
      "epoch": 265.11,
      "grad_norm": 9.527416229248047,
      "learning_rate": 1.1631631631631631e-06,
      "loss": 0.1505,
      "step": 88280
    },
    {
      "epoch": 265.14,
      "grad_norm": 10.460866928100586,
      "learning_rate": 1.1621621621621624e-06,
      "loss": 0.1917,
      "step": 88290
    },
    {
      "epoch": 265.17,
      "grad_norm": 13.573442459106445,
      "learning_rate": 1.1611611611611614e-06,
      "loss": 0.2019,
      "step": 88300
    },
    {
      "epoch": 265.2,
      "grad_norm": 14.612360000610352,
      "learning_rate": 1.1601601601601604e-06,
      "loss": 0.2362,
      "step": 88310
    },
    {
      "epoch": 265.23,
      "grad_norm": 8.814614295959473,
      "learning_rate": 1.1591591591591592e-06,
      "loss": 0.1902,
      "step": 88320
    },
    {
      "epoch": 265.26,
      "grad_norm": 8.236265182495117,
      "learning_rate": 1.1581581581581582e-06,
      "loss": 0.2159,
      "step": 88330
    },
    {
      "epoch": 265.29,
      "grad_norm": 15.353469848632812,
      "learning_rate": 1.1571571571571573e-06,
      "loss": 0.1779,
      "step": 88340
    },
    {
      "epoch": 265.32,
      "grad_norm": 6.584125518798828,
      "learning_rate": 1.1561561561561563e-06,
      "loss": 0.1492,
      "step": 88350
    },
    {
      "epoch": 265.35,
      "grad_norm": 9.276329040527344,
      "learning_rate": 1.1551551551551553e-06,
      "loss": 0.1628,
      "step": 88360
    },
    {
      "epoch": 265.38,
      "grad_norm": 12.019739151000977,
      "learning_rate": 1.1541541541541543e-06,
      "loss": 0.2107,
      "step": 88370
    },
    {
      "epoch": 265.41,
      "grad_norm": 7.731393337249756,
      "learning_rate": 1.1531531531531533e-06,
      "loss": 0.1638,
      "step": 88380
    },
    {
      "epoch": 265.44,
      "grad_norm": 14.65656852722168,
      "learning_rate": 1.1521521521521523e-06,
      "loss": 0.1834,
      "step": 88390
    },
    {
      "epoch": 265.47,
      "grad_norm": 11.729318618774414,
      "learning_rate": 1.1511511511511512e-06,
      "loss": 0.1897,
      "step": 88400
    },
    {
      "epoch": 265.5,
      "grad_norm": 9.135805130004883,
      "learning_rate": 1.1501501501501502e-06,
      "loss": 0.1903,
      "step": 88410
    },
    {
      "epoch": 265.53,
      "grad_norm": 9.010747909545898,
      "learning_rate": 1.1491491491491492e-06,
      "loss": 0.1683,
      "step": 88420
    },
    {
      "epoch": 265.56,
      "grad_norm": 6.216284275054932,
      "learning_rate": 1.1481481481481482e-06,
      "loss": 0.1559,
      "step": 88430
    },
    {
      "epoch": 265.59,
      "grad_norm": 17.318334579467773,
      "learning_rate": 1.1471471471471472e-06,
      "loss": 0.1961,
      "step": 88440
    },
    {
      "epoch": 265.62,
      "grad_norm": 8.112624168395996,
      "learning_rate": 1.1461461461461462e-06,
      "loss": 0.185,
      "step": 88450
    },
    {
      "epoch": 265.65,
      "grad_norm": 14.769083023071289,
      "learning_rate": 1.1451451451451453e-06,
      "loss": 0.1631,
      "step": 88460
    },
    {
      "epoch": 265.68,
      "grad_norm": 8.555747985839844,
      "learning_rate": 1.1441441441441443e-06,
      "loss": 0.1843,
      "step": 88470
    },
    {
      "epoch": 265.71,
      "grad_norm": 7.177957534790039,
      "learning_rate": 1.1431431431431433e-06,
      "loss": 0.1689,
      "step": 88480
    },
    {
      "epoch": 265.74,
      "grad_norm": 11.364341735839844,
      "learning_rate": 1.142142142142142e-06,
      "loss": 0.1625,
      "step": 88490
    },
    {
      "epoch": 265.77,
      "grad_norm": 6.455366134643555,
      "learning_rate": 1.1411411411411411e-06,
      "loss": 0.2184,
      "step": 88500
    },
    {
      "epoch": 265.8,
      "grad_norm": 7.029098987579346,
      "learning_rate": 1.1401401401401401e-06,
      "loss": 0.2228,
      "step": 88510
    },
    {
      "epoch": 265.83,
      "grad_norm": 6.166537761688232,
      "learning_rate": 1.1391391391391392e-06,
      "loss": 0.2043,
      "step": 88520
    },
    {
      "epoch": 265.86,
      "grad_norm": 14.4400053024292,
      "learning_rate": 1.1381381381381382e-06,
      "loss": 0.1742,
      "step": 88530
    },
    {
      "epoch": 265.89,
      "grad_norm": 6.704250335693359,
      "learning_rate": 1.1371371371371372e-06,
      "loss": 0.1833,
      "step": 88540
    },
    {
      "epoch": 265.92,
      "grad_norm": 11.387862205505371,
      "learning_rate": 1.1361361361361362e-06,
      "loss": 0.1902,
      "step": 88550
    },
    {
      "epoch": 265.95,
      "grad_norm": 7.025127410888672,
      "learning_rate": 1.1351351351351352e-06,
      "loss": 0.1751,
      "step": 88560
    },
    {
      "epoch": 265.98,
      "grad_norm": 8.46017837524414,
      "learning_rate": 1.1341341341341343e-06,
      "loss": 0.1897,
      "step": 88570
    },
    {
      "epoch": 266.0,
      "eval_accuracy": 0.923,
      "eval_loss": 0.3491867780685425,
      "eval_runtime": 29.8619,
      "eval_samples_per_second": 334.875,
      "eval_steps_per_second": 1.34,
      "step": 88578
    },
    {
      "epoch": 266.01,
      "grad_norm": 4.997856140136719,
      "learning_rate": 1.1331331331331333e-06,
      "loss": 0.2079,
      "step": 88580
    },
    {
      "epoch": 266.04,
      "grad_norm": 7.770934581756592,
      "learning_rate": 1.1321321321321323e-06,
      "loss": 0.2081,
      "step": 88590
    },
    {
      "epoch": 266.07,
      "grad_norm": 8.665752410888672,
      "learning_rate": 1.1311311311311313e-06,
      "loss": 0.2076,
      "step": 88600
    },
    {
      "epoch": 266.1,
      "grad_norm": 8.523094177246094,
      "learning_rate": 1.1301301301301303e-06,
      "loss": 0.1736,
      "step": 88610
    },
    {
      "epoch": 266.13,
      "grad_norm": 8.209228515625,
      "learning_rate": 1.1291291291291293e-06,
      "loss": 0.1983,
      "step": 88620
    },
    {
      "epoch": 266.16,
      "grad_norm": 6.112668514251709,
      "learning_rate": 1.1281281281281281e-06,
      "loss": 0.1871,
      "step": 88630
    },
    {
      "epoch": 266.19,
      "grad_norm": 8.11951732635498,
      "learning_rate": 1.1271271271271272e-06,
      "loss": 0.2057,
      "step": 88640
    },
    {
      "epoch": 266.22,
      "grad_norm": 6.804521083831787,
      "learning_rate": 1.1261261261261262e-06,
      "loss": 0.1939,
      "step": 88650
    },
    {
      "epoch": 266.25,
      "grad_norm": 9.759822845458984,
      "learning_rate": 1.1251251251251252e-06,
      "loss": 0.1743,
      "step": 88660
    },
    {
      "epoch": 266.28,
      "grad_norm": 7.0534820556640625,
      "learning_rate": 1.1241241241241242e-06,
      "loss": 0.2223,
      "step": 88670
    },
    {
      "epoch": 266.31,
      "grad_norm": 10.747279167175293,
      "learning_rate": 1.1231231231231232e-06,
      "loss": 0.1993,
      "step": 88680
    },
    {
      "epoch": 266.34,
      "grad_norm": 10.019169807434082,
      "learning_rate": 1.1221221221221223e-06,
      "loss": 0.1828,
      "step": 88690
    },
    {
      "epoch": 266.37,
      "grad_norm": 8.275425910949707,
      "learning_rate": 1.1211211211211213e-06,
      "loss": 0.2086,
      "step": 88700
    },
    {
      "epoch": 266.4,
      "grad_norm": 6.796770095825195,
      "learning_rate": 1.1201201201201203e-06,
      "loss": 0.201,
      "step": 88710
    },
    {
      "epoch": 266.43,
      "grad_norm": 12.060229301452637,
      "learning_rate": 1.119119119119119e-06,
      "loss": 0.1845,
      "step": 88720
    },
    {
      "epoch": 266.46,
      "grad_norm": 6.999817371368408,
      "learning_rate": 1.1181181181181181e-06,
      "loss": 0.2124,
      "step": 88730
    },
    {
      "epoch": 266.49,
      "grad_norm": 5.849118709564209,
      "learning_rate": 1.1171171171171171e-06,
      "loss": 0.1551,
      "step": 88740
    },
    {
      "epoch": 266.52,
      "grad_norm": 8.721981048583984,
      "learning_rate": 1.1161161161161162e-06,
      "loss": 0.1898,
      "step": 88750
    },
    {
      "epoch": 266.55,
      "grad_norm": 10.10022258758545,
      "learning_rate": 1.1151151151151152e-06,
      "loss": 0.1738,
      "step": 88760
    },
    {
      "epoch": 266.58,
      "grad_norm": 7.734562873840332,
      "learning_rate": 1.1141141141141142e-06,
      "loss": 0.1803,
      "step": 88770
    },
    {
      "epoch": 266.61,
      "grad_norm": 10.31944465637207,
      "learning_rate": 1.1131131131131132e-06,
      "loss": 0.1894,
      "step": 88780
    },
    {
      "epoch": 266.64,
      "grad_norm": 6.646129608154297,
      "learning_rate": 1.1121121121121122e-06,
      "loss": 0.1757,
      "step": 88790
    },
    {
      "epoch": 266.67,
      "grad_norm": 11.610176086425781,
      "learning_rate": 1.111111111111111e-06,
      "loss": 0.1749,
      "step": 88800
    },
    {
      "epoch": 266.7,
      "grad_norm": 8.179570198059082,
      "learning_rate": 1.11011011011011e-06,
      "loss": 0.2077,
      "step": 88810
    },
    {
      "epoch": 266.73,
      "grad_norm": 14.254944801330566,
      "learning_rate": 1.1091091091091093e-06,
      "loss": 0.2055,
      "step": 88820
    },
    {
      "epoch": 266.76,
      "grad_norm": 11.0759916305542,
      "learning_rate": 1.1081081081081083e-06,
      "loss": 0.1952,
      "step": 88830
    },
    {
      "epoch": 266.79,
      "grad_norm": 3.986703634262085,
      "learning_rate": 1.1071071071071073e-06,
      "loss": 0.1819,
      "step": 88840
    },
    {
      "epoch": 266.82,
      "grad_norm": 4.783156394958496,
      "learning_rate": 1.1061061061061063e-06,
      "loss": 0.1751,
      "step": 88850
    },
    {
      "epoch": 266.85,
      "grad_norm": 8.63057804107666,
      "learning_rate": 1.1051051051051051e-06,
      "loss": 0.1911,
      "step": 88860
    },
    {
      "epoch": 266.88,
      "grad_norm": 12.310174942016602,
      "learning_rate": 1.1041041041041042e-06,
      "loss": 0.2033,
      "step": 88870
    },
    {
      "epoch": 266.91,
      "grad_norm": 6.318748474121094,
      "learning_rate": 1.1031031031031032e-06,
      "loss": 0.2013,
      "step": 88880
    },
    {
      "epoch": 266.94,
      "grad_norm": 6.4629716873168945,
      "learning_rate": 1.1021021021021022e-06,
      "loss": 0.1967,
      "step": 88890
    },
    {
      "epoch": 266.97,
      "grad_norm": 9.273946762084961,
      "learning_rate": 1.1011011011011012e-06,
      "loss": 0.1938,
      "step": 88900
    },
    {
      "epoch": 267.0,
      "grad_norm": 8.163138389587402,
      "learning_rate": 1.1001001001001002e-06,
      "loss": 0.1941,
      "step": 88910
    },
    {
      "epoch": 267.0,
      "eval_accuracy": 0.9237,
      "eval_loss": 0.35044243931770325,
      "eval_runtime": 30.2584,
      "eval_samples_per_second": 330.486,
      "eval_steps_per_second": 1.322,
      "step": 88911
    },
    {
      "epoch": 267.03,
      "grad_norm": 10.05307388305664,
      "learning_rate": 1.0990990990990993e-06,
      "loss": 0.1737,
      "step": 88920
    },
    {
      "epoch": 267.06,
      "grad_norm": 9.379416465759277,
      "learning_rate": 1.0980980980980983e-06,
      "loss": 0.1916,
      "step": 88930
    },
    {
      "epoch": 267.09,
      "grad_norm": 8.78580093383789,
      "learning_rate": 1.0970970970970973e-06,
      "loss": 0.1681,
      "step": 88940
    },
    {
      "epoch": 267.12,
      "grad_norm": 6.047155857086182,
      "learning_rate": 1.096096096096096e-06,
      "loss": 0.1902,
      "step": 88950
    },
    {
      "epoch": 267.15,
      "grad_norm": 6.566320419311523,
      "learning_rate": 1.0950950950950951e-06,
      "loss": 0.1748,
      "step": 88960
    },
    {
      "epoch": 267.18,
      "grad_norm": 8.893243789672852,
      "learning_rate": 1.0940940940940941e-06,
      "loss": 0.1861,
      "step": 88970
    },
    {
      "epoch": 267.21,
      "grad_norm": 5.88163948059082,
      "learning_rate": 1.0930930930930932e-06,
      "loss": 0.2433,
      "step": 88980
    },
    {
      "epoch": 267.24,
      "grad_norm": 6.191723823547363,
      "learning_rate": 1.0920920920920922e-06,
      "loss": 0.1817,
      "step": 88990
    },
    {
      "epoch": 267.27,
      "grad_norm": 14.310038566589355,
      "learning_rate": 1.0910910910910912e-06,
      "loss": 0.2045,
      "step": 89000
    },
    {
      "epoch": 267.3,
      "grad_norm": 6.928933620452881,
      "learning_rate": 1.0900900900900902e-06,
      "loss": 0.2047,
      "step": 89010
    },
    {
      "epoch": 267.33,
      "grad_norm": 7.571589469909668,
      "learning_rate": 1.0890890890890892e-06,
      "loss": 0.221,
      "step": 89020
    },
    {
      "epoch": 267.36,
      "grad_norm": 5.559576988220215,
      "learning_rate": 1.088088088088088e-06,
      "loss": 0.1903,
      "step": 89030
    },
    {
      "epoch": 267.39,
      "grad_norm": 9.689630508422852,
      "learning_rate": 1.087087087087087e-06,
      "loss": 0.1691,
      "step": 89040
    },
    {
      "epoch": 267.42,
      "grad_norm": 9.81722640991211,
      "learning_rate": 1.086086086086086e-06,
      "loss": 0.18,
      "step": 89050
    },
    {
      "epoch": 267.45,
      "grad_norm": 7.796257972717285,
      "learning_rate": 1.085085085085085e-06,
      "loss": 0.1601,
      "step": 89060
    },
    {
      "epoch": 267.48,
      "grad_norm": 7.7921576499938965,
      "learning_rate": 1.0840840840840841e-06,
      "loss": 0.165,
      "step": 89070
    },
    {
      "epoch": 267.51,
      "grad_norm": 12.64129638671875,
      "learning_rate": 1.0830830830830831e-06,
      "loss": 0.2079,
      "step": 89080
    },
    {
      "epoch": 267.54,
      "grad_norm": 9.287129402160645,
      "learning_rate": 1.0820820820820821e-06,
      "loss": 0.1818,
      "step": 89090
    },
    {
      "epoch": 267.57,
      "grad_norm": 8.013032913208008,
      "learning_rate": 1.0810810810810812e-06,
      "loss": 0.2171,
      "step": 89100
    },
    {
      "epoch": 267.6,
      "grad_norm": 8.769704818725586,
      "learning_rate": 1.0800800800800802e-06,
      "loss": 0.215,
      "step": 89110
    },
    {
      "epoch": 267.63,
      "grad_norm": 7.647989749908447,
      "learning_rate": 1.0790790790790792e-06,
      "loss": 0.1847,
      "step": 89120
    },
    {
      "epoch": 267.66,
      "grad_norm": 11.269083976745605,
      "learning_rate": 1.0780780780780782e-06,
      "loss": 0.1991,
      "step": 89130
    },
    {
      "epoch": 267.69,
      "grad_norm": 6.775044918060303,
      "learning_rate": 1.0770770770770772e-06,
      "loss": 0.1821,
      "step": 89140
    },
    {
      "epoch": 267.72,
      "grad_norm": 10.502656936645508,
      "learning_rate": 1.0760760760760763e-06,
      "loss": 0.1884,
      "step": 89150
    },
    {
      "epoch": 267.75,
      "grad_norm": 6.691751003265381,
      "learning_rate": 1.0750750750750753e-06,
      "loss": 0.1763,
      "step": 89160
    },
    {
      "epoch": 267.78,
      "grad_norm": 6.925626754760742,
      "learning_rate": 1.074074074074074e-06,
      "loss": 0.2019,
      "step": 89170
    },
    {
      "epoch": 267.81,
      "grad_norm": 4.57762336730957,
      "learning_rate": 1.073073073073073e-06,
      "loss": 0.169,
      "step": 89180
    },
    {
      "epoch": 267.84,
      "grad_norm": 9.11938762664795,
      "learning_rate": 1.0720720720720721e-06,
      "loss": 0.2242,
      "step": 89190
    },
    {
      "epoch": 267.87,
      "grad_norm": 8.24806022644043,
      "learning_rate": 1.0710710710710711e-06,
      "loss": 0.165,
      "step": 89200
    },
    {
      "epoch": 267.9,
      "grad_norm": 7.0392632484436035,
      "learning_rate": 1.0700700700700702e-06,
      "loss": 0.1819,
      "step": 89210
    },
    {
      "epoch": 267.93,
      "grad_norm": 5.508830547332764,
      "learning_rate": 1.0690690690690692e-06,
      "loss": 0.1726,
      "step": 89220
    },
    {
      "epoch": 267.96,
      "grad_norm": 8.525140762329102,
      "learning_rate": 1.0680680680680682e-06,
      "loss": 0.196,
      "step": 89230
    },
    {
      "epoch": 267.99,
      "grad_norm": 9.514302253723145,
      "learning_rate": 1.0670670670670672e-06,
      "loss": 0.197,
      "step": 89240
    },
    {
      "epoch": 268.0,
      "eval_accuracy": 0.923,
      "eval_loss": 0.3504386842250824,
      "eval_runtime": 30.1525,
      "eval_samples_per_second": 331.648,
      "eval_steps_per_second": 1.327,
      "step": 89244
    },
    {
      "epoch": 268.02,
      "grad_norm": 7.771040439605713,
      "learning_rate": 1.0660660660660662e-06,
      "loss": 0.1822,
      "step": 89250
    },
    {
      "epoch": 268.05,
      "grad_norm": 9.047700881958008,
      "learning_rate": 1.065065065065065e-06,
      "loss": 0.1975,
      "step": 89260
    },
    {
      "epoch": 268.08,
      "grad_norm": 12.223260879516602,
      "learning_rate": 1.064064064064064e-06,
      "loss": 0.172,
      "step": 89270
    },
    {
      "epoch": 268.11,
      "grad_norm": 7.560888767242432,
      "learning_rate": 1.063063063063063e-06,
      "loss": 0.1946,
      "step": 89280
    },
    {
      "epoch": 268.14,
      "grad_norm": 5.882164001464844,
      "learning_rate": 1.062062062062062e-06,
      "loss": 0.1905,
      "step": 89290
    },
    {
      "epoch": 268.17,
      "grad_norm": 9.714170455932617,
      "learning_rate": 1.0610610610610611e-06,
      "loss": 0.211,
      "step": 89300
    },
    {
      "epoch": 268.2,
      "grad_norm": 8.206031799316406,
      "learning_rate": 1.0600600600600601e-06,
      "loss": 0.2108,
      "step": 89310
    },
    {
      "epoch": 268.23,
      "grad_norm": 8.07179069519043,
      "learning_rate": 1.0590590590590591e-06,
      "loss": 0.1716,
      "step": 89320
    },
    {
      "epoch": 268.26,
      "grad_norm": 7.633755683898926,
      "learning_rate": 1.0580580580580582e-06,
      "loss": 0.1496,
      "step": 89330
    },
    {
      "epoch": 268.29,
      "grad_norm": 16.435277938842773,
      "learning_rate": 1.0570570570570572e-06,
      "loss": 0.2109,
      "step": 89340
    },
    {
      "epoch": 268.32,
      "grad_norm": 5.637465476989746,
      "learning_rate": 1.0560560560560562e-06,
      "loss": 0.1705,
      "step": 89350
    },
    {
      "epoch": 268.35,
      "grad_norm": 5.428961277008057,
      "learning_rate": 1.0550550550550552e-06,
      "loss": 0.1993,
      "step": 89360
    },
    {
      "epoch": 268.38,
      "grad_norm": 6.796525955200195,
      "learning_rate": 1.0540540540540542e-06,
      "loss": 0.1929,
      "step": 89370
    },
    {
      "epoch": 268.41,
      "grad_norm": 5.077475070953369,
      "learning_rate": 1.0530530530530533e-06,
      "loss": 0.1765,
      "step": 89380
    },
    {
      "epoch": 268.44,
      "grad_norm": 8.571958541870117,
      "learning_rate": 1.0520520520520523e-06,
      "loss": 0.2071,
      "step": 89390
    },
    {
      "epoch": 268.47,
      "grad_norm": 9.17179012298584,
      "learning_rate": 1.051051051051051e-06,
      "loss": 0.1424,
      "step": 89400
    },
    {
      "epoch": 268.5,
      "grad_norm": 18.69776153564453,
      "learning_rate": 1.05005005005005e-06,
      "loss": 0.1913,
      "step": 89410
    },
    {
      "epoch": 268.53,
      "grad_norm": 13.301143646240234,
      "learning_rate": 1.0490490490490491e-06,
      "loss": 0.191,
      "step": 89420
    },
    {
      "epoch": 268.56,
      "grad_norm": 6.574741840362549,
      "learning_rate": 1.0480480480480481e-06,
      "loss": 0.1681,
      "step": 89430
    },
    {
      "epoch": 268.59,
      "grad_norm": 9.567334175109863,
      "learning_rate": 1.0470470470470472e-06,
      "loss": 0.2043,
      "step": 89440
    },
    {
      "epoch": 268.62,
      "grad_norm": 12.359048843383789,
      "learning_rate": 1.0460460460460462e-06,
      "loss": 0.2155,
      "step": 89450
    },
    {
      "epoch": 268.65,
      "grad_norm": 17.99010467529297,
      "learning_rate": 1.0450450450450452e-06,
      "loss": 0.2183,
      "step": 89460
    },
    {
      "epoch": 268.68,
      "grad_norm": 16.681711196899414,
      "learning_rate": 1.0440440440440442e-06,
      "loss": 0.2025,
      "step": 89470
    },
    {
      "epoch": 268.71,
      "grad_norm": 7.914881706237793,
      "learning_rate": 1.0430430430430432e-06,
      "loss": 0.1968,
      "step": 89480
    },
    {
      "epoch": 268.74,
      "grad_norm": 11.171416282653809,
      "learning_rate": 1.042042042042042e-06,
      "loss": 0.1927,
      "step": 89490
    },
    {
      "epoch": 268.77,
      "grad_norm": 7.594454288482666,
      "learning_rate": 1.041041041041041e-06,
      "loss": 0.1879,
      "step": 89500
    },
    {
      "epoch": 268.8,
      "grad_norm": 8.14065170288086,
      "learning_rate": 1.04004004004004e-06,
      "loss": 0.1632,
      "step": 89510
    },
    {
      "epoch": 268.83,
      "grad_norm": 10.314549446105957,
      "learning_rate": 1.039039039039039e-06,
      "loss": 0.178,
      "step": 89520
    },
    {
      "epoch": 268.86,
      "grad_norm": 12.352797508239746,
      "learning_rate": 1.038038038038038e-06,
      "loss": 0.184,
      "step": 89530
    },
    {
      "epoch": 268.89,
      "grad_norm": 6.349777698516846,
      "learning_rate": 1.0370370370370371e-06,
      "loss": 0.1438,
      "step": 89540
    },
    {
      "epoch": 268.92,
      "grad_norm": 15.855127334594727,
      "learning_rate": 1.0360360360360361e-06,
      "loss": 0.1757,
      "step": 89550
    },
    {
      "epoch": 268.95,
      "grad_norm": 13.224331855773926,
      "learning_rate": 1.0350350350350352e-06,
      "loss": 0.2099,
      "step": 89560
    },
    {
      "epoch": 268.98,
      "grad_norm": 12.895501136779785,
      "learning_rate": 1.0340340340340342e-06,
      "loss": 0.1933,
      "step": 89570
    },
    {
      "epoch": 269.0,
      "eval_accuracy": 0.9227,
      "eval_loss": 0.3484911024570465,
      "eval_runtime": 29.8201,
      "eval_samples_per_second": 335.344,
      "eval_steps_per_second": 1.341,
      "step": 89577
    },
    {
      "epoch": 269.01,
      "grad_norm": 5.834805488586426,
      "learning_rate": 1.033033033033033e-06,
      "loss": 0.1807,
      "step": 89580
    },
    {
      "epoch": 269.04,
      "grad_norm": 13.4645414352417,
      "learning_rate": 1.032032032032032e-06,
      "loss": 0.1922,
      "step": 89590
    },
    {
      "epoch": 269.07,
      "grad_norm": 7.7975287437438965,
      "learning_rate": 1.031031031031031e-06,
      "loss": 0.1963,
      "step": 89600
    },
    {
      "epoch": 269.1,
      "grad_norm": 4.892468452453613,
      "learning_rate": 1.03003003003003e-06,
      "loss": 0.1852,
      "step": 89610
    },
    {
      "epoch": 269.13,
      "grad_norm": 7.345088958740234,
      "learning_rate": 1.0290290290290293e-06,
      "loss": 0.2109,
      "step": 89620
    },
    {
      "epoch": 269.16,
      "grad_norm": 4.766383171081543,
      "learning_rate": 1.028028028028028e-06,
      "loss": 0.1757,
      "step": 89630
    },
    {
      "epoch": 269.19,
      "grad_norm": 6.1111345291137695,
      "learning_rate": 1.027027027027027e-06,
      "loss": 0.2072,
      "step": 89640
    },
    {
      "epoch": 269.22,
      "grad_norm": 65.410400390625,
      "learning_rate": 1.0260260260260261e-06,
      "loss": 0.214,
      "step": 89650
    },
    {
      "epoch": 269.25,
      "grad_norm": 6.734743595123291,
      "learning_rate": 1.0250250250250251e-06,
      "loss": 0.1883,
      "step": 89660
    },
    {
      "epoch": 269.28,
      "grad_norm": 24.18647575378418,
      "learning_rate": 1.0240240240240242e-06,
      "loss": 0.1876,
      "step": 89670
    },
    {
      "epoch": 269.31,
      "grad_norm": 11.402543067932129,
      "learning_rate": 1.0230230230230232e-06,
      "loss": 0.1776,
      "step": 89680
    },
    {
      "epoch": 269.34,
      "grad_norm": 7.104577541351318,
      "learning_rate": 1.0220220220220222e-06,
      "loss": 0.1592,
      "step": 89690
    },
    {
      "epoch": 269.37,
      "grad_norm": 8.27873420715332,
      "learning_rate": 1.0210210210210212e-06,
      "loss": 0.1701,
      "step": 89700
    },
    {
      "epoch": 269.4,
      "grad_norm": 5.478456020355225,
      "learning_rate": 1.0200200200200202e-06,
      "loss": 0.1377,
      "step": 89710
    },
    {
      "epoch": 269.43,
      "grad_norm": 9.469157218933105,
      "learning_rate": 1.019019019019019e-06,
      "loss": 0.1949,
      "step": 89720
    },
    {
      "epoch": 269.46,
      "grad_norm": 8.622227668762207,
      "learning_rate": 1.018018018018018e-06,
      "loss": 0.1919,
      "step": 89730
    },
    {
      "epoch": 269.49,
      "grad_norm": 6.333393096923828,
      "learning_rate": 1.017017017017017e-06,
      "loss": 0.1804,
      "step": 89740
    },
    {
      "epoch": 269.52,
      "grad_norm": 15.088855743408203,
      "learning_rate": 1.016016016016016e-06,
      "loss": 0.1574,
      "step": 89750
    },
    {
      "epoch": 269.55,
      "grad_norm": 7.123387813568115,
      "learning_rate": 1.015015015015015e-06,
      "loss": 0.2049,
      "step": 89760
    },
    {
      "epoch": 269.58,
      "grad_norm": 4.878567695617676,
      "learning_rate": 1.0140140140140141e-06,
      "loss": 0.1464,
      "step": 89770
    },
    {
      "epoch": 269.61,
      "grad_norm": 8.94984245300293,
      "learning_rate": 1.0130130130130131e-06,
      "loss": 0.1675,
      "step": 89780
    },
    {
      "epoch": 269.64,
      "grad_norm": 9.760276794433594,
      "learning_rate": 1.0120120120120122e-06,
      "loss": 0.1769,
      "step": 89790
    },
    {
      "epoch": 269.67,
      "grad_norm": 11.690254211425781,
      "learning_rate": 1.011011011011011e-06,
      "loss": 0.1743,
      "step": 89800
    },
    {
      "epoch": 269.7,
      "grad_norm": 9.678510665893555,
      "learning_rate": 1.01001001001001e-06,
      "loss": 0.1898,
      "step": 89810
    },
    {
      "epoch": 269.73,
      "grad_norm": 13.516404151916504,
      "learning_rate": 1.009009009009009e-06,
      "loss": 0.2062,
      "step": 89820
    },
    {
      "epoch": 269.76,
      "grad_norm": 9.148297309875488,
      "learning_rate": 1.008008008008008e-06,
      "loss": 0.1316,
      "step": 89830
    },
    {
      "epoch": 269.79,
      "grad_norm": 6.094362735748291,
      "learning_rate": 1.007007007007007e-06,
      "loss": 0.2133,
      "step": 89840
    },
    {
      "epoch": 269.82,
      "grad_norm": 8.038824081420898,
      "learning_rate": 1.006006006006006e-06,
      "loss": 0.1979,
      "step": 89850
    },
    {
      "epoch": 269.85,
      "grad_norm": 7.922441482543945,
      "learning_rate": 1.005005005005005e-06,
      "loss": 0.1913,
      "step": 89860
    },
    {
      "epoch": 269.88,
      "grad_norm": 19.16408920288086,
      "learning_rate": 1.004004004004004e-06,
      "loss": 0.2256,
      "step": 89870
    },
    {
      "epoch": 269.91,
      "grad_norm": 10.083487510681152,
      "learning_rate": 1.0030030030030031e-06,
      "loss": 0.2166,
      "step": 89880
    },
    {
      "epoch": 269.94,
      "grad_norm": 7.712009906768799,
      "learning_rate": 1.0020020020020021e-06,
      "loss": 0.231,
      "step": 89890
    },
    {
      "epoch": 269.97,
      "grad_norm": 14.2257719039917,
      "learning_rate": 1.0010010010010011e-06,
      "loss": 0.2144,
      "step": 89900
    },
    {
      "epoch": 270.0,
      "grad_norm": 4.269519805908203,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.1585,
      "step": 89910
    },
    {
      "epoch": 270.0,
      "eval_accuracy": 0.9237,
      "eval_loss": 0.34879547357559204,
      "eval_runtime": 30.4828,
      "eval_samples_per_second": 328.053,
      "eval_steps_per_second": 1.312,
      "step": 89910
    },
    {
      "epoch": 270.03,
      "grad_norm": 8.022486686706543,
      "learning_rate": 9.989989989989992e-07,
      "loss": 0.2084,
      "step": 89920
    },
    {
      "epoch": 270.06,
      "grad_norm": 14.142882347106934,
      "learning_rate": 9.979979979979982e-07,
      "loss": 0.1895,
      "step": 89930
    },
    {
      "epoch": 270.09,
      "grad_norm": 8.676467895507812,
      "learning_rate": 9.969969969969972e-07,
      "loss": 0.2046,
      "step": 89940
    },
    {
      "epoch": 270.12,
      "grad_norm": 12.300312042236328,
      "learning_rate": 9.95995995995996e-07,
      "loss": 0.1922,
      "step": 89950
    },
    {
      "epoch": 270.15,
      "grad_norm": 8.576961517333984,
      "learning_rate": 9.94994994994995e-07,
      "loss": 0.1941,
      "step": 89960
    },
    {
      "epoch": 270.18,
      "grad_norm": 13.564138412475586,
      "learning_rate": 9.93993993993994e-07,
      "loss": 0.2159,
      "step": 89970
    },
    {
      "epoch": 270.21,
      "grad_norm": 8.109700202941895,
      "learning_rate": 9.92992992992993e-07,
      "loss": 0.2209,
      "step": 89980
    },
    {
      "epoch": 270.24,
      "grad_norm": 9.770171165466309,
      "learning_rate": 9.91991991991992e-07,
      "loss": 0.1733,
      "step": 89990
    },
    {
      "epoch": 270.27,
      "grad_norm": 8.307682991027832,
      "learning_rate": 9.909909909909911e-07,
      "loss": 0.1948,
      "step": 90000
    },
    {
      "epoch": 270.3,
      "grad_norm": 9.049535751342773,
      "learning_rate": 9.899899899899901e-07,
      "loss": 0.1801,
      "step": 90010
    },
    {
      "epoch": 270.33,
      "grad_norm": 11.670942306518555,
      "learning_rate": 9.889889889889892e-07,
      "loss": 0.1984,
      "step": 90020
    },
    {
      "epoch": 270.36,
      "grad_norm": 8.904656410217285,
      "learning_rate": 9.87987987987988e-07,
      "loss": 0.1821,
      "step": 90030
    },
    {
      "epoch": 270.39,
      "grad_norm": 8.799362182617188,
      "learning_rate": 9.86986986986987e-07,
      "loss": 0.1794,
      "step": 90040
    },
    {
      "epoch": 270.42,
      "grad_norm": 10.60417652130127,
      "learning_rate": 9.85985985985986e-07,
      "loss": 0.2068,
      "step": 90050
    },
    {
      "epoch": 270.45,
      "grad_norm": 3.2962839603424072,
      "learning_rate": 9.84984984984985e-07,
      "loss": 0.1599,
      "step": 90060
    },
    {
      "epoch": 270.48,
      "grad_norm": 9.615166664123535,
      "learning_rate": 9.83983983983984e-07,
      "loss": 0.2165,
      "step": 90070
    },
    {
      "epoch": 270.51,
      "grad_norm": 11.424586296081543,
      "learning_rate": 9.82982982982983e-07,
      "loss": 0.2035,
      "step": 90080
    },
    {
      "epoch": 270.54,
      "grad_norm": 5.625278949737549,
      "learning_rate": 9.81981981981982e-07,
      "loss": 0.1975,
      "step": 90090
    },
    {
      "epoch": 270.57,
      "grad_norm": 10.510758399963379,
      "learning_rate": 9.80980980980981e-07,
      "loss": 0.1805,
      "step": 90100
    },
    {
      "epoch": 270.6,
      "grad_norm": 9.026445388793945,
      "learning_rate": 9.799799799799801e-07,
      "loss": 0.2182,
      "step": 90110
    },
    {
      "epoch": 270.63,
      "grad_norm": 7.9913411140441895,
      "learning_rate": 9.78978978978979e-07,
      "loss": 0.1935,
      "step": 90120
    },
    {
      "epoch": 270.66,
      "grad_norm": 5.624424457550049,
      "learning_rate": 9.77977977977978e-07,
      "loss": 0.2144,
      "step": 90130
    },
    {
      "epoch": 270.69,
      "grad_norm": 10.169697761535645,
      "learning_rate": 9.76976976976977e-07,
      "loss": 0.1965,
      "step": 90140
    },
    {
      "epoch": 270.72,
      "grad_norm": 22.528905868530273,
      "learning_rate": 9.75975975975976e-07,
      "loss": 0.1806,
      "step": 90150
    },
    {
      "epoch": 270.75,
      "grad_norm": 12.388751983642578,
      "learning_rate": 9.749749749749752e-07,
      "loss": 0.194,
      "step": 90160
    },
    {
      "epoch": 270.78,
      "grad_norm": 22.72612762451172,
      "learning_rate": 9.73973973973974e-07,
      "loss": 0.1924,
      "step": 90170
    },
    {
      "epoch": 270.81,
      "grad_norm": 10.033129692077637,
      "learning_rate": 9.72972972972973e-07,
      "loss": 0.1856,
      "step": 90180
    },
    {
      "epoch": 270.84,
      "grad_norm": 6.791257381439209,
      "learning_rate": 9.71971971971972e-07,
      "loss": 0.1686,
      "step": 90190
    },
    {
      "epoch": 270.87,
      "grad_norm": 6.203569412231445,
      "learning_rate": 9.70970970970971e-07,
      "loss": 0.2169,
      "step": 90200
    },
    {
      "epoch": 270.9,
      "grad_norm": 4.654849052429199,
      "learning_rate": 9.6996996996997e-07,
      "loss": 0.1575,
      "step": 90210
    },
    {
      "epoch": 270.93,
      "grad_norm": 9.299792289733887,
      "learning_rate": 9.68968968968969e-07,
      "loss": 0.2335,
      "step": 90220
    },
    {
      "epoch": 270.96,
      "grad_norm": 15.917144775390625,
      "learning_rate": 9.679679679679681e-07,
      "loss": 0.1913,
      "step": 90230
    },
    {
      "epoch": 270.99,
      "grad_norm": 3.4583349227905273,
      "learning_rate": 9.669669669669671e-07,
      "loss": 0.1994,
      "step": 90240
    },
    {
      "epoch": 271.0,
      "eval_accuracy": 0.9223,
      "eval_loss": 0.3487699627876282,
      "eval_runtime": 30.3746,
      "eval_samples_per_second": 329.222,
      "eval_steps_per_second": 1.317,
      "step": 90243
    },
    {
      "epoch": 271.02,
      "grad_norm": 9.225135803222656,
      "learning_rate": 9.659659659659662e-07,
      "loss": 0.1474,
      "step": 90250
    },
    {
      "epoch": 271.05,
      "grad_norm": 7.874370574951172,
      "learning_rate": 9.64964964964965e-07,
      "loss": 0.1708,
      "step": 90260
    },
    {
      "epoch": 271.08,
      "grad_norm": 7.529922008514404,
      "learning_rate": 9.63963963963964e-07,
      "loss": 0.1923,
      "step": 90270
    },
    {
      "epoch": 271.11,
      "grad_norm": 4.974855899810791,
      "learning_rate": 9.62962962962963e-07,
      "loss": 0.1389,
      "step": 90280
    },
    {
      "epoch": 271.14,
      "grad_norm": 8.979231834411621,
      "learning_rate": 9.61961961961962e-07,
      "loss": 0.1618,
      "step": 90290
    },
    {
      "epoch": 271.17,
      "grad_norm": 9.757363319396973,
      "learning_rate": 9.60960960960961e-07,
      "loss": 0.1937,
      "step": 90300
    },
    {
      "epoch": 271.2,
      "grad_norm": 22.210609436035156,
      "learning_rate": 9.5995995995996e-07,
      "loss": 0.2049,
      "step": 90310
    },
    {
      "epoch": 271.23,
      "grad_norm": 6.696158409118652,
      "learning_rate": 9.58958958958959e-07,
      "loss": 0.1944,
      "step": 90320
    },
    {
      "epoch": 271.26,
      "grad_norm": 9.043782234191895,
      "learning_rate": 9.57957957957958e-07,
      "loss": 0.2004,
      "step": 90330
    },
    {
      "epoch": 271.29,
      "grad_norm": 10.279471397399902,
      "learning_rate": 9.569569569569571e-07,
      "loss": 0.2507,
      "step": 90340
    },
    {
      "epoch": 271.32,
      "grad_norm": 18.100419998168945,
      "learning_rate": 9.55955955955956e-07,
      "loss": 0.1725,
      "step": 90350
    },
    {
      "epoch": 271.35,
      "grad_norm": 9.240782737731934,
      "learning_rate": 9.54954954954955e-07,
      "loss": 0.2035,
      "step": 90360
    },
    {
      "epoch": 271.38,
      "grad_norm": 13.811389923095703,
      "learning_rate": 9.53953953953954e-07,
      "loss": 0.1445,
      "step": 90370
    },
    {
      "epoch": 271.41,
      "grad_norm": 6.618707180023193,
      "learning_rate": 9.52952952952953e-07,
      "loss": 0.1896,
      "step": 90380
    },
    {
      "epoch": 271.44,
      "grad_norm": 11.408964157104492,
      "learning_rate": 9.51951951951952e-07,
      "loss": 0.2024,
      "step": 90390
    },
    {
      "epoch": 271.47,
      "grad_norm": 4.692192554473877,
      "learning_rate": 9.50950950950951e-07,
      "loss": 0.1723,
      "step": 90400
    },
    {
      "epoch": 271.5,
      "grad_norm": 6.980201721191406,
      "learning_rate": 9.499499499499499e-07,
      "loss": 0.1945,
      "step": 90410
    },
    {
      "epoch": 271.53,
      "grad_norm": 5.905778884887695,
      "learning_rate": 9.48948948948949e-07,
      "loss": 0.1638,
      "step": 90420
    },
    {
      "epoch": 271.56,
      "grad_norm": 7.097388744354248,
      "learning_rate": 9.479479479479481e-07,
      "loss": 0.1813,
      "step": 90430
    },
    {
      "epoch": 271.59,
      "grad_norm": 8.286694526672363,
      "learning_rate": 9.469469469469471e-07,
      "loss": 0.1513,
      "step": 90440
    },
    {
      "epoch": 271.62,
      "grad_norm": 10.149962425231934,
      "learning_rate": 9.459459459459461e-07,
      "loss": 0.1726,
      "step": 90450
    },
    {
      "epoch": 271.65,
      "grad_norm": 5.5808258056640625,
      "learning_rate": 9.449449449449451e-07,
      "loss": 0.1688,
      "step": 90460
    },
    {
      "epoch": 271.68,
      "grad_norm": 5.092613697052002,
      "learning_rate": 9.43943943943944e-07,
      "loss": 0.1925,
      "step": 90470
    },
    {
      "epoch": 271.71,
      "grad_norm": 7.953741550445557,
      "learning_rate": 9.42942942942943e-07,
      "loss": 0.1498,
      "step": 90480
    },
    {
      "epoch": 271.74,
      "grad_norm": 7.817978858947754,
      "learning_rate": 9.419419419419421e-07,
      "loss": 0.2153,
      "step": 90490
    },
    {
      "epoch": 271.77,
      "grad_norm": 10.804422378540039,
      "learning_rate": 9.409409409409411e-07,
      "loss": 0.1589,
      "step": 90500
    },
    {
      "epoch": 271.8,
      "grad_norm": 9.042570114135742,
      "learning_rate": 9.3993993993994e-07,
      "loss": 0.197,
      "step": 90510
    },
    {
      "epoch": 271.83,
      "grad_norm": 9.822269439697266,
      "learning_rate": 9.38938938938939e-07,
      "loss": 0.2194,
      "step": 90520
    },
    {
      "epoch": 271.86,
      "grad_norm": 7.582170009613037,
      "learning_rate": 9.37937937937938e-07,
      "loss": 0.1553,
      "step": 90530
    },
    {
      "epoch": 271.89,
      "grad_norm": 6.637485504150391,
      "learning_rate": 9.36936936936937e-07,
      "loss": 0.1989,
      "step": 90540
    },
    {
      "epoch": 271.92,
      "grad_norm": 10.629439353942871,
      "learning_rate": 9.35935935935936e-07,
      "loss": 0.1548,
      "step": 90550
    },
    {
      "epoch": 271.95,
      "grad_norm": 14.120959281921387,
      "learning_rate": 9.34934934934935e-07,
      "loss": 0.1805,
      "step": 90560
    },
    {
      "epoch": 271.98,
      "grad_norm": 9.087392807006836,
      "learning_rate": 9.33933933933934e-07,
      "loss": 0.1562,
      "step": 90570
    },
    {
      "epoch": 272.0,
      "eval_accuracy": 0.922,
      "eval_loss": 0.34817421436309814,
      "eval_runtime": 30.2817,
      "eval_samples_per_second": 330.232,
      "eval_steps_per_second": 1.321,
      "step": 90576
    },
    {
      "epoch": 272.01,
      "grad_norm": 8.507233619689941,
      "learning_rate": 9.32932932932933e-07,
      "loss": 0.1531,
      "step": 90580
    },
    {
      "epoch": 272.04,
      "grad_norm": 12.120017051696777,
      "learning_rate": 9.31931931931932e-07,
      "loss": 0.1853,
      "step": 90590
    },
    {
      "epoch": 272.07,
      "grad_norm": 19.745256423950195,
      "learning_rate": 9.30930930930931e-07,
      "loss": 0.2292,
      "step": 90600
    },
    {
      "epoch": 272.1,
      "grad_norm": 8.620681762695312,
      "learning_rate": 9.2992992992993e-07,
      "loss": 0.1933,
      "step": 90610
    },
    {
      "epoch": 272.13,
      "grad_norm": 10.105342864990234,
      "learning_rate": 9.28928928928929e-07,
      "loss": 0.2045,
      "step": 90620
    },
    {
      "epoch": 272.16,
      "grad_norm": 10.691093444824219,
      "learning_rate": 9.27927927927928e-07,
      "loss": 0.1664,
      "step": 90630
    },
    {
      "epoch": 272.19,
      "grad_norm": 6.323592185974121,
      "learning_rate": 9.269269269269269e-07,
      "loss": 0.1645,
      "step": 90640
    },
    {
      "epoch": 272.22,
      "grad_norm": 10.270243644714355,
      "learning_rate": 9.259259259259259e-07,
      "loss": 0.1973,
      "step": 90650
    },
    {
      "epoch": 272.25,
      "grad_norm": 5.415646553039551,
      "learning_rate": 9.24924924924925e-07,
      "loss": 0.1643,
      "step": 90660
    },
    {
      "epoch": 272.28,
      "grad_norm": 8.935891151428223,
      "learning_rate": 9.23923923923924e-07,
      "loss": 0.1725,
      "step": 90670
    },
    {
      "epoch": 272.31,
      "grad_norm": 16.111886978149414,
      "learning_rate": 9.229229229229229e-07,
      "loss": 0.2011,
      "step": 90680
    },
    {
      "epoch": 272.34,
      "grad_norm": 8.689400672912598,
      "learning_rate": 9.219219219219221e-07,
      "loss": 0.198,
      "step": 90690
    },
    {
      "epoch": 272.37,
      "grad_norm": 5.976129531860352,
      "learning_rate": 9.20920920920921e-07,
      "loss": 0.1577,
      "step": 90700
    },
    {
      "epoch": 272.4,
      "grad_norm": 7.185765743255615,
      "learning_rate": 9.1991991991992e-07,
      "loss": 0.1981,
      "step": 90710
    },
    {
      "epoch": 272.43,
      "grad_norm": 13.83364486694336,
      "learning_rate": 9.189189189189191e-07,
      "loss": 0.1939,
      "step": 90720
    },
    {
      "epoch": 272.46,
      "grad_norm": 10.846379280090332,
      "learning_rate": 9.179179179179181e-07,
      "loss": 0.1991,
      "step": 90730
    },
    {
      "epoch": 272.49,
      "grad_norm": 5.492271423339844,
      "learning_rate": 9.16916916916917e-07,
      "loss": 0.2261,
      "step": 90740
    },
    {
      "epoch": 272.52,
      "grad_norm": 8.713478088378906,
      "learning_rate": 9.15915915915916e-07,
      "loss": 0.1927,
      "step": 90750
    },
    {
      "epoch": 272.55,
      "grad_norm": 9.686235427856445,
      "learning_rate": 9.14914914914915e-07,
      "loss": 0.1645,
      "step": 90760
    },
    {
      "epoch": 272.58,
      "grad_norm": 10.741779327392578,
      "learning_rate": 9.13913913913914e-07,
      "loss": 0.2039,
      "step": 90770
    },
    {
      "epoch": 272.61,
      "grad_norm": 13.364851951599121,
      "learning_rate": 9.12912912912913e-07,
      "loss": 0.1872,
      "step": 90780
    },
    {
      "epoch": 272.64,
      "grad_norm": 7.497980117797852,
      "learning_rate": 9.11911911911912e-07,
      "loss": 0.23,
      "step": 90790
    },
    {
      "epoch": 272.67,
      "grad_norm": 12.793495178222656,
      "learning_rate": 9.10910910910911e-07,
      "loss": 0.1734,
      "step": 90800
    },
    {
      "epoch": 272.7,
      "grad_norm": 7.17387580871582,
      "learning_rate": 9.0990990990991e-07,
      "loss": 0.1673,
      "step": 90810
    },
    {
      "epoch": 272.73,
      "grad_norm": 16.309301376342773,
      "learning_rate": 9.08908908908909e-07,
      "loss": 0.2097,
      "step": 90820
    },
    {
      "epoch": 272.76,
      "grad_norm": 9.3052396774292,
      "learning_rate": 9.07907907907908e-07,
      "loss": 0.2266,
      "step": 90830
    },
    {
      "epoch": 272.79,
      "grad_norm": 10.10366153717041,
      "learning_rate": 9.06906906906907e-07,
      "loss": 0.2048,
      "step": 90840
    },
    {
      "epoch": 272.82,
      "grad_norm": 9.367642402648926,
      "learning_rate": 9.05905905905906e-07,
      "loss": 0.2175,
      "step": 90850
    },
    {
      "epoch": 272.85,
      "grad_norm": 7.025120735168457,
      "learning_rate": 9.04904904904905e-07,
      "loss": 0.1872,
      "step": 90860
    },
    {
      "epoch": 272.88,
      "grad_norm": 10.522810935974121,
      "learning_rate": 9.039039039039039e-07,
      "loss": 0.1911,
      "step": 90870
    },
    {
      "epoch": 272.91,
      "grad_norm": 7.691766738891602,
      "learning_rate": 9.029029029029029e-07,
      "loss": 0.1608,
      "step": 90880
    },
    {
      "epoch": 272.94,
      "grad_norm": 5.097179889678955,
      "learning_rate": 9.01901901901902e-07,
      "loss": 0.1941,
      "step": 90890
    },
    {
      "epoch": 272.97,
      "grad_norm": 14.56262493133545,
      "learning_rate": 9.00900900900901e-07,
      "loss": 0.1804,
      "step": 90900
    },
    {
      "epoch": 273.0,
      "eval_accuracy": 0.9214,
      "eval_loss": 0.348690390586853,
      "eval_runtime": 30.4883,
      "eval_samples_per_second": 327.995,
      "eval_steps_per_second": 1.312,
      "step": 90909
    },
    {
      "epoch": 273.0,
      "grad_norm": 6.005967140197754,
      "learning_rate": 8.998998998998999e-07,
      "loss": 0.4341,
      "step": 90910
    },
    {
      "epoch": 273.03,
      "grad_norm": 8.757821083068848,
      "learning_rate": 8.988988988988989e-07,
      "loss": 0.1845,
      "step": 90920
    },
    {
      "epoch": 273.06,
      "grad_norm": 4.981235980987549,
      "learning_rate": 8.978978978978979e-07,
      "loss": 0.1909,
      "step": 90930
    },
    {
      "epoch": 273.09,
      "grad_norm": 15.58055305480957,
      "learning_rate": 8.968968968968969e-07,
      "loss": 0.1836,
      "step": 90940
    },
    {
      "epoch": 273.12,
      "grad_norm": 10.398634910583496,
      "learning_rate": 8.958958958958958e-07,
      "loss": 0.1573,
      "step": 90950
    },
    {
      "epoch": 273.15,
      "grad_norm": 5.023301601409912,
      "learning_rate": 8.948948948948951e-07,
      "loss": 0.1389,
      "step": 90960
    },
    {
      "epoch": 273.18,
      "grad_norm": 151.21865844726562,
      "learning_rate": 8.93893893893894e-07,
      "loss": 0.1841,
      "step": 90970
    },
    {
      "epoch": 273.21,
      "grad_norm": 12.406375885009766,
      "learning_rate": 8.92892892892893e-07,
      "loss": 0.1859,
      "step": 90980
    },
    {
      "epoch": 273.24,
      "grad_norm": 11.229181289672852,
      "learning_rate": 8.91891891891892e-07,
      "loss": 0.219,
      "step": 90990
    },
    {
      "epoch": 273.27,
      "grad_norm": 14.549325942993164,
      "learning_rate": 8.90890890890891e-07,
      "loss": 0.2088,
      "step": 91000
    },
    {
      "epoch": 273.3,
      "grad_norm": 12.763517379760742,
      "learning_rate": 8.8988988988989e-07,
      "loss": 0.1972,
      "step": 91010
    },
    {
      "epoch": 273.33,
      "grad_norm": 5.383101940155029,
      "learning_rate": 8.88888888888889e-07,
      "loss": 0.1861,
      "step": 91020
    },
    {
      "epoch": 273.36,
      "grad_norm": 8.046422958374023,
      "learning_rate": 8.87887887887888e-07,
      "loss": 0.1961,
      "step": 91030
    },
    {
      "epoch": 273.39,
      "grad_norm": 5.871699333190918,
      "learning_rate": 8.86886886886887e-07,
      "loss": 0.2078,
      "step": 91040
    },
    {
      "epoch": 273.42,
      "grad_norm": 10.00069522857666,
      "learning_rate": 8.858858858858859e-07,
      "loss": 0.2047,
      "step": 91050
    },
    {
      "epoch": 273.45,
      "grad_norm": 4.811551094055176,
      "learning_rate": 8.848848848848849e-07,
      "loss": 0.1301,
      "step": 91060
    },
    {
      "epoch": 273.48,
      "grad_norm": 16.389236450195312,
      "learning_rate": 8.83883883883884e-07,
      "loss": 0.2067,
      "step": 91070
    },
    {
      "epoch": 273.51,
      "grad_norm": 7.010746002197266,
      "learning_rate": 8.82882882882883e-07,
      "loss": 0.1916,
      "step": 91080
    },
    {
      "epoch": 273.54,
      "grad_norm": 8.6441650390625,
      "learning_rate": 8.81881881881882e-07,
      "loss": 0.196,
      "step": 91090
    },
    {
      "epoch": 273.57,
      "grad_norm": 10.989593505859375,
      "learning_rate": 8.808808808808809e-07,
      "loss": 0.19,
      "step": 91100
    },
    {
      "epoch": 273.6,
      "grad_norm": 10.231937408447266,
      "learning_rate": 8.798798798798799e-07,
      "loss": 0.1628,
      "step": 91110
    },
    {
      "epoch": 273.63,
      "grad_norm": 13.978028297424316,
      "learning_rate": 8.78878878878879e-07,
      "loss": 0.1749,
      "step": 91120
    },
    {
      "epoch": 273.66,
      "grad_norm": 8.244202613830566,
      "learning_rate": 8.77877877877878e-07,
      "loss": 0.1657,
      "step": 91130
    },
    {
      "epoch": 273.69,
      "grad_norm": 10.744071006774902,
      "learning_rate": 8.768768768768769e-07,
      "loss": 0.1729,
      "step": 91140
    },
    {
      "epoch": 273.72,
      "grad_norm": 11.86174201965332,
      "learning_rate": 8.758758758758759e-07,
      "loss": 0.2062,
      "step": 91150
    },
    {
      "epoch": 273.75,
      "grad_norm": 9.102922439575195,
      "learning_rate": 8.748748748748749e-07,
      "loss": 0.1896,
      "step": 91160
    },
    {
      "epoch": 273.78,
      "grad_norm": 8.443825721740723,
      "learning_rate": 8.738738738738739e-07,
      "loss": 0.2185,
      "step": 91170
    },
    {
      "epoch": 273.81,
      "grad_norm": 11.443572998046875,
      "learning_rate": 8.728728728728728e-07,
      "loss": 0.1786,
      "step": 91180
    },
    {
      "epoch": 273.84,
      "grad_norm": 8.300676345825195,
      "learning_rate": 8.718718718718719e-07,
      "loss": 0.1609,
      "step": 91190
    },
    {
      "epoch": 273.87,
      "grad_norm": 9.307548522949219,
      "learning_rate": 8.708708708708709e-07,
      "loss": 0.1868,
      "step": 91200
    },
    {
      "epoch": 273.9,
      "grad_norm": 5.828485012054443,
      "learning_rate": 8.698698698698699e-07,
      "loss": 0.1831,
      "step": 91210
    },
    {
      "epoch": 273.93,
      "grad_norm": 11.946285247802734,
      "learning_rate": 8.68868868868869e-07,
      "loss": 0.1729,
      "step": 91220
    },
    {
      "epoch": 273.96,
      "grad_norm": 4.123525619506836,
      "learning_rate": 8.67867867867868e-07,
      "loss": 0.1365,
      "step": 91230
    },
    {
      "epoch": 273.99,
      "grad_norm": 7.676427841186523,
      "learning_rate": 8.66866866866867e-07,
      "loss": 0.2202,
      "step": 91240
    },
    {
      "epoch": 274.0,
      "eval_accuracy": 0.9215,
      "eval_loss": 0.3508787751197815,
      "eval_runtime": 29.9719,
      "eval_samples_per_second": 333.646,
      "eval_steps_per_second": 1.335,
      "step": 91242
    },
    {
      "epoch": 274.02,
      "grad_norm": 10.699176788330078,
      "learning_rate": 8.65865865865866e-07,
      "loss": 0.1574,
      "step": 91250
    },
    {
      "epoch": 274.05,
      "grad_norm": 9.360703468322754,
      "learning_rate": 8.64864864864865e-07,
      "loss": 0.2245,
      "step": 91260
    },
    {
      "epoch": 274.08,
      "grad_norm": 8.03820514678955,
      "learning_rate": 8.63863863863864e-07,
      "loss": 0.2055,
      "step": 91270
    },
    {
      "epoch": 274.11,
      "grad_norm": 4.4602179527282715,
      "learning_rate": 8.628628628628629e-07,
      "loss": 0.1536,
      "step": 91280
    },
    {
      "epoch": 274.14,
      "grad_norm": 6.988121509552002,
      "learning_rate": 8.618618618618619e-07,
      "loss": 0.1659,
      "step": 91290
    },
    {
      "epoch": 274.17,
      "grad_norm": 11.688573837280273,
      "learning_rate": 8.60860860860861e-07,
      "loss": 0.1898,
      "step": 91300
    },
    {
      "epoch": 274.2,
      "grad_norm": 7.689059257507324,
      "learning_rate": 8.5985985985986e-07,
      "loss": 0.1585,
      "step": 91310
    },
    {
      "epoch": 274.23,
      "grad_norm": 11.546932220458984,
      "learning_rate": 8.58858858858859e-07,
      "loss": 0.2164,
      "step": 91320
    },
    {
      "epoch": 274.26,
      "grad_norm": 8.809536933898926,
      "learning_rate": 8.578578578578579e-07,
      "loss": 0.1568,
      "step": 91330
    },
    {
      "epoch": 274.29,
      "grad_norm": 13.755922317504883,
      "learning_rate": 8.568568568568569e-07,
      "loss": 0.145,
      "step": 91340
    },
    {
      "epoch": 274.32,
      "grad_norm": 8.537341117858887,
      "learning_rate": 8.55855855855856e-07,
      "loss": 0.2298,
      "step": 91350
    },
    {
      "epoch": 274.35,
      "grad_norm": 5.964023113250732,
      "learning_rate": 8.54854854854855e-07,
      "loss": 0.1784,
      "step": 91360
    },
    {
      "epoch": 274.38,
      "grad_norm": 8.537186622619629,
      "learning_rate": 8.538538538538539e-07,
      "loss": 0.1726,
      "step": 91370
    },
    {
      "epoch": 274.41,
      "grad_norm": 24.956275939941406,
      "learning_rate": 8.528528528528529e-07,
      "loss": 0.185,
      "step": 91380
    },
    {
      "epoch": 274.44,
      "grad_norm": 12.16773509979248,
      "learning_rate": 8.518518518518519e-07,
      "loss": 0.1792,
      "step": 91390
    },
    {
      "epoch": 274.47,
      "grad_norm": 7.295405864715576,
      "learning_rate": 8.508508508508509e-07,
      "loss": 0.2237,
      "step": 91400
    },
    {
      "epoch": 274.5,
      "grad_norm": 11.372282028198242,
      "learning_rate": 8.498498498498498e-07,
      "loss": 0.1835,
      "step": 91410
    },
    {
      "epoch": 274.53,
      "grad_norm": 7.533218860626221,
      "learning_rate": 8.488488488488489e-07,
      "loss": 0.1986,
      "step": 91420
    },
    {
      "epoch": 274.56,
      "grad_norm": 6.168046474456787,
      "learning_rate": 8.478478478478479e-07,
      "loss": 0.1469,
      "step": 91430
    },
    {
      "epoch": 274.59,
      "grad_norm": 17.495323181152344,
      "learning_rate": 8.468468468468469e-07,
      "loss": 0.1955,
      "step": 91440
    },
    {
      "epoch": 274.62,
      "grad_norm": 8.688248634338379,
      "learning_rate": 8.458458458458458e-07,
      "loss": 0.1768,
      "step": 91450
    },
    {
      "epoch": 274.65,
      "grad_norm": 5.905192852020264,
      "learning_rate": 8.448448448448448e-07,
      "loss": 0.1768,
      "step": 91460
    },
    {
      "epoch": 274.68,
      "grad_norm": 6.16829776763916,
      "learning_rate": 8.438438438438439e-07,
      "loss": 0.2112,
      "step": 91470
    },
    {
      "epoch": 274.71,
      "grad_norm": 10.61315631866455,
      "learning_rate": 8.428428428428429e-07,
      "loss": 0.2401,
      "step": 91480
    },
    {
      "epoch": 274.74,
      "grad_norm": 7.824102878570557,
      "learning_rate": 8.41841841841842e-07,
      "loss": 0.2262,
      "step": 91490
    },
    {
      "epoch": 274.77,
      "grad_norm": 7.085625648498535,
      "learning_rate": 8.40840840840841e-07,
      "loss": 0.1537,
      "step": 91500
    },
    {
      "epoch": 274.8,
      "grad_norm": 8.457878112792969,
      "learning_rate": 8.398398398398399e-07,
      "loss": 0.1662,
      "step": 91510
    },
    {
      "epoch": 274.83,
      "grad_norm": 8.10151195526123,
      "learning_rate": 8.388388388388389e-07,
      "loss": 0.1617,
      "step": 91520
    },
    {
      "epoch": 274.86,
      "grad_norm": 8.72238540649414,
      "learning_rate": 8.37837837837838e-07,
      "loss": 0.189,
      "step": 91530
    },
    {
      "epoch": 274.89,
      "grad_norm": 7.4450531005859375,
      "learning_rate": 8.36836836836837e-07,
      "loss": 0.202,
      "step": 91540
    },
    {
      "epoch": 274.92,
      "grad_norm": 9.713227272033691,
      "learning_rate": 8.358358358358359e-07,
      "loss": 0.1824,
      "step": 91550
    },
    {
      "epoch": 274.95,
      "grad_norm": 10.386857986450195,
      "learning_rate": 8.348348348348349e-07,
      "loss": 0.1854,
      "step": 91560
    },
    {
      "epoch": 274.98,
      "grad_norm": 6.122295379638672,
      "learning_rate": 8.338338338338339e-07,
      "loss": 0.1804,
      "step": 91570
    },
    {
      "epoch": 275.0,
      "eval_accuracy": 0.9227,
      "eval_loss": 0.35024166107177734,
      "eval_runtime": 30.391,
      "eval_samples_per_second": 329.045,
      "eval_steps_per_second": 1.316,
      "step": 91575
    },
    {
      "epoch": 275.02,
      "grad_norm": 6.1142778396606445,
      "learning_rate": 8.328328328328329e-07,
      "loss": 0.1356,
      "step": 91580
    },
    {
      "epoch": 275.05,
      "grad_norm": 7.438205718994141,
      "learning_rate": 8.31831831831832e-07,
      "loss": 0.1602,
      "step": 91590
    },
    {
      "epoch": 275.08,
      "grad_norm": 15.214101791381836,
      "learning_rate": 8.308308308308309e-07,
      "loss": 0.2074,
      "step": 91600
    },
    {
      "epoch": 275.11,
      "grad_norm": 11.082049369812012,
      "learning_rate": 8.298298298298299e-07,
      "loss": 0.1924,
      "step": 91610
    },
    {
      "epoch": 275.14,
      "grad_norm": 9.185382843017578,
      "learning_rate": 8.288288288288289e-07,
      "loss": 0.2029,
      "step": 91620
    },
    {
      "epoch": 275.17,
      "grad_norm": 8.849352836608887,
      "learning_rate": 8.278278278278279e-07,
      "loss": 0.1882,
      "step": 91630
    },
    {
      "epoch": 275.2,
      "grad_norm": 6.922469615936279,
      "learning_rate": 8.268268268268268e-07,
      "loss": 0.1787,
      "step": 91640
    },
    {
      "epoch": 275.23,
      "grad_norm": 8.060291290283203,
      "learning_rate": 8.258258258258259e-07,
      "loss": 0.1919,
      "step": 91650
    },
    {
      "epoch": 275.26,
      "grad_norm": 14.404717445373535,
      "learning_rate": 8.248248248248249e-07,
      "loss": 0.1755,
      "step": 91660
    },
    {
      "epoch": 275.29,
      "grad_norm": 8.0862455368042,
      "learning_rate": 8.238238238238239e-07,
      "loss": 0.1788,
      "step": 91670
    },
    {
      "epoch": 275.32,
      "grad_norm": 15.028704643249512,
      "learning_rate": 8.228228228228228e-07,
      "loss": 0.1969,
      "step": 91680
    },
    {
      "epoch": 275.35,
      "grad_norm": 8.093008995056152,
      "learning_rate": 8.218218218218218e-07,
      "loss": 0.1936,
      "step": 91690
    },
    {
      "epoch": 275.38,
      "grad_norm": 22.866722106933594,
      "learning_rate": 8.208208208208208e-07,
      "loss": 0.2167,
      "step": 91700
    },
    {
      "epoch": 275.41,
      "grad_norm": 5.900140285491943,
      "learning_rate": 8.198198198198199e-07,
      "loss": 0.209,
      "step": 91710
    },
    {
      "epoch": 275.44,
      "grad_norm": 18.971721649169922,
      "learning_rate": 8.188188188188189e-07,
      "loss": 0.1918,
      "step": 91720
    },
    {
      "epoch": 275.47,
      "grad_norm": 14.155665397644043,
      "learning_rate": 8.178178178178178e-07,
      "loss": 0.2569,
      "step": 91730
    },
    {
      "epoch": 275.5,
      "grad_norm": 20.45295524597168,
      "learning_rate": 8.168168168168168e-07,
      "loss": 0.1556,
      "step": 91740
    },
    {
      "epoch": 275.53,
      "grad_norm": 6.0278000831604,
      "learning_rate": 8.158158158158158e-07,
      "loss": 0.1945,
      "step": 91750
    },
    {
      "epoch": 275.56,
      "grad_norm": 12.325450897216797,
      "learning_rate": 8.14814814814815e-07,
      "loss": 0.192,
      "step": 91760
    },
    {
      "epoch": 275.59,
      "grad_norm": 7.747560977935791,
      "learning_rate": 8.13813813813814e-07,
      "loss": 0.1709,
      "step": 91770
    },
    {
      "epoch": 275.62,
      "grad_norm": 8.097228050231934,
      "learning_rate": 8.128128128128129e-07,
      "loss": 0.199,
      "step": 91780
    },
    {
      "epoch": 275.65,
      "grad_norm": 7.821126937866211,
      "learning_rate": 8.118118118118119e-07,
      "loss": 0.1949,
      "step": 91790
    },
    {
      "epoch": 275.68,
      "grad_norm": 8.0576753616333,
      "learning_rate": 8.108108108108109e-07,
      "loss": 0.1825,
      "step": 91800
    },
    {
      "epoch": 275.71,
      "grad_norm": 6.3059821128845215,
      "learning_rate": 8.098098098098099e-07,
      "loss": 0.1836,
      "step": 91810
    },
    {
      "epoch": 275.74,
      "grad_norm": 6.190617561340332,
      "learning_rate": 8.08808808808809e-07,
      "loss": 0.2172,
      "step": 91820
    },
    {
      "epoch": 275.77,
      "grad_norm": 8.426007270812988,
      "learning_rate": 8.078078078078079e-07,
      "loss": 0.1852,
      "step": 91830
    },
    {
      "epoch": 275.8,
      "grad_norm": 12.321769714355469,
      "learning_rate": 8.068068068068069e-07,
      "loss": 0.2032,
      "step": 91840
    },
    {
      "epoch": 275.83,
      "grad_norm": 5.943408012390137,
      "learning_rate": 8.058058058058059e-07,
      "loss": 0.1431,
      "step": 91850
    },
    {
      "epoch": 275.86,
      "grad_norm": 9.514937400817871,
      "learning_rate": 8.048048048048049e-07,
      "loss": 0.2132,
      "step": 91860
    },
    {
      "epoch": 275.89,
      "grad_norm": 5.887760162353516,
      "learning_rate": 8.038038038038038e-07,
      "loss": 0.1493,
      "step": 91870
    },
    {
      "epoch": 275.92,
      "grad_norm": 5.572448253631592,
      "learning_rate": 8.028028028028029e-07,
      "loss": 0.1918,
      "step": 91880
    },
    {
      "epoch": 275.95,
      "grad_norm": 11.756245613098145,
      "learning_rate": 8.018018018018019e-07,
      "loss": 0.2197,
      "step": 91890
    },
    {
      "epoch": 275.98,
      "grad_norm": 7.37849760055542,
      "learning_rate": 8.008008008008009e-07,
      "loss": 0.1542,
      "step": 91900
    },
    {
      "epoch": 276.0,
      "eval_accuracy": 0.9229,
      "eval_loss": 0.34959039092063904,
      "eval_runtime": 30.3616,
      "eval_samples_per_second": 329.364,
      "eval_steps_per_second": 1.317,
      "step": 91908
    },
    {
      "epoch": 276.01,
      "grad_norm": 7.547004699707031,
      "learning_rate": 7.997997997997998e-07,
      "loss": 0.1926,
      "step": 91910
    },
    {
      "epoch": 276.04,
      "grad_norm": 11.517204284667969,
      "learning_rate": 7.987987987987988e-07,
      "loss": 0.2201,
      "step": 91920
    },
    {
      "epoch": 276.07,
      "grad_norm": 11.912858009338379,
      "learning_rate": 7.977977977977978e-07,
      "loss": 0.1934,
      "step": 91930
    },
    {
      "epoch": 276.1,
      "grad_norm": 8.050774574279785,
      "learning_rate": 7.967967967967969e-07,
      "loss": 0.1973,
      "step": 91940
    },
    {
      "epoch": 276.13,
      "grad_norm": 25.22945213317871,
      "learning_rate": 7.957957957957958e-07,
      "loss": 0.1952,
      "step": 91950
    },
    {
      "epoch": 276.16,
      "grad_norm": 4.601602077484131,
      "learning_rate": 7.947947947947948e-07,
      "loss": 0.1989,
      "step": 91960
    },
    {
      "epoch": 276.19,
      "grad_norm": 8.468716621398926,
      "learning_rate": 7.937937937937938e-07,
      "loss": 0.2193,
      "step": 91970
    },
    {
      "epoch": 276.22,
      "grad_norm": 9.864021301269531,
      "learning_rate": 7.927927927927928e-07,
      "loss": 0.1877,
      "step": 91980
    },
    {
      "epoch": 276.25,
      "grad_norm": 9.509844779968262,
      "learning_rate": 7.917917917917919e-07,
      "loss": 0.1921,
      "step": 91990
    },
    {
      "epoch": 276.28,
      "grad_norm": 7.579611301422119,
      "learning_rate": 7.907907907907908e-07,
      "loss": 0.2097,
      "step": 92000
    },
    {
      "epoch": 276.31,
      "grad_norm": 17.913307189941406,
      "learning_rate": 7.897897897897898e-07,
      "loss": 0.1828,
      "step": 92010
    },
    {
      "epoch": 276.34,
      "grad_norm": 12.086991310119629,
      "learning_rate": 7.887887887887888e-07,
      "loss": 0.1818,
      "step": 92020
    },
    {
      "epoch": 276.37,
      "grad_norm": 7.509232997894287,
      "learning_rate": 7.877877877877879e-07,
      "loss": 0.1919,
      "step": 92030
    },
    {
      "epoch": 276.4,
      "grad_norm": 9.362010955810547,
      "learning_rate": 7.867867867867869e-07,
      "loss": 0.1513,
      "step": 92040
    },
    {
      "epoch": 276.43,
      "grad_norm": 6.234588146209717,
      "learning_rate": 7.857857857857859e-07,
      "loss": 0.1929,
      "step": 92050
    },
    {
      "epoch": 276.46,
      "grad_norm": 12.175146102905273,
      "learning_rate": 7.847847847847849e-07,
      "loss": 0.1773,
      "step": 92060
    },
    {
      "epoch": 276.49,
      "grad_norm": 5.041324138641357,
      "learning_rate": 7.837837837837839e-07,
      "loss": 0.1976,
      "step": 92070
    },
    {
      "epoch": 276.52,
      "grad_norm": 9.547052383422852,
      "learning_rate": 7.827827827827829e-07,
      "loss": 0.1622,
      "step": 92080
    },
    {
      "epoch": 276.55,
      "grad_norm": 10.639104843139648,
      "learning_rate": 7.817817817817819e-07,
      "loss": 0.2124,
      "step": 92090
    },
    {
      "epoch": 276.58,
      "grad_norm": 7.385708332061768,
      "learning_rate": 7.807807807807808e-07,
      "loss": 0.1882,
      "step": 92100
    },
    {
      "epoch": 276.61,
      "grad_norm": 3.707102060317993,
      "learning_rate": 7.797797797797799e-07,
      "loss": 0.2006,
      "step": 92110
    },
    {
      "epoch": 276.64,
      "grad_norm": 8.716361045837402,
      "learning_rate": 7.787787787787789e-07,
      "loss": 0.1797,
      "step": 92120
    },
    {
      "epoch": 276.67,
      "grad_norm": 8.952524185180664,
      "learning_rate": 7.777777777777779e-07,
      "loss": 0.2044,
      "step": 92130
    },
    {
      "epoch": 276.7,
      "grad_norm": 4.383061408996582,
      "learning_rate": 7.767767767767768e-07,
      "loss": 0.2114,
      "step": 92140
    },
    {
      "epoch": 276.73,
      "grad_norm": 9.117514610290527,
      "learning_rate": 7.757757757757758e-07,
      "loss": 0.1663,
      "step": 92150
    },
    {
      "epoch": 276.76,
      "grad_norm": 8.893731117248535,
      "learning_rate": 7.747747747747748e-07,
      "loss": 0.1956,
      "step": 92160
    },
    {
      "epoch": 276.79,
      "grad_norm": 13.22093677520752,
      "learning_rate": 7.737737737737739e-07,
      "loss": 0.2072,
      "step": 92170
    },
    {
      "epoch": 276.82,
      "grad_norm": 8.830599784851074,
      "learning_rate": 7.727727727727728e-07,
      "loss": 0.2132,
      "step": 92180
    },
    {
      "epoch": 276.85,
      "grad_norm": 9.125177383422852,
      "learning_rate": 7.717717717717718e-07,
      "loss": 0.2034,
      "step": 92190
    },
    {
      "epoch": 276.88,
      "grad_norm": 9.151504516601562,
      "learning_rate": 7.707707707707708e-07,
      "loss": 0.1686,
      "step": 92200
    },
    {
      "epoch": 276.91,
      "grad_norm": 10.556206703186035,
      "learning_rate": 7.697697697697698e-07,
      "loss": 0.1797,
      "step": 92210
    },
    {
      "epoch": 276.94,
      "grad_norm": 6.698130130767822,
      "learning_rate": 7.687687687687688e-07,
      "loss": 0.1435,
      "step": 92220
    },
    {
      "epoch": 276.97,
      "grad_norm": 10.389276504516602,
      "learning_rate": 7.677677677677678e-07,
      "loss": 0.1807,
      "step": 92230
    },
    {
      "epoch": 277.0,
      "grad_norm": 6.247650146484375,
      "learning_rate": 7.667667667667668e-07,
      "loss": 0.1744,
      "step": 92240
    },
    {
      "epoch": 277.0,
      "eval_accuracy": 0.9226,
      "eval_loss": 0.348612904548645,
      "eval_runtime": 30.0489,
      "eval_samples_per_second": 332.79,
      "eval_steps_per_second": 1.331,
      "step": 92241
    },
    {
      "epoch": 277.03,
      "grad_norm": 6.425793170928955,
      "learning_rate": 7.657657657657658e-07,
      "loss": 0.1815,
      "step": 92250
    },
    {
      "epoch": 277.06,
      "grad_norm": 7.438033103942871,
      "learning_rate": 7.647647647647648e-07,
      "loss": 0.1825,
      "step": 92260
    },
    {
      "epoch": 277.09,
      "grad_norm": 9.516624450683594,
      "learning_rate": 7.637637637637637e-07,
      "loss": 0.1956,
      "step": 92270
    },
    {
      "epoch": 277.12,
      "grad_norm": 6.098445892333984,
      "learning_rate": 7.627627627627627e-07,
      "loss": 0.1366,
      "step": 92280
    },
    {
      "epoch": 277.15,
      "grad_norm": 5.587775707244873,
      "learning_rate": 7.617617617617619e-07,
      "loss": 0.1929,
      "step": 92290
    },
    {
      "epoch": 277.18,
      "grad_norm": 9.475522994995117,
      "learning_rate": 7.607607607607609e-07,
      "loss": 0.1568,
      "step": 92300
    },
    {
      "epoch": 277.21,
      "grad_norm": 10.771020889282227,
      "learning_rate": 7.597597597597599e-07,
      "loss": 0.1832,
      "step": 92310
    },
    {
      "epoch": 277.24,
      "grad_norm": 7.350369453430176,
      "learning_rate": 7.587587587587588e-07,
      "loss": 0.1798,
      "step": 92320
    },
    {
      "epoch": 277.27,
      "grad_norm": 10.485979080200195,
      "learning_rate": 7.577577577577578e-07,
      "loss": 0.1819,
      "step": 92330
    },
    {
      "epoch": 277.3,
      "grad_norm": 8.053985595703125,
      "learning_rate": 7.567567567567569e-07,
      "loss": 0.1655,
      "step": 92340
    },
    {
      "epoch": 277.33,
      "grad_norm": 9.377668380737305,
      "learning_rate": 7.557557557557559e-07,
      "loss": 0.2254,
      "step": 92350
    },
    {
      "epoch": 277.36,
      "grad_norm": 7.820313453674316,
      "learning_rate": 7.547547547547549e-07,
      "loss": 0.2146,
      "step": 92360
    },
    {
      "epoch": 277.39,
      "grad_norm": 9.382530212402344,
      "learning_rate": 7.537537537537538e-07,
      "loss": 0.1758,
      "step": 92370
    },
    {
      "epoch": 277.42,
      "grad_norm": 6.953212261199951,
      "learning_rate": 7.527527527527528e-07,
      "loss": 0.2111,
      "step": 92380
    },
    {
      "epoch": 277.45,
      "grad_norm": 8.748991966247559,
      "learning_rate": 7.517517517517518e-07,
      "loss": 0.1876,
      "step": 92390
    },
    {
      "epoch": 277.48,
      "grad_norm": 3.882725477218628,
      "learning_rate": 7.507507507507509e-07,
      "loss": 0.143,
      "step": 92400
    },
    {
      "epoch": 277.51,
      "grad_norm": 7.822537422180176,
      "learning_rate": 7.497497497497498e-07,
      "loss": 0.1896,
      "step": 92410
    },
    {
      "epoch": 277.54,
      "grad_norm": 13.135604858398438,
      "learning_rate": 7.487487487487488e-07,
      "loss": 0.1969,
      "step": 92420
    },
    {
      "epoch": 277.57,
      "grad_norm": 10.035101890563965,
      "learning_rate": 7.477477477477478e-07,
      "loss": 0.1625,
      "step": 92430
    },
    {
      "epoch": 277.6,
      "grad_norm": 5.777451515197754,
      "learning_rate": 7.467467467467468e-07,
      "loss": 0.1714,
      "step": 92440
    },
    {
      "epoch": 277.63,
      "grad_norm": 12.348820686340332,
      "learning_rate": 7.457457457457457e-07,
      "loss": 0.194,
      "step": 92450
    },
    {
      "epoch": 277.66,
      "grad_norm": 5.185982704162598,
      "learning_rate": 7.447447447447448e-07,
      "loss": 0.1848,
      "step": 92460
    },
    {
      "epoch": 277.69,
      "grad_norm": 9.268671035766602,
      "learning_rate": 7.437437437437438e-07,
      "loss": 0.1934,
      "step": 92470
    },
    {
      "epoch": 277.72,
      "grad_norm": 8.485932350158691,
      "learning_rate": 7.427427427427428e-07,
      "loss": 0.2038,
      "step": 92480
    },
    {
      "epoch": 277.75,
      "grad_norm": 14.882165908813477,
      "learning_rate": 7.417417417417418e-07,
      "loss": 0.2114,
      "step": 92490
    },
    {
      "epoch": 277.78,
      "grad_norm": 16.157241821289062,
      "learning_rate": 7.407407407407407e-07,
      "loss": 0.1449,
      "step": 92500
    },
    {
      "epoch": 277.81,
      "grad_norm": 11.725473403930664,
      "learning_rate": 7.397397397397397e-07,
      "loss": 0.2049,
      "step": 92510
    },
    {
      "epoch": 277.84,
      "grad_norm": 13.23259162902832,
      "learning_rate": 7.387387387387388e-07,
      "loss": 0.1462,
      "step": 92520
    },
    {
      "epoch": 277.87,
      "grad_norm": 22.680757522583008,
      "learning_rate": 7.377377377377378e-07,
      "loss": 0.1723,
      "step": 92530
    },
    {
      "epoch": 277.9,
      "grad_norm": 5.945184230804443,
      "learning_rate": 7.367367367367367e-07,
      "loss": 0.1709,
      "step": 92540
    },
    {
      "epoch": 277.93,
      "grad_norm": 15.038105964660645,
      "learning_rate": 7.357357357357357e-07,
      "loss": 0.2039,
      "step": 92550
    },
    {
      "epoch": 277.96,
      "grad_norm": 5.159658908843994,
      "learning_rate": 7.347347347347348e-07,
      "loss": 0.1608,
      "step": 92560
    },
    {
      "epoch": 277.99,
      "grad_norm": 5.669399261474609,
      "learning_rate": 7.337337337337339e-07,
      "loss": 0.1779,
      "step": 92570
    },
    {
      "epoch": 278.0,
      "eval_accuracy": 0.9228,
      "eval_loss": 0.3483006954193115,
      "eval_runtime": 30.0533,
      "eval_samples_per_second": 332.742,
      "eval_steps_per_second": 1.331,
      "step": 92574
    },
    {
      "epoch": 278.02,
      "grad_norm": 12.969282150268555,
      "learning_rate": 7.327327327327329e-07,
      "loss": 0.188,
      "step": 92580
    },
    {
      "epoch": 278.05,
      "grad_norm": 5.987633228302002,
      "learning_rate": 7.317317317317319e-07,
      "loss": 0.1627,
      "step": 92590
    },
    {
      "epoch": 278.08,
      "grad_norm": 3.1498403549194336,
      "learning_rate": 7.307307307307308e-07,
      "loss": 0.1756,
      "step": 92600
    },
    {
      "epoch": 278.11,
      "grad_norm": 5.837270259857178,
      "learning_rate": 7.297297297297298e-07,
      "loss": 0.1641,
      "step": 92610
    },
    {
      "epoch": 278.14,
      "grad_norm": 12.201165199279785,
      "learning_rate": 7.287287287287288e-07,
      "loss": 0.1693,
      "step": 92620
    },
    {
      "epoch": 278.17,
      "grad_norm": 6.028960704803467,
      "learning_rate": 7.277277277277279e-07,
      "loss": 0.1614,
      "step": 92630
    },
    {
      "epoch": 278.2,
      "grad_norm": 8.963961601257324,
      "learning_rate": 7.267267267267268e-07,
      "loss": 0.2005,
      "step": 92640
    },
    {
      "epoch": 278.23,
      "grad_norm": 11.516244888305664,
      "learning_rate": 7.257257257257258e-07,
      "loss": 0.1931,
      "step": 92650
    },
    {
      "epoch": 278.26,
      "grad_norm": 7.685690879821777,
      "learning_rate": 7.247247247247248e-07,
      "loss": 0.1582,
      "step": 92660
    },
    {
      "epoch": 278.29,
      "grad_norm": 15.3729248046875,
      "learning_rate": 7.237237237237238e-07,
      "loss": 0.2409,
      "step": 92670
    },
    {
      "epoch": 278.32,
      "grad_norm": 8.150314331054688,
      "learning_rate": 7.227227227227227e-07,
      "loss": 0.2008,
      "step": 92680
    },
    {
      "epoch": 278.35,
      "grad_norm": 13.204696655273438,
      "learning_rate": 7.217217217217218e-07,
      "loss": 0.174,
      "step": 92690
    },
    {
      "epoch": 278.38,
      "grad_norm": 5.518671035766602,
      "learning_rate": 7.207207207207208e-07,
      "loss": 0.1964,
      "step": 92700
    },
    {
      "epoch": 278.41,
      "grad_norm": 12.500702857971191,
      "learning_rate": 7.197197197197198e-07,
      "loss": 0.1616,
      "step": 92710
    },
    {
      "epoch": 278.44,
      "grad_norm": 17.583572387695312,
      "learning_rate": 7.187187187187188e-07,
      "loss": 0.1817,
      "step": 92720
    },
    {
      "epoch": 278.47,
      "grad_norm": 7.954564094543457,
      "learning_rate": 7.177177177177177e-07,
      "loss": 0.1796,
      "step": 92730
    },
    {
      "epoch": 278.5,
      "grad_norm": 10.01884937286377,
      "learning_rate": 7.167167167167167e-07,
      "loss": 0.1974,
      "step": 92740
    },
    {
      "epoch": 278.53,
      "grad_norm": 9.666257858276367,
      "learning_rate": 7.157157157157158e-07,
      "loss": 0.1867,
      "step": 92750
    },
    {
      "epoch": 278.56,
      "grad_norm": 9.096963882446289,
      "learning_rate": 7.147147147147148e-07,
      "loss": 0.1587,
      "step": 92760
    },
    {
      "epoch": 278.59,
      "grad_norm": 9.99958610534668,
      "learning_rate": 7.137137137137137e-07,
      "loss": 0.182,
      "step": 92770
    },
    {
      "epoch": 278.62,
      "grad_norm": 8.918646812438965,
      "learning_rate": 7.127127127127127e-07,
      "loss": 0.1884,
      "step": 92780
    },
    {
      "epoch": 278.65,
      "grad_norm": 7.1500444412231445,
      "learning_rate": 7.117117117117117e-07,
      "loss": 0.1887,
      "step": 92790
    },
    {
      "epoch": 278.68,
      "grad_norm": 9.665763854980469,
      "learning_rate": 7.107107107107107e-07,
      "loss": 0.206,
      "step": 92800
    },
    {
      "epoch": 278.71,
      "grad_norm": 13.490777015686035,
      "learning_rate": 7.097097097097097e-07,
      "loss": 0.1669,
      "step": 92810
    },
    {
      "epoch": 278.74,
      "grad_norm": 19.546531677246094,
      "learning_rate": 7.087087087087087e-07,
      "loss": 0.1964,
      "step": 92820
    },
    {
      "epoch": 278.77,
      "grad_norm": 9.986693382263184,
      "learning_rate": 7.077077077077078e-07,
      "loss": 0.2045,
      "step": 92830
    },
    {
      "epoch": 278.8,
      "grad_norm": 12.731371879577637,
      "learning_rate": 7.067067067067068e-07,
      "loss": 0.2225,
      "step": 92840
    },
    {
      "epoch": 278.83,
      "grad_norm": 7.088918685913086,
      "learning_rate": 7.057057057057058e-07,
      "loss": 0.1888,
      "step": 92850
    },
    {
      "epoch": 278.86,
      "grad_norm": 7.145175457000732,
      "learning_rate": 7.047047047047049e-07,
      "loss": 0.1639,
      "step": 92860
    },
    {
      "epoch": 278.89,
      "grad_norm": 8.948333740234375,
      "learning_rate": 7.037037037037038e-07,
      "loss": 0.2081,
      "step": 92870
    },
    {
      "epoch": 278.92,
      "grad_norm": 12.479610443115234,
      "learning_rate": 7.027027027027028e-07,
      "loss": 0.202,
      "step": 92880
    },
    {
      "epoch": 278.95,
      "grad_norm": 7.667999744415283,
      "learning_rate": 7.017017017017018e-07,
      "loss": 0.1749,
      "step": 92890
    },
    {
      "epoch": 278.98,
      "grad_norm": 7.06297492980957,
      "learning_rate": 7.007007007007008e-07,
      "loss": 0.1396,
      "step": 92900
    },
    {
      "epoch": 279.0,
      "eval_accuracy": 0.9228,
      "eval_loss": 0.3494606018066406,
      "eval_runtime": 30.3878,
      "eval_samples_per_second": 329.08,
      "eval_steps_per_second": 1.316,
      "step": 92907
    },
    {
      "epoch": 279.01,
      "grad_norm": 9.287768363952637,
      "learning_rate": 6.996996996996997e-07,
      "loss": 0.1819,
      "step": 92910
    },
    {
      "epoch": 279.04,
      "grad_norm": 9.37019157409668,
      "learning_rate": 6.986986986986988e-07,
      "loss": 0.1767,
      "step": 92920
    },
    {
      "epoch": 279.07,
      "grad_norm": 6.519312858581543,
      "learning_rate": 6.976976976976978e-07,
      "loss": 0.184,
      "step": 92930
    },
    {
      "epoch": 279.1,
      "grad_norm": 10.531512260437012,
      "learning_rate": 6.966966966966968e-07,
      "loss": 0.2,
      "step": 92940
    },
    {
      "epoch": 279.13,
      "grad_norm": 29.19515037536621,
      "learning_rate": 6.956956956956957e-07,
      "loss": 0.2387,
      "step": 92950
    },
    {
      "epoch": 279.16,
      "grad_norm": 4.296021938323975,
      "learning_rate": 6.946946946946947e-07,
      "loss": 0.1905,
      "step": 92960
    },
    {
      "epoch": 279.19,
      "grad_norm": 8.085820198059082,
      "learning_rate": 6.936936936936937e-07,
      "loss": 0.1984,
      "step": 92970
    },
    {
      "epoch": 279.22,
      "grad_norm": 12.278340339660645,
      "learning_rate": 6.926926926926928e-07,
      "loss": 0.1816,
      "step": 92980
    },
    {
      "epoch": 279.25,
      "grad_norm": 7.307237148284912,
      "learning_rate": 6.916916916916918e-07,
      "loss": 0.1581,
      "step": 92990
    },
    {
      "epoch": 279.28,
      "grad_norm": 8.668619155883789,
      "learning_rate": 6.906906906906907e-07,
      "loss": 0.1727,
      "step": 93000
    },
    {
      "epoch": 279.31,
      "grad_norm": 9.67676067352295,
      "learning_rate": 6.896896896896897e-07,
      "loss": 0.2041,
      "step": 93010
    },
    {
      "epoch": 279.34,
      "grad_norm": 9.745436668395996,
      "learning_rate": 6.886886886886887e-07,
      "loss": 0.2123,
      "step": 93020
    },
    {
      "epoch": 279.37,
      "grad_norm": 8.133345603942871,
      "learning_rate": 6.876876876876877e-07,
      "loss": 0.2078,
      "step": 93030
    },
    {
      "epoch": 279.4,
      "grad_norm": 12.13882827758789,
      "learning_rate": 6.866866866866867e-07,
      "loss": 0.1812,
      "step": 93040
    },
    {
      "epoch": 279.43,
      "grad_norm": 4.484041690826416,
      "learning_rate": 6.856856856856857e-07,
      "loss": 0.2043,
      "step": 93050
    },
    {
      "epoch": 279.46,
      "grad_norm": 10.569123268127441,
      "learning_rate": 6.846846846846847e-07,
      "loss": 0.1917,
      "step": 93060
    },
    {
      "epoch": 279.49,
      "grad_norm": 8.036044120788574,
      "learning_rate": 6.836836836836837e-07,
      "loss": 0.1913,
      "step": 93070
    },
    {
      "epoch": 279.52,
      "grad_norm": 11.973335266113281,
      "learning_rate": 6.826826826826826e-07,
      "loss": 0.1886,
      "step": 93080
    },
    {
      "epoch": 279.55,
      "grad_norm": 15.80704116821289,
      "learning_rate": 6.816816816816819e-07,
      "loss": 0.2008,
      "step": 93090
    },
    {
      "epoch": 279.58,
      "grad_norm": 7.810923099517822,
      "learning_rate": 6.806806806806808e-07,
      "loss": 0.1687,
      "step": 93100
    },
    {
      "epoch": 279.61,
      "grad_norm": 9.648831367492676,
      "learning_rate": 6.796796796796798e-07,
      "loss": 0.1978,
      "step": 93110
    },
    {
      "epoch": 279.64,
      "grad_norm": 8.244765281677246,
      "learning_rate": 6.786786786786788e-07,
      "loss": 0.2036,
      "step": 93120
    },
    {
      "epoch": 279.67,
      "grad_norm": 5.94614315032959,
      "learning_rate": 6.776776776776778e-07,
      "loss": 0.1724,
      "step": 93130
    },
    {
      "epoch": 279.7,
      "grad_norm": 11.583590507507324,
      "learning_rate": 6.766766766766767e-07,
      "loss": 0.1608,
      "step": 93140
    },
    {
      "epoch": 279.73,
      "grad_norm": 13.816680908203125,
      "learning_rate": 6.756756756756758e-07,
      "loss": 0.1655,
      "step": 93150
    },
    {
      "epoch": 279.76,
      "grad_norm": 7.566768169403076,
      "learning_rate": 6.746746746746748e-07,
      "loss": 0.1778,
      "step": 93160
    },
    {
      "epoch": 279.79,
      "grad_norm": 5.473132610321045,
      "learning_rate": 6.736736736736738e-07,
      "loss": 0.1655,
      "step": 93170
    },
    {
      "epoch": 279.82,
      "grad_norm": 9.53297233581543,
      "learning_rate": 6.726726726726727e-07,
      "loss": 0.1891,
      "step": 93180
    },
    {
      "epoch": 279.85,
      "grad_norm": 6.92855978012085,
      "learning_rate": 6.716716716716717e-07,
      "loss": 0.1738,
      "step": 93190
    },
    {
      "epoch": 279.88,
      "grad_norm": 31.804079055786133,
      "learning_rate": 6.706706706706707e-07,
      "loss": 0.1787,
      "step": 93200
    },
    {
      "epoch": 279.91,
      "grad_norm": 7.807969570159912,
      "learning_rate": 6.696696696696698e-07,
      "loss": 0.1554,
      "step": 93210
    },
    {
      "epoch": 279.94,
      "grad_norm": 12.320775985717773,
      "learning_rate": 6.686686686686688e-07,
      "loss": 0.1747,
      "step": 93220
    },
    {
      "epoch": 279.97,
      "grad_norm": 6.537836074829102,
      "learning_rate": 6.676676676676677e-07,
      "loss": 0.1852,
      "step": 93230
    },
    {
      "epoch": 280.0,
      "grad_norm": 32.48252868652344,
      "learning_rate": 6.666666666666667e-07,
      "loss": 0.1501,
      "step": 93240
    },
    {
      "epoch": 280.0,
      "eval_accuracy": 0.9232,
      "eval_loss": 0.34844404458999634,
      "eval_runtime": 30.3608,
      "eval_samples_per_second": 329.372,
      "eval_steps_per_second": 1.317,
      "step": 93240
    },
    {
      "epoch": 280.03,
      "grad_norm": 12.149720191955566,
      "learning_rate": 6.656656656656657e-07,
      "loss": 0.2377,
      "step": 93250
    },
    {
      "epoch": 280.06,
      "grad_norm": 10.123480796813965,
      "learning_rate": 6.646646646646647e-07,
      "loss": 0.193,
      "step": 93260
    },
    {
      "epoch": 280.09,
      "grad_norm": 6.229051113128662,
      "learning_rate": 6.636636636636637e-07,
      "loss": 0.192,
      "step": 93270
    },
    {
      "epoch": 280.12,
      "grad_norm": 17.550392150878906,
      "learning_rate": 6.626626626626627e-07,
      "loss": 0.2155,
      "step": 93280
    },
    {
      "epoch": 280.15,
      "grad_norm": 15.615151405334473,
      "learning_rate": 6.616616616616617e-07,
      "loss": 0.1798,
      "step": 93290
    },
    {
      "epoch": 280.18,
      "grad_norm": 11.352758407592773,
      "learning_rate": 6.606606606606607e-07,
      "loss": 0.1858,
      "step": 93300
    },
    {
      "epoch": 280.21,
      "grad_norm": 7.769526958465576,
      "learning_rate": 6.596596596596596e-07,
      "loss": 0.173,
      "step": 93310
    },
    {
      "epoch": 280.24,
      "grad_norm": 7.91024923324585,
      "learning_rate": 6.586586586586586e-07,
      "loss": 0.1542,
      "step": 93320
    },
    {
      "epoch": 280.27,
      "grad_norm": 9.962808609008789,
      "learning_rate": 6.576576576576577e-07,
      "loss": 0.1875,
      "step": 93330
    },
    {
      "epoch": 280.3,
      "grad_norm": 11.549215316772461,
      "learning_rate": 6.566566566566567e-07,
      "loss": 0.1623,
      "step": 93340
    },
    {
      "epoch": 280.33,
      "grad_norm": 7.846828937530518,
      "learning_rate": 6.556556556556556e-07,
      "loss": 0.192,
      "step": 93350
    },
    {
      "epoch": 280.36,
      "grad_norm": 48.70316696166992,
      "learning_rate": 6.546546546546548e-07,
      "loss": 0.2016,
      "step": 93360
    },
    {
      "epoch": 280.39,
      "grad_norm": 12.281060218811035,
      "learning_rate": 6.536536536536537e-07,
      "loss": 0.1809,
      "step": 93370
    },
    {
      "epoch": 280.42,
      "grad_norm": 5.33806037902832,
      "learning_rate": 6.526526526526528e-07,
      "loss": 0.2089,
      "step": 93380
    },
    {
      "epoch": 280.45,
      "grad_norm": 6.108957767486572,
      "learning_rate": 6.516516516516518e-07,
      "loss": 0.1792,
      "step": 93390
    },
    {
      "epoch": 280.48,
      "grad_norm": 5.368511199951172,
      "learning_rate": 6.506506506506508e-07,
      "loss": 0.1871,
      "step": 93400
    },
    {
      "epoch": 280.51,
      "grad_norm": 6.744450092315674,
      "learning_rate": 6.496496496496497e-07,
      "loss": 0.2089,
      "step": 93410
    },
    {
      "epoch": 280.54,
      "grad_norm": 7.362810134887695,
      "learning_rate": 6.486486486486487e-07,
      "loss": 0.1868,
      "step": 93420
    },
    {
      "epoch": 280.57,
      "grad_norm": 9.632051467895508,
      "learning_rate": 6.476476476476477e-07,
      "loss": 0.2038,
      "step": 93430
    },
    {
      "epoch": 280.6,
      "grad_norm": 21.63732147216797,
      "learning_rate": 6.466466466466468e-07,
      "loss": 0.193,
      "step": 93440
    },
    {
      "epoch": 280.63,
      "grad_norm": 13.561774253845215,
      "learning_rate": 6.456456456456457e-07,
      "loss": 0.1924,
      "step": 93450
    },
    {
      "epoch": 280.66,
      "grad_norm": 4.9927873611450195,
      "learning_rate": 6.446446446446447e-07,
      "loss": 0.1519,
      "step": 93460
    },
    {
      "epoch": 280.69,
      "grad_norm": 7.275759696960449,
      "learning_rate": 6.436436436436437e-07,
      "loss": 0.1623,
      "step": 93470
    },
    {
      "epoch": 280.72,
      "grad_norm": 7.824166297912598,
      "learning_rate": 6.426426426426427e-07,
      "loss": 0.1866,
      "step": 93480
    },
    {
      "epoch": 280.75,
      "grad_norm": 15.864542007446289,
      "learning_rate": 6.416416416416417e-07,
      "loss": 0.213,
      "step": 93490
    },
    {
      "epoch": 280.78,
      "grad_norm": 9.497118949890137,
      "learning_rate": 6.406406406406407e-07,
      "loss": 0.1739,
      "step": 93500
    },
    {
      "epoch": 280.81,
      "grad_norm": 11.265289306640625,
      "learning_rate": 6.396396396396397e-07,
      "loss": 0.2022,
      "step": 93510
    },
    {
      "epoch": 280.84,
      "grad_norm": 7.469902038574219,
      "learning_rate": 6.386386386386387e-07,
      "loss": 0.21,
      "step": 93520
    },
    {
      "epoch": 280.87,
      "grad_norm": 12.9201021194458,
      "learning_rate": 6.376376376376377e-07,
      "loss": 0.183,
      "step": 93530
    },
    {
      "epoch": 280.9,
      "grad_norm": 5.300271511077881,
      "learning_rate": 6.366366366366366e-07,
      "loss": 0.1758,
      "step": 93540
    },
    {
      "epoch": 280.93,
      "grad_norm": 4.769259452819824,
      "learning_rate": 6.356356356356356e-07,
      "loss": 0.16,
      "step": 93550
    },
    {
      "epoch": 280.96,
      "grad_norm": 6.864993572235107,
      "learning_rate": 6.346346346346347e-07,
      "loss": 0.2162,
      "step": 93560
    },
    {
      "epoch": 280.99,
      "grad_norm": 6.948833465576172,
      "learning_rate": 6.336336336336337e-07,
      "loss": 0.1808,
      "step": 93570
    },
    {
      "epoch": 281.0,
      "eval_accuracy": 0.9227,
      "eval_loss": 0.35027581453323364,
      "eval_runtime": 29.823,
      "eval_samples_per_second": 335.312,
      "eval_steps_per_second": 1.341,
      "step": 93573
    },
    {
      "epoch": 281.02,
      "grad_norm": 9.775230407714844,
      "learning_rate": 6.326326326326326e-07,
      "loss": 0.148,
      "step": 93580
    },
    {
      "epoch": 281.05,
      "grad_norm": 13.888908386230469,
      "learning_rate": 6.316316316316316e-07,
      "loss": 0.1809,
      "step": 93590
    },
    {
      "epoch": 281.08,
      "grad_norm": 9.402486801147461,
      "learning_rate": 6.306306306306306e-07,
      "loss": 0.1954,
      "step": 93600
    },
    {
      "epoch": 281.11,
      "grad_norm": 5.247973442077637,
      "learning_rate": 6.296296296296296e-07,
      "loss": 0.1793,
      "step": 93610
    },
    {
      "epoch": 281.14,
      "grad_norm": 14.54990291595459,
      "learning_rate": 6.286286286286287e-07,
      "loss": 0.1898,
      "step": 93620
    },
    {
      "epoch": 281.17,
      "grad_norm": 8.074441909790039,
      "learning_rate": 6.276276276276278e-07,
      "loss": 0.2017,
      "step": 93630
    },
    {
      "epoch": 281.2,
      "grad_norm": 9.701157569885254,
      "learning_rate": 6.266266266266267e-07,
      "loss": 0.2061,
      "step": 93640
    },
    {
      "epoch": 281.23,
      "grad_norm": 9.232666969299316,
      "learning_rate": 6.256256256256257e-07,
      "loss": 0.1727,
      "step": 93650
    },
    {
      "epoch": 281.26,
      "grad_norm": 5.250031471252441,
      "learning_rate": 6.246246246246246e-07,
      "loss": 0.1689,
      "step": 93660
    },
    {
      "epoch": 281.29,
      "grad_norm": 4.4208197593688965,
      "learning_rate": 6.236236236236236e-07,
      "loss": 0.2004,
      "step": 93670
    },
    {
      "epoch": 281.32,
      "grad_norm": 8.678750038146973,
      "learning_rate": 6.226226226226227e-07,
      "loss": 0.1764,
      "step": 93680
    },
    {
      "epoch": 281.35,
      "grad_norm": 3.7998883724212646,
      "learning_rate": 6.216216216216217e-07,
      "loss": 0.1745,
      "step": 93690
    },
    {
      "epoch": 281.38,
      "grad_norm": 8.855843544006348,
      "learning_rate": 6.206206206206207e-07,
      "loss": 0.1682,
      "step": 93700
    },
    {
      "epoch": 281.41,
      "grad_norm": 7.77747917175293,
      "learning_rate": 6.196196196196197e-07,
      "loss": 0.1806,
      "step": 93710
    },
    {
      "epoch": 281.44,
      "grad_norm": 6.188965797424316,
      "learning_rate": 6.186186186186187e-07,
      "loss": 0.2176,
      "step": 93720
    },
    {
      "epoch": 281.47,
      "grad_norm": 9.609407424926758,
      "learning_rate": 6.176176176176177e-07,
      "loss": 0.2166,
      "step": 93730
    },
    {
      "epoch": 281.5,
      "grad_norm": 6.1971940994262695,
      "learning_rate": 6.166166166166167e-07,
      "loss": 0.2208,
      "step": 93740
    },
    {
      "epoch": 281.53,
      "grad_norm": 9.02153205871582,
      "learning_rate": 6.156156156156157e-07,
      "loss": 0.1495,
      "step": 93750
    },
    {
      "epoch": 281.56,
      "grad_norm": 6.195413112640381,
      "learning_rate": 6.146146146146147e-07,
      "loss": 0.1642,
      "step": 93760
    },
    {
      "epoch": 281.59,
      "grad_norm": 5.703518867492676,
      "learning_rate": 6.136136136136136e-07,
      "loss": 0.1999,
      "step": 93770
    },
    {
      "epoch": 281.62,
      "grad_norm": 4.874156951904297,
      "learning_rate": 6.126126126126126e-07,
      "loss": 0.2147,
      "step": 93780
    },
    {
      "epoch": 281.65,
      "grad_norm": 8.158533096313477,
      "learning_rate": 6.116116116116117e-07,
      "loss": 0.1705,
      "step": 93790
    },
    {
      "epoch": 281.68,
      "grad_norm": 9.90766716003418,
      "learning_rate": 6.106106106106107e-07,
      "loss": 0.2023,
      "step": 93800
    },
    {
      "epoch": 281.71,
      "grad_norm": 7.135465145111084,
      "learning_rate": 6.096096096096096e-07,
      "loss": 0.1754,
      "step": 93810
    },
    {
      "epoch": 281.74,
      "grad_norm": 8.11416244506836,
      "learning_rate": 6.086086086086086e-07,
      "loss": 0.1716,
      "step": 93820
    },
    {
      "epoch": 281.77,
      "grad_norm": 9.946249008178711,
      "learning_rate": 6.076076076076077e-07,
      "loss": 0.2125,
      "step": 93830
    },
    {
      "epoch": 281.8,
      "grad_norm": 10.247566223144531,
      "learning_rate": 6.066066066066066e-07,
      "loss": 0.1915,
      "step": 93840
    },
    {
      "epoch": 281.83,
      "grad_norm": 8.322664260864258,
      "learning_rate": 6.056056056056057e-07,
      "loss": 0.1827,
      "step": 93850
    },
    {
      "epoch": 281.86,
      "grad_norm": 8.903614044189453,
      "learning_rate": 6.046046046046047e-07,
      "loss": 0.1803,
      "step": 93860
    },
    {
      "epoch": 281.89,
      "grad_norm": 7.389410018920898,
      "learning_rate": 6.036036036036037e-07,
      "loss": 0.144,
      "step": 93870
    },
    {
      "epoch": 281.92,
      "grad_norm": 5.89046049118042,
      "learning_rate": 6.026026026026026e-07,
      "loss": 0.1741,
      "step": 93880
    },
    {
      "epoch": 281.95,
      "grad_norm": 6.4074201583862305,
      "learning_rate": 6.016016016016016e-07,
      "loss": 0.219,
      "step": 93890
    },
    {
      "epoch": 281.98,
      "grad_norm": 6.204282760620117,
      "learning_rate": 6.006006006006006e-07,
      "loss": 0.1749,
      "step": 93900
    },
    {
      "epoch": 282.0,
      "eval_accuracy": 0.9218,
      "eval_loss": 0.34923070669174194,
      "eval_runtime": 30.0444,
      "eval_samples_per_second": 332.841,
      "eval_steps_per_second": 1.331,
      "step": 93906
    },
    {
      "epoch": 282.01,
      "grad_norm": 9.81662654876709,
      "learning_rate": 5.995995995995997e-07,
      "loss": 0.2067,
      "step": 93910
    },
    {
      "epoch": 282.04,
      "grad_norm": 8.250527381896973,
      "learning_rate": 5.985985985985987e-07,
      "loss": 0.2025,
      "step": 93920
    },
    {
      "epoch": 282.07,
      "grad_norm": 9.411706924438477,
      "learning_rate": 5.975975975975976e-07,
      "loss": 0.1643,
      "step": 93930
    },
    {
      "epoch": 282.1,
      "grad_norm": 13.292866706848145,
      "learning_rate": 5.965965965965966e-07,
      "loss": 0.2123,
      "step": 93940
    },
    {
      "epoch": 282.13,
      "grad_norm": 29.6545352935791,
      "learning_rate": 5.955955955955956e-07,
      "loss": 0.1691,
      "step": 93950
    },
    {
      "epoch": 282.16,
      "grad_norm": 7.744854927062988,
      "learning_rate": 5.945945945945947e-07,
      "loss": 0.192,
      "step": 93960
    },
    {
      "epoch": 282.19,
      "grad_norm": 17.168014526367188,
      "learning_rate": 5.935935935935937e-07,
      "loss": 0.1676,
      "step": 93970
    },
    {
      "epoch": 282.22,
      "grad_norm": 11.471046447753906,
      "learning_rate": 5.925925925925927e-07,
      "loss": 0.1901,
      "step": 93980
    },
    {
      "epoch": 282.25,
      "grad_norm": 6.568400859832764,
      "learning_rate": 5.915915915915917e-07,
      "loss": 0.1934,
      "step": 93990
    },
    {
      "epoch": 282.28,
      "grad_norm": 9.215129852294922,
      "learning_rate": 5.905905905905906e-07,
      "loss": 0.1859,
      "step": 94000
    },
    {
      "epoch": 282.31,
      "grad_norm": 6.579705238342285,
      "learning_rate": 5.895895895895896e-07,
      "loss": 0.1829,
      "step": 94010
    },
    {
      "epoch": 282.34,
      "grad_norm": 8.719270706176758,
      "learning_rate": 5.885885885885887e-07,
      "loss": 0.1979,
      "step": 94020
    },
    {
      "epoch": 282.37,
      "grad_norm": 6.2895708084106445,
      "learning_rate": 5.875875875875877e-07,
      "loss": 0.1932,
      "step": 94030
    },
    {
      "epoch": 282.4,
      "grad_norm": 4.246201038360596,
      "learning_rate": 5.865865865865866e-07,
      "loss": 0.1844,
      "step": 94040
    },
    {
      "epoch": 282.43,
      "grad_norm": 8.381440162658691,
      "learning_rate": 5.855855855855856e-07,
      "loss": 0.2049,
      "step": 94050
    },
    {
      "epoch": 282.46,
      "grad_norm": 13.026788711547852,
      "learning_rate": 5.845845845845846e-07,
      "loss": 0.1959,
      "step": 94060
    },
    {
      "epoch": 282.49,
      "grad_norm": 7.2175679206848145,
      "learning_rate": 5.835835835835836e-07,
      "loss": 0.1885,
      "step": 94070
    },
    {
      "epoch": 282.52,
      "grad_norm": 6.4432783126831055,
      "learning_rate": 5.825825825825826e-07,
      "loss": 0.2064,
      "step": 94080
    },
    {
      "epoch": 282.55,
      "grad_norm": 17.453201293945312,
      "learning_rate": 5.815815815815816e-07,
      "loss": 0.1756,
      "step": 94090
    },
    {
      "epoch": 282.58,
      "grad_norm": 8.117071151733398,
      "learning_rate": 5.805805805805807e-07,
      "loss": 0.1903,
      "step": 94100
    },
    {
      "epoch": 282.61,
      "grad_norm": 6.975236892700195,
      "learning_rate": 5.795795795795796e-07,
      "loss": 0.2065,
      "step": 94110
    },
    {
      "epoch": 282.64,
      "grad_norm": 10.81727409362793,
      "learning_rate": 5.785785785785786e-07,
      "loss": 0.2314,
      "step": 94120
    },
    {
      "epoch": 282.67,
      "grad_norm": 16.06839370727539,
      "learning_rate": 5.775775775775776e-07,
      "loss": 0.1438,
      "step": 94130
    },
    {
      "epoch": 282.7,
      "grad_norm": 7.4369659423828125,
      "learning_rate": 5.765765765765767e-07,
      "loss": 0.1851,
      "step": 94140
    },
    {
      "epoch": 282.73,
      "grad_norm": 7.700915336608887,
      "learning_rate": 5.755755755755756e-07,
      "loss": 0.1816,
      "step": 94150
    },
    {
      "epoch": 282.76,
      "grad_norm": 11.128373146057129,
      "learning_rate": 5.745745745745746e-07,
      "loss": 0.2065,
      "step": 94160
    },
    {
      "epoch": 282.79,
      "grad_norm": 12.14608383178711,
      "learning_rate": 5.735735735735736e-07,
      "loss": 0.1925,
      "step": 94170
    },
    {
      "epoch": 282.82,
      "grad_norm": 4.501680374145508,
      "learning_rate": 5.725725725725726e-07,
      "loss": 0.1551,
      "step": 94180
    },
    {
      "epoch": 282.85,
      "grad_norm": 12.111481666564941,
      "learning_rate": 5.715715715715716e-07,
      "loss": 0.1614,
      "step": 94190
    },
    {
      "epoch": 282.88,
      "grad_norm": 6.0821943283081055,
      "learning_rate": 5.705705705705706e-07,
      "loss": 0.1453,
      "step": 94200
    },
    {
      "epoch": 282.91,
      "grad_norm": 8.645496368408203,
      "learning_rate": 5.695695695695696e-07,
      "loss": 0.188,
      "step": 94210
    },
    {
      "epoch": 282.94,
      "grad_norm": 4.244828701019287,
      "learning_rate": 5.685685685685686e-07,
      "loss": 0.1889,
      "step": 94220
    },
    {
      "epoch": 282.97,
      "grad_norm": 10.827722549438477,
      "learning_rate": 5.675675675675676e-07,
      "loss": 0.2295,
      "step": 94230
    },
    {
      "epoch": 283.0,
      "eval_accuracy": 0.9216,
      "eval_loss": 0.34925174713134766,
      "eval_runtime": 30.1131,
      "eval_samples_per_second": 332.081,
      "eval_steps_per_second": 1.328,
      "step": 94239
    },
    {
      "epoch": 283.0,
      "grad_norm": 8.936872482299805,
      "learning_rate": 5.665665665665666e-07,
      "loss": 0.1568,
      "step": 94240
    },
    {
      "epoch": 283.03,
      "grad_norm": 11.198091506958008,
      "learning_rate": 5.655655655655657e-07,
      "loss": 0.1539,
      "step": 94250
    },
    {
      "epoch": 283.06,
      "grad_norm": 8.481804847717285,
      "learning_rate": 5.645645645645647e-07,
      "loss": 0.1835,
      "step": 94260
    },
    {
      "epoch": 283.09,
      "grad_norm": 6.851963996887207,
      "learning_rate": 5.635635635635636e-07,
      "loss": 0.1797,
      "step": 94270
    },
    {
      "epoch": 283.12,
      "grad_norm": 14.650593757629395,
      "learning_rate": 5.625625625625626e-07,
      "loss": 0.2061,
      "step": 94280
    },
    {
      "epoch": 283.15,
      "grad_norm": 17.928924560546875,
      "learning_rate": 5.615615615615616e-07,
      "loss": 0.2118,
      "step": 94290
    },
    {
      "epoch": 283.18,
      "grad_norm": 4.794007301330566,
      "learning_rate": 5.605605605605606e-07,
      "loss": 0.1516,
      "step": 94300
    },
    {
      "epoch": 283.21,
      "grad_norm": 23.92691421508789,
      "learning_rate": 5.595595595595596e-07,
      "loss": 0.1973,
      "step": 94310
    },
    {
      "epoch": 283.24,
      "grad_norm": 7.790602684020996,
      "learning_rate": 5.585585585585586e-07,
      "loss": 0.1866,
      "step": 94320
    },
    {
      "epoch": 283.27,
      "grad_norm": 10.867959022521973,
      "learning_rate": 5.575575575575576e-07,
      "loss": 0.1871,
      "step": 94330
    },
    {
      "epoch": 283.3,
      "grad_norm": 79.43052673339844,
      "learning_rate": 5.565565565565566e-07,
      "loss": 0.175,
      "step": 94340
    },
    {
      "epoch": 283.33,
      "grad_norm": 7.877035617828369,
      "learning_rate": 5.555555555555555e-07,
      "loss": 0.2051,
      "step": 94350
    },
    {
      "epoch": 283.36,
      "grad_norm": 10.99435806274414,
      "learning_rate": 5.545545545545546e-07,
      "loss": 0.1999,
      "step": 94360
    },
    {
      "epoch": 283.39,
      "grad_norm": 7.962160587310791,
      "learning_rate": 5.535535535535537e-07,
      "loss": 0.1464,
      "step": 94370
    },
    {
      "epoch": 283.42,
      "grad_norm": 10.05582332611084,
      "learning_rate": 5.525525525525526e-07,
      "loss": 0.1444,
      "step": 94380
    },
    {
      "epoch": 283.45,
      "grad_norm": 6.796876430511475,
      "learning_rate": 5.515515515515516e-07,
      "loss": 0.1927,
      "step": 94390
    },
    {
      "epoch": 283.48,
      "grad_norm": 6.270811557769775,
      "learning_rate": 5.505505505505506e-07,
      "loss": 0.1745,
      "step": 94400
    },
    {
      "epoch": 283.51,
      "grad_norm": 6.675833702087402,
      "learning_rate": 5.495495495495496e-07,
      "loss": 0.1738,
      "step": 94410
    },
    {
      "epoch": 283.54,
      "grad_norm": 10.13005256652832,
      "learning_rate": 5.485485485485486e-07,
      "loss": 0.1682,
      "step": 94420
    },
    {
      "epoch": 283.57,
      "grad_norm": 9.481799125671387,
      "learning_rate": 5.475475475475476e-07,
      "loss": 0.1696,
      "step": 94430
    },
    {
      "epoch": 283.6,
      "grad_norm": 4.942862510681152,
      "learning_rate": 5.465465465465466e-07,
      "loss": 0.1594,
      "step": 94440
    },
    {
      "epoch": 283.63,
      "grad_norm": 10.011496543884277,
      "learning_rate": 5.455455455455456e-07,
      "loss": 0.1621,
      "step": 94450
    },
    {
      "epoch": 283.66,
      "grad_norm": 16.061229705810547,
      "learning_rate": 5.445445445445446e-07,
      "loss": 0.1869,
      "step": 94460
    },
    {
      "epoch": 283.69,
      "grad_norm": 7.582136631011963,
      "learning_rate": 5.435435435435435e-07,
      "loss": 0.2174,
      "step": 94470
    },
    {
      "epoch": 283.72,
      "grad_norm": 11.908390045166016,
      "learning_rate": 5.425425425425425e-07,
      "loss": 0.1925,
      "step": 94480
    },
    {
      "epoch": 283.75,
      "grad_norm": 7.167508602142334,
      "learning_rate": 5.415415415415416e-07,
      "loss": 0.1863,
      "step": 94490
    },
    {
      "epoch": 283.78,
      "grad_norm": 11.765337944030762,
      "learning_rate": 5.405405405405406e-07,
      "loss": 0.1989,
      "step": 94500
    },
    {
      "epoch": 283.81,
      "grad_norm": 7.536864280700684,
      "learning_rate": 5.395395395395396e-07,
      "loss": 0.1767,
      "step": 94510
    },
    {
      "epoch": 283.84,
      "grad_norm": 7.4894537925720215,
      "learning_rate": 5.385385385385386e-07,
      "loss": 0.2049,
      "step": 94520
    },
    {
      "epoch": 283.87,
      "grad_norm": 7.868075370788574,
      "learning_rate": 5.375375375375376e-07,
      "loss": 0.1999,
      "step": 94530
    },
    {
      "epoch": 283.9,
      "grad_norm": 5.448962688446045,
      "learning_rate": 5.365365365365365e-07,
      "loss": 0.1921,
      "step": 94540
    },
    {
      "epoch": 283.93,
      "grad_norm": 3.394087076187134,
      "learning_rate": 5.355355355355356e-07,
      "loss": 0.1737,
      "step": 94550
    },
    {
      "epoch": 283.96,
      "grad_norm": 9.814059257507324,
      "learning_rate": 5.345345345345346e-07,
      "loss": 0.1458,
      "step": 94560
    },
    {
      "epoch": 283.99,
      "grad_norm": 5.9682135581970215,
      "learning_rate": 5.335335335335336e-07,
      "loss": 0.1695,
      "step": 94570
    },
    {
      "epoch": 284.0,
      "eval_accuracy": 0.9219,
      "eval_loss": 0.34910914301872253,
      "eval_runtime": 30.2897,
      "eval_samples_per_second": 330.145,
      "eval_steps_per_second": 1.321,
      "step": 94572
    },
    {
      "epoch": 284.02,
      "grad_norm": 4.850225925445557,
      "learning_rate": 5.325325325325325e-07,
      "loss": 0.2496,
      "step": 94580
    },
    {
      "epoch": 284.05,
      "grad_norm": 51.408164978027344,
      "learning_rate": 5.315315315315315e-07,
      "loss": 0.1771,
      "step": 94590
    },
    {
      "epoch": 284.08,
      "grad_norm": 18.794231414794922,
      "learning_rate": 5.305305305305306e-07,
      "loss": 0.1797,
      "step": 94600
    },
    {
      "epoch": 284.11,
      "grad_norm": 11.393068313598633,
      "learning_rate": 5.295295295295296e-07,
      "loss": 0.2092,
      "step": 94610
    },
    {
      "epoch": 284.14,
      "grad_norm": 6.675703525543213,
      "learning_rate": 5.285285285285286e-07,
      "loss": 0.1657,
      "step": 94620
    },
    {
      "epoch": 284.17,
      "grad_norm": 8.380932807922363,
      "learning_rate": 5.275275275275276e-07,
      "loss": 0.2127,
      "step": 94630
    },
    {
      "epoch": 284.2,
      "grad_norm": 9.57076644897461,
      "learning_rate": 5.265265265265266e-07,
      "loss": 0.1963,
      "step": 94640
    },
    {
      "epoch": 284.23,
      "grad_norm": 5.876262664794922,
      "learning_rate": 5.255255255255255e-07,
      "loss": 0.2172,
      "step": 94650
    },
    {
      "epoch": 284.26,
      "grad_norm": 6.727551460266113,
      "learning_rate": 5.245245245245246e-07,
      "loss": 0.1545,
      "step": 94660
    },
    {
      "epoch": 284.29,
      "grad_norm": 6.392829418182373,
      "learning_rate": 5.235235235235236e-07,
      "loss": 0.2091,
      "step": 94670
    },
    {
      "epoch": 284.32,
      "grad_norm": 10.17914867401123,
      "learning_rate": 5.225225225225226e-07,
      "loss": 0.1896,
      "step": 94680
    },
    {
      "epoch": 284.35,
      "grad_norm": 13.885440826416016,
      "learning_rate": 5.215215215215216e-07,
      "loss": 0.1964,
      "step": 94690
    },
    {
      "epoch": 284.38,
      "grad_norm": 11.61838150024414,
      "learning_rate": 5.205205205205205e-07,
      "loss": 0.1577,
      "step": 94700
    },
    {
      "epoch": 284.41,
      "grad_norm": 10.008115768432617,
      "learning_rate": 5.195195195195195e-07,
      "loss": 0.2375,
      "step": 94710
    },
    {
      "epoch": 284.44,
      "grad_norm": 5.152472019195557,
      "learning_rate": 5.185185185185186e-07,
      "loss": 0.1811,
      "step": 94720
    },
    {
      "epoch": 284.47,
      "grad_norm": 6.668306350708008,
      "learning_rate": 5.175175175175176e-07,
      "loss": 0.1841,
      "step": 94730
    },
    {
      "epoch": 284.5,
      "grad_norm": 9.555723190307617,
      "learning_rate": 5.165165165165165e-07,
      "loss": 0.155,
      "step": 94740
    },
    {
      "epoch": 284.53,
      "grad_norm": 8.553205490112305,
      "learning_rate": 5.155155155155155e-07,
      "loss": 0.1805,
      "step": 94750
    },
    {
      "epoch": 284.56,
      "grad_norm": 9.926412582397461,
      "learning_rate": 5.145145145145146e-07,
      "loss": 0.1741,
      "step": 94760
    },
    {
      "epoch": 284.59,
      "grad_norm": 10.089454650878906,
      "learning_rate": 5.135135135135135e-07,
      "loss": 0.2021,
      "step": 94770
    },
    {
      "epoch": 284.62,
      "grad_norm": 38.9876708984375,
      "learning_rate": 5.125125125125126e-07,
      "loss": 0.1898,
      "step": 94780
    },
    {
      "epoch": 284.65,
      "grad_norm": 11.25538158416748,
      "learning_rate": 5.115115115115116e-07,
      "loss": 0.1859,
      "step": 94790
    },
    {
      "epoch": 284.68,
      "grad_norm": 6.999671936035156,
      "learning_rate": 5.105105105105106e-07,
      "loss": 0.1699,
      "step": 94800
    },
    {
      "epoch": 284.71,
      "grad_norm": 41.32756042480469,
      "learning_rate": 5.095095095095095e-07,
      "loss": 0.1879,
      "step": 94810
    },
    {
      "epoch": 284.74,
      "grad_norm": 48.02335739135742,
      "learning_rate": 5.085085085085085e-07,
      "loss": 0.1953,
      "step": 94820
    },
    {
      "epoch": 284.77,
      "grad_norm": 11.522754669189453,
      "learning_rate": 5.075075075075076e-07,
      "loss": 0.1746,
      "step": 94830
    },
    {
      "epoch": 284.8,
      "grad_norm": 7.453914642333984,
      "learning_rate": 5.065065065065066e-07,
      "loss": 0.1686,
      "step": 94840
    },
    {
      "epoch": 284.83,
      "grad_norm": 7.847193717956543,
      "learning_rate": 5.055055055055055e-07,
      "loss": 0.161,
      "step": 94850
    },
    {
      "epoch": 284.86,
      "grad_norm": 8.516251564025879,
      "learning_rate": 5.045045045045045e-07,
      "loss": 0.1786,
      "step": 94860
    },
    {
      "epoch": 284.89,
      "grad_norm": 9.196722984313965,
      "learning_rate": 5.035035035035035e-07,
      "loss": 0.1951,
      "step": 94870
    },
    {
      "epoch": 284.92,
      "grad_norm": 8.682083129882812,
      "learning_rate": 5.025025025025025e-07,
      "loss": 0.1757,
      "step": 94880
    },
    {
      "epoch": 284.95,
      "grad_norm": 4.999848365783691,
      "learning_rate": 5.015015015015016e-07,
      "loss": 0.1994,
      "step": 94890
    },
    {
      "epoch": 284.98,
      "grad_norm": 9.464710235595703,
      "learning_rate": 5.005005005005006e-07,
      "loss": 0.1859,
      "step": 94900
    },
    {
      "epoch": 285.0,
      "eval_accuracy": 0.9222,
      "eval_loss": 0.35018086433410645,
      "eval_runtime": 29.7051,
      "eval_samples_per_second": 336.642,
      "eval_steps_per_second": 1.347,
      "step": 94905
    },
    {
      "epoch": 285.02,
      "grad_norm": 9.307838439941406,
      "learning_rate": 4.994994994994996e-07,
      "loss": 0.1623,
      "step": 94910
    },
    {
      "epoch": 285.05,
      "grad_norm": 28.56440544128418,
      "learning_rate": 4.984984984984986e-07,
      "loss": 0.2122,
      "step": 94920
    },
    {
      "epoch": 285.08,
      "grad_norm": 6.306981563568115,
      "learning_rate": 4.974974974974975e-07,
      "loss": 0.1623,
      "step": 94930
    },
    {
      "epoch": 285.11,
      "grad_norm": 9.130448341369629,
      "learning_rate": 4.964964964964965e-07,
      "loss": 0.1916,
      "step": 94940
    },
    {
      "epoch": 285.14,
      "grad_norm": 9.843271255493164,
      "learning_rate": 4.954954954954956e-07,
      "loss": 0.1728,
      "step": 94950
    },
    {
      "epoch": 285.17,
      "grad_norm": 6.6711297035217285,
      "learning_rate": 4.944944944944946e-07,
      "loss": 0.1723,
      "step": 94960
    },
    {
      "epoch": 285.2,
      "grad_norm": 7.209865093231201,
      "learning_rate": 4.934934934934935e-07,
      "loss": 0.1725,
      "step": 94970
    },
    {
      "epoch": 285.23,
      "grad_norm": 8.495177268981934,
      "learning_rate": 4.924924924924925e-07,
      "loss": 0.1775,
      "step": 94980
    },
    {
      "epoch": 285.26,
      "grad_norm": 6.971652984619141,
      "learning_rate": 4.914914914914915e-07,
      "loss": 0.2396,
      "step": 94990
    },
    {
      "epoch": 285.29,
      "grad_norm": 8.527682304382324,
      "learning_rate": 4.904904904904905e-07,
      "loss": 0.1795,
      "step": 95000
    },
    {
      "epoch": 285.32,
      "grad_norm": 13.04582691192627,
      "learning_rate": 4.894894894894895e-07,
      "loss": 0.1982,
      "step": 95010
    },
    {
      "epoch": 285.35,
      "grad_norm": 20.321781158447266,
      "learning_rate": 4.884884884884885e-07,
      "loss": 0.1774,
      "step": 95020
    },
    {
      "epoch": 285.38,
      "grad_norm": 12.240036010742188,
      "learning_rate": 4.874874874874876e-07,
      "loss": 0.197,
      "step": 95030
    },
    {
      "epoch": 285.41,
      "grad_norm": 7.494658470153809,
      "learning_rate": 4.864864864864865e-07,
      "loss": 0.2053,
      "step": 95040
    },
    {
      "epoch": 285.44,
      "grad_norm": 9.447043418884277,
      "learning_rate": 4.854854854854855e-07,
      "loss": 0.2223,
      "step": 95050
    },
    {
      "epoch": 285.47,
      "grad_norm": 33.25238037109375,
      "learning_rate": 4.844844844844845e-07,
      "loss": 0.2062,
      "step": 95060
    },
    {
      "epoch": 285.5,
      "grad_norm": 8.61745548248291,
      "learning_rate": 4.834834834834836e-07,
      "loss": 0.167,
      "step": 95070
    },
    {
      "epoch": 285.53,
      "grad_norm": 5.8508219718933105,
      "learning_rate": 4.824824824824825e-07,
      "loss": 0.2449,
      "step": 95080
    },
    {
      "epoch": 285.56,
      "grad_norm": 5.904736518859863,
      "learning_rate": 4.814814814814815e-07,
      "loss": 0.1792,
      "step": 95090
    },
    {
      "epoch": 285.59,
      "grad_norm": 9.604451179504395,
      "learning_rate": 4.804804804804805e-07,
      "loss": 0.1755,
      "step": 95100
    },
    {
      "epoch": 285.62,
      "grad_norm": 9.858085632324219,
      "learning_rate": 4.794794794794795e-07,
      "loss": 0.1714,
      "step": 95110
    },
    {
      "epoch": 285.65,
      "grad_norm": 9.107651710510254,
      "learning_rate": 4.784784784784786e-07,
      "loss": 0.1457,
      "step": 95120
    },
    {
      "epoch": 285.68,
      "grad_norm": 8.432609558105469,
      "learning_rate": 4.774774774774775e-07,
      "loss": 0.1975,
      "step": 95130
    },
    {
      "epoch": 285.71,
      "grad_norm": 5.885417461395264,
      "learning_rate": 4.764764764764765e-07,
      "loss": 0.1924,
      "step": 95140
    },
    {
      "epoch": 285.74,
      "grad_norm": 11.298110961914062,
      "learning_rate": 4.754754754754755e-07,
      "loss": 0.1706,
      "step": 95150
    },
    {
      "epoch": 285.77,
      "grad_norm": 24.261489868164062,
      "learning_rate": 4.744744744744745e-07,
      "loss": 0.1613,
      "step": 95160
    },
    {
      "epoch": 285.8,
      "grad_norm": 6.564768314361572,
      "learning_rate": 4.7347347347347354e-07,
      "loss": 0.1713,
      "step": 95170
    },
    {
      "epoch": 285.83,
      "grad_norm": 9.69290828704834,
      "learning_rate": 4.7247247247247256e-07,
      "loss": 0.1666,
      "step": 95180
    },
    {
      "epoch": 285.86,
      "grad_norm": 9.562887191772461,
      "learning_rate": 4.714714714714715e-07,
      "loss": 0.1548,
      "step": 95190
    },
    {
      "epoch": 285.89,
      "grad_norm": 10.377954483032227,
      "learning_rate": 4.7047047047047054e-07,
      "loss": 0.1903,
      "step": 95200
    },
    {
      "epoch": 285.92,
      "grad_norm": 8.311456680297852,
      "learning_rate": 4.694694694694695e-07,
      "loss": 0.2281,
      "step": 95210
    },
    {
      "epoch": 285.95,
      "grad_norm": 11.366046905517578,
      "learning_rate": 4.684684684684685e-07,
      "loss": 0.196,
      "step": 95220
    },
    {
      "epoch": 285.98,
      "grad_norm": 8.109556198120117,
      "learning_rate": 4.674674674674675e-07,
      "loss": 0.1891,
      "step": 95230
    },
    {
      "epoch": 286.0,
      "eval_accuracy": 0.9226,
      "eval_loss": 0.35050201416015625,
      "eval_runtime": 30.5524,
      "eval_samples_per_second": 327.306,
      "eval_steps_per_second": 1.309,
      "step": 95238
    },
    {
      "epoch": 286.01,
      "grad_norm": 9.273707389831543,
      "learning_rate": 4.664664664664665e-07,
      "loss": 0.1462,
      "step": 95240
    },
    {
      "epoch": 286.04,
      "grad_norm": 6.225035667419434,
      "learning_rate": 4.654654654654655e-07,
      "loss": 0.1586,
      "step": 95250
    },
    {
      "epoch": 286.07,
      "grad_norm": 8.075520515441895,
      "learning_rate": 4.644644644644645e-07,
      "loss": 0.177,
      "step": 95260
    },
    {
      "epoch": 286.1,
      "grad_norm": 8.377695083618164,
      "learning_rate": 4.6346346346346346e-07,
      "loss": 0.2053,
      "step": 95270
    },
    {
      "epoch": 286.13,
      "grad_norm": 15.135931968688965,
      "learning_rate": 4.624624624624625e-07,
      "loss": 0.14,
      "step": 95280
    },
    {
      "epoch": 286.16,
      "grad_norm": 9.570109367370605,
      "learning_rate": 4.6146146146146144e-07,
      "loss": 0.2169,
      "step": 95290
    },
    {
      "epoch": 286.19,
      "grad_norm": 8.303789138793945,
      "learning_rate": 4.604604604604605e-07,
      "loss": 0.1666,
      "step": 95300
    },
    {
      "epoch": 286.22,
      "grad_norm": 13.612077713012695,
      "learning_rate": 4.5945945945945953e-07,
      "loss": 0.1972,
      "step": 95310
    },
    {
      "epoch": 286.25,
      "grad_norm": 15.526899337768555,
      "learning_rate": 4.584584584584585e-07,
      "loss": 0.2055,
      "step": 95320
    },
    {
      "epoch": 286.28,
      "grad_norm": 8.46012020111084,
      "learning_rate": 4.574574574574575e-07,
      "loss": 0.1576,
      "step": 95330
    },
    {
      "epoch": 286.31,
      "grad_norm": 4.892340660095215,
      "learning_rate": 4.564564564564565e-07,
      "loss": 0.1874,
      "step": 95340
    },
    {
      "epoch": 286.34,
      "grad_norm": 5.555288314819336,
      "learning_rate": 4.554554554554555e-07,
      "loss": 0.1627,
      "step": 95350
    },
    {
      "epoch": 286.37,
      "grad_norm": 6.299892902374268,
      "learning_rate": 4.544544544544545e-07,
      "loss": 0.1922,
      "step": 95360
    },
    {
      "epoch": 286.4,
      "grad_norm": 7.705307483673096,
      "learning_rate": 4.534534534534535e-07,
      "loss": 0.1726,
      "step": 95370
    },
    {
      "epoch": 286.43,
      "grad_norm": 9.585205078125,
      "learning_rate": 4.524524524524525e-07,
      "loss": 0.2022,
      "step": 95380
    },
    {
      "epoch": 286.46,
      "grad_norm": 26.713407516479492,
      "learning_rate": 4.5145145145145147e-07,
      "loss": 0.1828,
      "step": 95390
    },
    {
      "epoch": 286.49,
      "grad_norm": 9.380204200744629,
      "learning_rate": 4.504504504504505e-07,
      "loss": 0.1673,
      "step": 95400
    },
    {
      "epoch": 286.52,
      "grad_norm": 12.198960304260254,
      "learning_rate": 4.4944944944944945e-07,
      "loss": 0.2189,
      "step": 95410
    },
    {
      "epoch": 286.55,
      "grad_norm": 5.957647800445557,
      "learning_rate": 4.4844844844844847e-07,
      "loss": 0.1597,
      "step": 95420
    },
    {
      "epoch": 286.58,
      "grad_norm": 9.902233123779297,
      "learning_rate": 4.4744744744744754e-07,
      "loss": 0.2012,
      "step": 95430
    },
    {
      "epoch": 286.61,
      "grad_norm": 6.081257343292236,
      "learning_rate": 4.464464464464465e-07,
      "loss": 0.1796,
      "step": 95440
    },
    {
      "epoch": 286.64,
      "grad_norm": 9.488055229187012,
      "learning_rate": 4.454454454454455e-07,
      "loss": 0.1625,
      "step": 95450
    },
    {
      "epoch": 286.67,
      "grad_norm": 7.234827518463135,
      "learning_rate": 4.444444444444445e-07,
      "loss": 0.1929,
      "step": 95460
    },
    {
      "epoch": 286.7,
      "grad_norm": 16.00482940673828,
      "learning_rate": 4.434434434434435e-07,
      "loss": 0.1704,
      "step": 95470
    },
    {
      "epoch": 286.73,
      "grad_norm": 12.325714111328125,
      "learning_rate": 4.4244244244244247e-07,
      "loss": 0.1703,
      "step": 95480
    },
    {
      "epoch": 286.76,
      "grad_norm": 5.147406101226807,
      "learning_rate": 4.414414414414415e-07,
      "loss": 0.19,
      "step": 95490
    },
    {
      "epoch": 286.79,
      "grad_norm": 7.945609092712402,
      "learning_rate": 4.4044044044044046e-07,
      "loss": 0.1761,
      "step": 95500
    },
    {
      "epoch": 286.82,
      "grad_norm": 11.046072006225586,
      "learning_rate": 4.394394394394395e-07,
      "loss": 0.1845,
      "step": 95510
    },
    {
      "epoch": 286.85,
      "grad_norm": 5.655889511108398,
      "learning_rate": 4.3843843843843844e-07,
      "loss": 0.136,
      "step": 95520
    },
    {
      "epoch": 286.88,
      "grad_norm": 5.84739875793457,
      "learning_rate": 4.3743743743743746e-07,
      "loss": 0.1567,
      "step": 95530
    },
    {
      "epoch": 286.91,
      "grad_norm": 9.700919151306152,
      "learning_rate": 4.364364364364364e-07,
      "loss": 0.203,
      "step": 95540
    },
    {
      "epoch": 286.94,
      "grad_norm": 10.658275604248047,
      "learning_rate": 4.3543543543543544e-07,
      "loss": 0.1773,
      "step": 95550
    },
    {
      "epoch": 286.97,
      "grad_norm": 6.626459121704102,
      "learning_rate": 4.344344344344345e-07,
      "loss": 0.2137,
      "step": 95560
    },
    {
      "epoch": 287.0,
      "grad_norm": 7.708876132965088,
      "learning_rate": 4.334334334334335e-07,
      "loss": 0.1681,
      "step": 95570
    },
    {
      "epoch": 287.0,
      "eval_accuracy": 0.9222,
      "eval_loss": 0.3513043522834778,
      "eval_runtime": 30.135,
      "eval_samples_per_second": 331.84,
      "eval_steps_per_second": 1.327,
      "step": 95571
    },
    {
      "epoch": 287.03,
      "grad_norm": 9.463316917419434,
      "learning_rate": 4.324324324324325e-07,
      "loss": 0.1871,
      "step": 95580
    },
    {
      "epoch": 287.06,
      "grad_norm": 5.515777111053467,
      "learning_rate": 4.3143143143143146e-07,
      "loss": 0.1948,
      "step": 95590
    },
    {
      "epoch": 287.09,
      "grad_norm": 10.131659507751465,
      "learning_rate": 4.304304304304305e-07,
      "loss": 0.1633,
      "step": 95600
    },
    {
      "epoch": 287.12,
      "grad_norm": 10.565319061279297,
      "learning_rate": 4.294294294294295e-07,
      "loss": 0.2027,
      "step": 95610
    },
    {
      "epoch": 287.15,
      "grad_norm": 6.783892631530762,
      "learning_rate": 4.2842842842842846e-07,
      "loss": 0.1933,
      "step": 95620
    },
    {
      "epoch": 287.18,
      "grad_norm": 9.602127075195312,
      "learning_rate": 4.274274274274275e-07,
      "loss": 0.155,
      "step": 95630
    },
    {
      "epoch": 287.21,
      "grad_norm": 7.23532247543335,
      "learning_rate": 4.2642642642642645e-07,
      "loss": 0.173,
      "step": 95640
    },
    {
      "epoch": 287.24,
      "grad_norm": 21.138702392578125,
      "learning_rate": 4.2542542542542547e-07,
      "loss": 0.1983,
      "step": 95650
    },
    {
      "epoch": 287.27,
      "grad_norm": 12.429274559020996,
      "learning_rate": 4.2442442442442443e-07,
      "loss": 0.1902,
      "step": 95660
    },
    {
      "epoch": 287.3,
      "grad_norm": 7.976635456085205,
      "learning_rate": 4.2342342342342345e-07,
      "loss": 0.1895,
      "step": 95670
    },
    {
      "epoch": 287.33,
      "grad_norm": 14.226107597351074,
      "learning_rate": 4.224224224224224e-07,
      "loss": 0.1858,
      "step": 95680
    },
    {
      "epoch": 287.36,
      "grad_norm": 29.359359741210938,
      "learning_rate": 4.2142142142142143e-07,
      "loss": 0.1424,
      "step": 95690
    },
    {
      "epoch": 287.39,
      "grad_norm": 8.23883056640625,
      "learning_rate": 4.204204204204205e-07,
      "loss": 0.1739,
      "step": 95700
    },
    {
      "epoch": 287.42,
      "grad_norm": 43.82486343383789,
      "learning_rate": 4.1941941941941947e-07,
      "loss": 0.2026,
      "step": 95710
    },
    {
      "epoch": 287.45,
      "grad_norm": 12.800736427307129,
      "learning_rate": 4.184184184184185e-07,
      "loss": 0.2003,
      "step": 95720
    },
    {
      "epoch": 287.48,
      "grad_norm": 12.155216217041016,
      "learning_rate": 4.1741741741741745e-07,
      "loss": 0.1982,
      "step": 95730
    },
    {
      "epoch": 287.51,
      "grad_norm": 58.86433410644531,
      "learning_rate": 4.1641641641641647e-07,
      "loss": 0.1981,
      "step": 95740
    },
    {
      "epoch": 287.54,
      "grad_norm": 7.0858964920043945,
      "learning_rate": 4.1541541541541544e-07,
      "loss": 0.194,
      "step": 95750
    },
    {
      "epoch": 287.57,
      "grad_norm": 18.576940536499023,
      "learning_rate": 4.1441441441441446e-07,
      "loss": 0.1545,
      "step": 95760
    },
    {
      "epoch": 287.6,
      "grad_norm": 8.78947925567627,
      "learning_rate": 4.134134134134134e-07,
      "loss": 0.2041,
      "step": 95770
    },
    {
      "epoch": 287.63,
      "grad_norm": 6.29293966293335,
      "learning_rate": 4.1241241241241244e-07,
      "loss": 0.2145,
      "step": 95780
    },
    {
      "epoch": 287.66,
      "grad_norm": 4.1282854080200195,
      "learning_rate": 4.114114114114114e-07,
      "loss": 0.1678,
      "step": 95790
    },
    {
      "epoch": 287.69,
      "grad_norm": 6.43406867980957,
      "learning_rate": 4.104104104104104e-07,
      "loss": 0.1702,
      "step": 95800
    },
    {
      "epoch": 287.72,
      "grad_norm": 4.325094699859619,
      "learning_rate": 4.0940940940940944e-07,
      "loss": 0.1589,
      "step": 95810
    },
    {
      "epoch": 287.75,
      "grad_norm": 9.326000213623047,
      "learning_rate": 4.084084084084084e-07,
      "loss": 0.1869,
      "step": 95820
    },
    {
      "epoch": 287.78,
      "grad_norm": 11.061281204223633,
      "learning_rate": 4.074074074074075e-07,
      "loss": 0.1734,
      "step": 95830
    },
    {
      "epoch": 287.81,
      "grad_norm": 12.5880765914917,
      "learning_rate": 4.0640640640640644e-07,
      "loss": 0.1564,
      "step": 95840
    },
    {
      "epoch": 287.84,
      "grad_norm": 12.08588695526123,
      "learning_rate": 4.0540540540540546e-07,
      "loss": 0.1958,
      "step": 95850
    },
    {
      "epoch": 287.87,
      "grad_norm": 5.639970779418945,
      "learning_rate": 4.044044044044045e-07,
      "loss": 0.1652,
      "step": 95860
    },
    {
      "epoch": 287.9,
      "grad_norm": 6.728212356567383,
      "learning_rate": 4.0340340340340345e-07,
      "loss": 0.1782,
      "step": 95870
    },
    {
      "epoch": 287.93,
      "grad_norm": 7.387587070465088,
      "learning_rate": 4.0240240240240246e-07,
      "loss": 0.1826,
      "step": 95880
    },
    {
      "epoch": 287.96,
      "grad_norm": 9.0371732711792,
      "learning_rate": 4.0140140140140143e-07,
      "loss": 0.1962,
      "step": 95890
    },
    {
      "epoch": 287.99,
      "grad_norm": 6.228532314300537,
      "learning_rate": 4.0040040040040045e-07,
      "loss": 0.1837,
      "step": 95900
    },
    {
      "epoch": 288.0,
      "eval_accuracy": 0.9217,
      "eval_loss": 0.35124605894088745,
      "eval_runtime": 30.3674,
      "eval_samples_per_second": 329.3,
      "eval_steps_per_second": 1.317,
      "step": 95904
    },
    {
      "epoch": 288.02,
      "grad_norm": 7.958787441253662,
      "learning_rate": 3.993993993993994e-07,
      "loss": 0.2067,
      "step": 95910
    },
    {
      "epoch": 288.05,
      "grad_norm": 16.831693649291992,
      "learning_rate": 3.9839839839839843e-07,
      "loss": 0.1883,
      "step": 95920
    },
    {
      "epoch": 288.08,
      "grad_norm": 5.480751037597656,
      "learning_rate": 3.973973973973974e-07,
      "loss": 0.1922,
      "step": 95930
    },
    {
      "epoch": 288.11,
      "grad_norm": 6.630716800689697,
      "learning_rate": 3.963963963963964e-07,
      "loss": 0.1649,
      "step": 95940
    },
    {
      "epoch": 288.14,
      "grad_norm": 7.405264377593994,
      "learning_rate": 3.953953953953954e-07,
      "loss": 0.1807,
      "step": 95950
    },
    {
      "epoch": 288.17,
      "grad_norm": 9.194046974182129,
      "learning_rate": 3.943943943943944e-07,
      "loss": 0.14,
      "step": 95960
    },
    {
      "epoch": 288.2,
      "grad_norm": 11.892428398132324,
      "learning_rate": 3.9339339339339347e-07,
      "loss": 0.2028,
      "step": 95970
    },
    {
      "epoch": 288.23,
      "grad_norm": 12.849623680114746,
      "learning_rate": 3.9239239239239244e-07,
      "loss": 0.1598,
      "step": 95980
    },
    {
      "epoch": 288.26,
      "grad_norm": 11.590575218200684,
      "learning_rate": 3.9139139139139145e-07,
      "loss": 0.173,
      "step": 95990
    },
    {
      "epoch": 288.29,
      "grad_norm": 11.886320114135742,
      "learning_rate": 3.903903903903904e-07,
      "loss": 0.2124,
      "step": 96000
    },
    {
      "epoch": 288.32,
      "grad_norm": 7.467533111572266,
      "learning_rate": 3.8938938938938944e-07,
      "loss": 0.2023,
      "step": 96010
    },
    {
      "epoch": 288.35,
      "grad_norm": 7.287947177886963,
      "learning_rate": 3.883883883883884e-07,
      "loss": 0.2318,
      "step": 96020
    },
    {
      "epoch": 288.38,
      "grad_norm": 8.058730125427246,
      "learning_rate": 3.873873873873874e-07,
      "loss": 0.1789,
      "step": 96030
    },
    {
      "epoch": 288.41,
      "grad_norm": 5.89462423324585,
      "learning_rate": 3.863863863863864e-07,
      "loss": 0.1532,
      "step": 96040
    },
    {
      "epoch": 288.44,
      "grad_norm": 7.697813510894775,
      "learning_rate": 3.853853853853854e-07,
      "loss": 0.1903,
      "step": 96050
    },
    {
      "epoch": 288.47,
      "grad_norm": 8.74051570892334,
      "learning_rate": 3.843843843843844e-07,
      "loss": 0.1783,
      "step": 96060
    },
    {
      "epoch": 288.5,
      "grad_norm": 8.013879776000977,
      "learning_rate": 3.833833833833834e-07,
      "loss": 0.1735,
      "step": 96070
    },
    {
      "epoch": 288.53,
      "grad_norm": 8.780600547790527,
      "learning_rate": 3.823823823823824e-07,
      "loss": 0.1941,
      "step": 96080
    },
    {
      "epoch": 288.56,
      "grad_norm": 20.38017463684082,
      "learning_rate": 3.8138138138138137e-07,
      "loss": 0.1905,
      "step": 96090
    },
    {
      "epoch": 288.59,
      "grad_norm": 6.356158256530762,
      "learning_rate": 3.8038038038038044e-07,
      "loss": 0.1773,
      "step": 96100
    },
    {
      "epoch": 288.62,
      "grad_norm": 8.73310661315918,
      "learning_rate": 3.793793793793794e-07,
      "loss": 0.1762,
      "step": 96110
    },
    {
      "epoch": 288.65,
      "grad_norm": 9.060988426208496,
      "learning_rate": 3.7837837837837843e-07,
      "loss": 0.2261,
      "step": 96120
    },
    {
      "epoch": 288.68,
      "grad_norm": 6.348284721374512,
      "learning_rate": 3.7737737737737745e-07,
      "loss": 0.1799,
      "step": 96130
    },
    {
      "epoch": 288.71,
      "grad_norm": 7.587226867675781,
      "learning_rate": 3.763763763763764e-07,
      "loss": 0.1463,
      "step": 96140
    },
    {
      "epoch": 288.74,
      "grad_norm": 14.326475143432617,
      "learning_rate": 3.7537537537537543e-07,
      "loss": 0.2025,
      "step": 96150
    },
    {
      "epoch": 288.77,
      "grad_norm": 19.941377639770508,
      "learning_rate": 3.743743743743744e-07,
      "loss": 0.1912,
      "step": 96160
    },
    {
      "epoch": 288.8,
      "grad_norm": 10.933481216430664,
      "learning_rate": 3.733733733733734e-07,
      "loss": 0.1933,
      "step": 96170
    },
    {
      "epoch": 288.83,
      "grad_norm": 5.959723949432373,
      "learning_rate": 3.723723723723724e-07,
      "loss": 0.1414,
      "step": 96180
    },
    {
      "epoch": 288.86,
      "grad_norm": 5.4270782470703125,
      "learning_rate": 3.713713713713714e-07,
      "loss": 0.1783,
      "step": 96190
    },
    {
      "epoch": 288.89,
      "grad_norm": 17.518526077270508,
      "learning_rate": 3.7037037037037036e-07,
      "loss": 0.1591,
      "step": 96200
    },
    {
      "epoch": 288.92,
      "grad_norm": 8.23892593383789,
      "learning_rate": 3.693693693693694e-07,
      "loss": 0.1848,
      "step": 96210
    },
    {
      "epoch": 288.95,
      "grad_norm": 6.585235118865967,
      "learning_rate": 3.6836836836836835e-07,
      "loss": 0.169,
      "step": 96220
    },
    {
      "epoch": 288.98,
      "grad_norm": 15.227516174316406,
      "learning_rate": 3.673673673673674e-07,
      "loss": 0.2181,
      "step": 96230
    },
    {
      "epoch": 289.0,
      "eval_accuracy": 0.9224,
      "eval_loss": 0.3511893153190613,
      "eval_runtime": 31.1067,
      "eval_samples_per_second": 321.474,
      "eval_steps_per_second": 1.286,
      "step": 96237
    },
    {
      "epoch": 289.01,
      "grad_norm": 10.13980484008789,
      "learning_rate": 3.6636636636636644e-07,
      "loss": 0.1471,
      "step": 96240
    },
    {
      "epoch": 289.04,
      "grad_norm": 6.072284698486328,
      "learning_rate": 3.653653653653654e-07,
      "loss": 0.2203,
      "step": 96250
    },
    {
      "epoch": 289.07,
      "grad_norm": 7.853864669799805,
      "learning_rate": 3.643643643643644e-07,
      "loss": 0.2001,
      "step": 96260
    },
    {
      "epoch": 289.1,
      "grad_norm": 7.392894744873047,
      "learning_rate": 3.633633633633634e-07,
      "loss": 0.1576,
      "step": 96270
    },
    {
      "epoch": 289.13,
      "grad_norm": 7.9130167961120605,
      "learning_rate": 3.623623623623624e-07,
      "loss": 0.176,
      "step": 96280
    },
    {
      "epoch": 289.16,
      "grad_norm": 7.225363731384277,
      "learning_rate": 3.6136136136136137e-07,
      "loss": 0.186,
      "step": 96290
    },
    {
      "epoch": 289.19,
      "grad_norm": 11.619985580444336,
      "learning_rate": 3.603603603603604e-07,
      "loss": 0.1604,
      "step": 96300
    },
    {
      "epoch": 289.22,
      "grad_norm": 11.890161514282227,
      "learning_rate": 3.593593593593594e-07,
      "loss": 0.1711,
      "step": 96310
    },
    {
      "epoch": 289.25,
      "grad_norm": 8.829264640808105,
      "learning_rate": 3.5835835835835837e-07,
      "loss": 0.2013,
      "step": 96320
    },
    {
      "epoch": 289.28,
      "grad_norm": 11.483244895935059,
      "learning_rate": 3.573573573573574e-07,
      "loss": 0.1848,
      "step": 96330
    },
    {
      "epoch": 289.31,
      "grad_norm": 27.52833366394043,
      "learning_rate": 3.5635635635635636e-07,
      "loss": 0.1814,
      "step": 96340
    },
    {
      "epoch": 289.34,
      "grad_norm": 7.8727498054504395,
      "learning_rate": 3.553553553553554e-07,
      "loss": 0.2152,
      "step": 96350
    },
    {
      "epoch": 289.37,
      "grad_norm": 14.143877983093262,
      "learning_rate": 3.5435435435435434e-07,
      "loss": 0.175,
      "step": 96360
    },
    {
      "epoch": 289.4,
      "grad_norm": 10.356393814086914,
      "learning_rate": 3.533533533533534e-07,
      "loss": 0.1887,
      "step": 96370
    },
    {
      "epoch": 289.43,
      "grad_norm": 9.41296672821045,
      "learning_rate": 3.5235235235235243e-07,
      "loss": 0.1692,
      "step": 96380
    },
    {
      "epoch": 289.46,
      "grad_norm": 9.513154029846191,
      "learning_rate": 3.513513513513514e-07,
      "loss": 0.1932,
      "step": 96390
    },
    {
      "epoch": 289.49,
      "grad_norm": 16.088510513305664,
      "learning_rate": 3.503503503503504e-07,
      "loss": 0.2376,
      "step": 96400
    },
    {
      "epoch": 289.52,
      "grad_norm": 7.964100360870361,
      "learning_rate": 3.493493493493494e-07,
      "loss": 0.175,
      "step": 96410
    },
    {
      "epoch": 289.55,
      "grad_norm": 7.604168891906738,
      "learning_rate": 3.483483483483484e-07,
      "loss": 0.1909,
      "step": 96420
    },
    {
      "epoch": 289.58,
      "grad_norm": 17.463308334350586,
      "learning_rate": 3.4734734734734736e-07,
      "loss": 0.1756,
      "step": 96430
    },
    {
      "epoch": 289.61,
      "grad_norm": 6.574659824371338,
      "learning_rate": 3.463463463463464e-07,
      "loss": 0.1608,
      "step": 96440
    },
    {
      "epoch": 289.64,
      "grad_norm": 6.821499824523926,
      "learning_rate": 3.4534534534534535e-07,
      "loss": 0.1699,
      "step": 96450
    },
    {
      "epoch": 289.67,
      "grad_norm": 7.894416809082031,
      "learning_rate": 3.4434434434434436e-07,
      "loss": 0.1982,
      "step": 96460
    },
    {
      "epoch": 289.7,
      "grad_norm": 9.052847862243652,
      "learning_rate": 3.4334334334334333e-07,
      "loss": 0.175,
      "step": 96470
    },
    {
      "epoch": 289.73,
      "grad_norm": 14.444825172424316,
      "learning_rate": 3.4234234234234235e-07,
      "loss": 0.1715,
      "step": 96480
    },
    {
      "epoch": 289.76,
      "grad_norm": 11.772759437561035,
      "learning_rate": 3.413413413413413e-07,
      "loss": 0.1611,
      "step": 96490
    },
    {
      "epoch": 289.79,
      "grad_norm": 10.97026252746582,
      "learning_rate": 3.403403403403404e-07,
      "loss": 0.2126,
      "step": 96500
    },
    {
      "epoch": 289.82,
      "grad_norm": 7.0093278884887695,
      "learning_rate": 3.393393393393394e-07,
      "loss": 0.1737,
      "step": 96510
    },
    {
      "epoch": 289.85,
      "grad_norm": 7.731503009796143,
      "learning_rate": 3.3833833833833837e-07,
      "loss": 0.1793,
      "step": 96520
    },
    {
      "epoch": 289.88,
      "grad_norm": 5.698312282562256,
      "learning_rate": 3.373373373373374e-07,
      "loss": 0.2143,
      "step": 96530
    },
    {
      "epoch": 289.91,
      "grad_norm": 10.125889778137207,
      "learning_rate": 3.3633633633633635e-07,
      "loss": 0.1629,
      "step": 96540
    },
    {
      "epoch": 289.94,
      "grad_norm": 9.121259689331055,
      "learning_rate": 3.3533533533533537e-07,
      "loss": 0.1743,
      "step": 96550
    },
    {
      "epoch": 289.97,
      "grad_norm": 20.07433319091797,
      "learning_rate": 3.343343343343344e-07,
      "loss": 0.1965,
      "step": 96560
    },
    {
      "epoch": 290.0,
      "grad_norm": 0.2971692979335785,
      "learning_rate": 3.3333333333333335e-07,
      "loss": 0.1637,
      "step": 96570
    },
    {
      "epoch": 290.0,
      "eval_accuracy": 0.9219,
      "eval_loss": 0.35135138034820557,
      "eval_runtime": 30.6205,
      "eval_samples_per_second": 326.579,
      "eval_steps_per_second": 1.306,
      "step": 96570
    },
    {
      "epoch": 290.03,
      "grad_norm": 7.412853717803955,
      "learning_rate": 3.3233233233233237e-07,
      "loss": 0.1607,
      "step": 96580
    },
    {
      "epoch": 290.06,
      "grad_norm": 12.164033889770508,
      "learning_rate": 3.3133133133133134e-07,
      "loss": 0.1815,
      "step": 96590
    },
    {
      "epoch": 290.09,
      "grad_norm": 7.839737892150879,
      "learning_rate": 3.3033033033033036e-07,
      "loss": 0.1512,
      "step": 96600
    },
    {
      "epoch": 290.12,
      "grad_norm": 8.431002616882324,
      "learning_rate": 3.293293293293293e-07,
      "loss": 0.1805,
      "step": 96610
    },
    {
      "epoch": 290.15,
      "grad_norm": 10.762585639953613,
      "learning_rate": 3.2832832832832834e-07,
      "loss": 0.2128,
      "step": 96620
    },
    {
      "epoch": 290.18,
      "grad_norm": 6.190657138824463,
      "learning_rate": 3.273273273273274e-07,
      "loss": 0.1565,
      "step": 96630
    },
    {
      "epoch": 290.21,
      "grad_norm": 7.108602523803711,
      "learning_rate": 3.263263263263264e-07,
      "loss": 0.1799,
      "step": 96640
    },
    {
      "epoch": 290.24,
      "grad_norm": 15.829144477844238,
      "learning_rate": 3.253253253253254e-07,
      "loss": 0.1927,
      "step": 96650
    },
    {
      "epoch": 290.27,
      "grad_norm": 12.627176284790039,
      "learning_rate": 3.2432432432432436e-07,
      "loss": 0.1703,
      "step": 96660
    },
    {
      "epoch": 290.3,
      "grad_norm": 10.91391658782959,
      "learning_rate": 3.233233233233234e-07,
      "loss": 0.1996,
      "step": 96670
    },
    {
      "epoch": 290.33,
      "grad_norm": 11.205720901489258,
      "learning_rate": 3.2232232232232234e-07,
      "loss": 0.1888,
      "step": 96680
    },
    {
      "epoch": 290.36,
      "grad_norm": 8.374791145324707,
      "learning_rate": 3.2132132132132136e-07,
      "loss": 0.1671,
      "step": 96690
    },
    {
      "epoch": 290.39,
      "grad_norm": 14.343439102172852,
      "learning_rate": 3.2032032032032033e-07,
      "loss": 0.2204,
      "step": 96700
    },
    {
      "epoch": 290.42,
      "grad_norm": 8.148462295532227,
      "learning_rate": 3.1931931931931935e-07,
      "loss": 0.1829,
      "step": 96710
    },
    {
      "epoch": 290.45,
      "grad_norm": 9.184608459472656,
      "learning_rate": 3.183183183183183e-07,
      "loss": 0.1899,
      "step": 96720
    },
    {
      "epoch": 290.48,
      "grad_norm": 6.719897747039795,
      "learning_rate": 3.1731731731731733e-07,
      "loss": 0.1988,
      "step": 96730
    },
    {
      "epoch": 290.51,
      "grad_norm": 7.558572769165039,
      "learning_rate": 3.163163163163163e-07,
      "loss": 0.2008,
      "step": 96740
    },
    {
      "epoch": 290.54,
      "grad_norm": 8.834803581237793,
      "learning_rate": 3.153153153153153e-07,
      "loss": 0.2121,
      "step": 96750
    },
    {
      "epoch": 290.57,
      "grad_norm": 10.367210388183594,
      "learning_rate": 3.1431431431431433e-07,
      "loss": 0.1623,
      "step": 96760
    },
    {
      "epoch": 290.6,
      "grad_norm": 10.69318675994873,
      "learning_rate": 3.1331331331331335e-07,
      "loss": 0.1848,
      "step": 96770
    },
    {
      "epoch": 290.63,
      "grad_norm": 5.263699054718018,
      "learning_rate": 3.123123123123123e-07,
      "loss": 0.1794,
      "step": 96780
    },
    {
      "epoch": 290.66,
      "grad_norm": 9.211396217346191,
      "learning_rate": 3.1131131131131133e-07,
      "loss": 0.191,
      "step": 96790
    },
    {
      "epoch": 290.69,
      "grad_norm": 12.554217338562012,
      "learning_rate": 3.1031031031031035e-07,
      "loss": 0.1698,
      "step": 96800
    },
    {
      "epoch": 290.72,
      "grad_norm": 10.365185737609863,
      "learning_rate": 3.0930930930930937e-07,
      "loss": 0.2082,
      "step": 96810
    },
    {
      "epoch": 290.75,
      "grad_norm": 10.725741386413574,
      "learning_rate": 3.0830830830830834e-07,
      "loss": 0.1811,
      "step": 96820
    },
    {
      "epoch": 290.78,
      "grad_norm": 7.682887077331543,
      "learning_rate": 3.0730730730730735e-07,
      "loss": 0.1717,
      "step": 96830
    },
    {
      "epoch": 290.81,
      "grad_norm": 7.560359477996826,
      "learning_rate": 3.063063063063063e-07,
      "loss": 0.202,
      "step": 96840
    },
    {
      "epoch": 290.84,
      "grad_norm": 12.413463592529297,
      "learning_rate": 3.0530530530530534e-07,
      "loss": 0.1605,
      "step": 96850
    },
    {
      "epoch": 290.87,
      "grad_norm": 10.084680557250977,
      "learning_rate": 3.043043043043043e-07,
      "loss": 0.2203,
      "step": 96860
    },
    {
      "epoch": 290.9,
      "grad_norm": 6.152895450592041,
      "learning_rate": 3.033033033033033e-07,
      "loss": 0.1779,
      "step": 96870
    },
    {
      "epoch": 290.93,
      "grad_norm": 11.01721477508545,
      "learning_rate": 3.0230230230230234e-07,
      "loss": 0.205,
      "step": 96880
    },
    {
      "epoch": 290.96,
      "grad_norm": 6.163355350494385,
      "learning_rate": 3.013013013013013e-07,
      "loss": 0.2078,
      "step": 96890
    },
    {
      "epoch": 290.99,
      "grad_norm": 8.1688232421875,
      "learning_rate": 3.003003003003003e-07,
      "loss": 0.1808,
      "step": 96900
    },
    {
      "epoch": 291.0,
      "eval_accuracy": 0.922,
      "eval_loss": 0.35111668705940247,
      "eval_runtime": 30.402,
      "eval_samples_per_second": 328.926,
      "eval_steps_per_second": 1.316,
      "step": 96903
    },
    {
      "epoch": 291.02,
      "grad_norm": 5.977980136871338,
      "learning_rate": 2.9929929929929934e-07,
      "loss": 0.1524,
      "step": 96910
    },
    {
      "epoch": 291.05,
      "grad_norm": 11.05988597869873,
      "learning_rate": 2.982982982982983e-07,
      "loss": 0.1836,
      "step": 96920
    },
    {
      "epoch": 291.08,
      "grad_norm": 14.774660110473633,
      "learning_rate": 2.972972972972973e-07,
      "loss": 0.1544,
      "step": 96930
    },
    {
      "epoch": 291.11,
      "grad_norm": 8.746810913085938,
      "learning_rate": 2.9629629629629634e-07,
      "loss": 0.1784,
      "step": 96940
    },
    {
      "epoch": 291.14,
      "grad_norm": 13.869183540344238,
      "learning_rate": 2.952952952952953e-07,
      "loss": 0.1614,
      "step": 96950
    },
    {
      "epoch": 291.17,
      "grad_norm": 11.246725082397461,
      "learning_rate": 2.9429429429429433e-07,
      "loss": 0.2052,
      "step": 96960
    },
    {
      "epoch": 291.2,
      "grad_norm": 17.561803817749023,
      "learning_rate": 2.932932932932933e-07,
      "loss": 0.1546,
      "step": 96970
    },
    {
      "epoch": 291.23,
      "grad_norm": 8.073896408081055,
      "learning_rate": 2.922922922922923e-07,
      "loss": 0.1398,
      "step": 96980
    },
    {
      "epoch": 291.26,
      "grad_norm": 12.199286460876465,
      "learning_rate": 2.912912912912913e-07,
      "loss": 0.2005,
      "step": 96990
    },
    {
      "epoch": 291.29,
      "grad_norm": 7.823231220245361,
      "learning_rate": 2.9029029029029035e-07,
      "loss": 0.176,
      "step": 97000
    },
    {
      "epoch": 291.32,
      "grad_norm": 8.55107307434082,
      "learning_rate": 2.892892892892893e-07,
      "loss": 0.1836,
      "step": 97010
    },
    {
      "epoch": 291.35,
      "grad_norm": 8.495357513427734,
      "learning_rate": 2.8828828828828833e-07,
      "loss": 0.2032,
      "step": 97020
    },
    {
      "epoch": 291.38,
      "grad_norm": 11.132481575012207,
      "learning_rate": 2.872872872872873e-07,
      "loss": 0.2218,
      "step": 97030
    },
    {
      "epoch": 291.41,
      "grad_norm": 10.816452026367188,
      "learning_rate": 2.862862862862863e-07,
      "loss": 0.1983,
      "step": 97040
    },
    {
      "epoch": 291.44,
      "grad_norm": 9.459407806396484,
      "learning_rate": 2.852852852852853e-07,
      "loss": 0.2287,
      "step": 97050
    },
    {
      "epoch": 291.47,
      "grad_norm": 11.077476501464844,
      "learning_rate": 2.842842842842843e-07,
      "loss": 0.1595,
      "step": 97060
    },
    {
      "epoch": 291.5,
      "grad_norm": 8.32515811920166,
      "learning_rate": 2.832832832832833e-07,
      "loss": 0.1435,
      "step": 97070
    },
    {
      "epoch": 291.53,
      "grad_norm": 6.450559616088867,
      "learning_rate": 2.8228228228228234e-07,
      "loss": 0.2461,
      "step": 97080
    },
    {
      "epoch": 291.56,
      "grad_norm": 6.475940704345703,
      "learning_rate": 2.812812812812813e-07,
      "loss": 0.1833,
      "step": 97090
    },
    {
      "epoch": 291.59,
      "grad_norm": 6.347726345062256,
      "learning_rate": 2.802802802802803e-07,
      "loss": 0.1709,
      "step": 97100
    },
    {
      "epoch": 291.62,
      "grad_norm": 5.686278343200684,
      "learning_rate": 2.792792792792793e-07,
      "loss": 0.1655,
      "step": 97110
    },
    {
      "epoch": 291.65,
      "grad_norm": 25.33323097229004,
      "learning_rate": 2.782782782782783e-07,
      "loss": 0.1724,
      "step": 97120
    },
    {
      "epoch": 291.68,
      "grad_norm": 4.689520835876465,
      "learning_rate": 2.772772772772773e-07,
      "loss": 0.1395,
      "step": 97130
    },
    {
      "epoch": 291.71,
      "grad_norm": 8.552033424377441,
      "learning_rate": 2.762762762762763e-07,
      "loss": 0.1814,
      "step": 97140
    },
    {
      "epoch": 291.74,
      "grad_norm": 7.146927356719971,
      "learning_rate": 2.752752752752753e-07,
      "loss": 0.1773,
      "step": 97150
    },
    {
      "epoch": 291.77,
      "grad_norm": 16.15311050415039,
      "learning_rate": 2.742742742742743e-07,
      "loss": 0.1592,
      "step": 97160
    },
    {
      "epoch": 291.8,
      "grad_norm": 8.083455085754395,
      "learning_rate": 2.732732732732733e-07,
      "loss": 0.168,
      "step": 97170
    },
    {
      "epoch": 291.83,
      "grad_norm": 9.019644737243652,
      "learning_rate": 2.722722722722723e-07,
      "loss": 0.2174,
      "step": 97180
    },
    {
      "epoch": 291.86,
      "grad_norm": 11.42094612121582,
      "learning_rate": 2.7127127127127127e-07,
      "loss": 0.1924,
      "step": 97190
    },
    {
      "epoch": 291.89,
      "grad_norm": 14.402693748474121,
      "learning_rate": 2.702702702702703e-07,
      "loss": 0.1882,
      "step": 97200
    },
    {
      "epoch": 291.92,
      "grad_norm": 11.41733169555664,
      "learning_rate": 2.692692692692693e-07,
      "loss": 0.1828,
      "step": 97210
    },
    {
      "epoch": 291.95,
      "grad_norm": 11.990534782409668,
      "learning_rate": 2.682682682682683e-07,
      "loss": 0.1427,
      "step": 97220
    },
    {
      "epoch": 291.98,
      "grad_norm": 7.1968302726745605,
      "learning_rate": 2.672672672672673e-07,
      "loss": 0.1935,
      "step": 97230
    },
    {
      "epoch": 292.0,
      "eval_accuracy": 0.9226,
      "eval_loss": 0.3511461615562439,
      "eval_runtime": 30.3985,
      "eval_samples_per_second": 328.964,
      "eval_steps_per_second": 1.316,
      "step": 97236
    },
    {
      "epoch": 292.01,
      "grad_norm": 9.064430236816406,
      "learning_rate": 2.6626626626626626e-07,
      "loss": 0.1434,
      "step": 97240
    },
    {
      "epoch": 292.04,
      "grad_norm": 5.879701614379883,
      "learning_rate": 2.652652652652653e-07,
      "loss": 0.166,
      "step": 97250
    },
    {
      "epoch": 292.07,
      "grad_norm": 15.763201713562012,
      "learning_rate": 2.642642642642643e-07,
      "loss": 0.2194,
      "step": 97260
    },
    {
      "epoch": 292.1,
      "grad_norm": 8.584965705871582,
      "learning_rate": 2.632632632632633e-07,
      "loss": 0.2013,
      "step": 97270
    },
    {
      "epoch": 292.13,
      "grad_norm": 7.146899223327637,
      "learning_rate": 2.622622622622623e-07,
      "loss": 0.2276,
      "step": 97280
    },
    {
      "epoch": 292.16,
      "grad_norm": 7.440613746643066,
      "learning_rate": 2.612612612612613e-07,
      "loss": 0.1638,
      "step": 97290
    },
    {
      "epoch": 292.19,
      "grad_norm": 9.060437202453613,
      "learning_rate": 2.6026026026026026e-07,
      "loss": 0.1791,
      "step": 97300
    },
    {
      "epoch": 292.22,
      "grad_norm": 10.524561882019043,
      "learning_rate": 2.592592592592593e-07,
      "loss": 0.2209,
      "step": 97310
    },
    {
      "epoch": 292.25,
      "grad_norm": 9.852057456970215,
      "learning_rate": 2.5825825825825825e-07,
      "loss": 0.1793,
      "step": 97320
    },
    {
      "epoch": 292.28,
      "grad_norm": 7.923940658569336,
      "learning_rate": 2.572572572572573e-07,
      "loss": 0.1715,
      "step": 97330
    },
    {
      "epoch": 292.31,
      "grad_norm": 49.20570755004883,
      "learning_rate": 2.562562562562563e-07,
      "loss": 0.2433,
      "step": 97340
    },
    {
      "epoch": 292.34,
      "grad_norm": 36.26921463012695,
      "learning_rate": 2.552552552552553e-07,
      "loss": 0.1601,
      "step": 97350
    },
    {
      "epoch": 292.37,
      "grad_norm": 14.222837448120117,
      "learning_rate": 2.5425425425425427e-07,
      "loss": 0.1937,
      "step": 97360
    },
    {
      "epoch": 292.4,
      "grad_norm": 10.002205848693848,
      "learning_rate": 2.532532532532533e-07,
      "loss": 0.1975,
      "step": 97370
    },
    {
      "epoch": 292.43,
      "grad_norm": 18.65520477294922,
      "learning_rate": 2.5225225225225225e-07,
      "loss": 0.1513,
      "step": 97380
    },
    {
      "epoch": 292.46,
      "grad_norm": 8.643986701965332,
      "learning_rate": 2.5125125125125127e-07,
      "loss": 0.2029,
      "step": 97390
    },
    {
      "epoch": 292.49,
      "grad_norm": 11.583308219909668,
      "learning_rate": 2.502502502502503e-07,
      "loss": 0.2218,
      "step": 97400
    },
    {
      "epoch": 292.52,
      "grad_norm": 9.699199676513672,
      "learning_rate": 2.492492492492493e-07,
      "loss": 0.1427,
      "step": 97410
    },
    {
      "epoch": 292.55,
      "grad_norm": 6.338335037231445,
      "learning_rate": 2.4824824824824827e-07,
      "loss": 0.2142,
      "step": 97420
    },
    {
      "epoch": 292.58,
      "grad_norm": 16.3291015625,
      "learning_rate": 2.472472472472473e-07,
      "loss": 0.1648,
      "step": 97430
    },
    {
      "epoch": 292.61,
      "grad_norm": 7.065120697021484,
      "learning_rate": 2.4624624624624625e-07,
      "loss": 0.15,
      "step": 97440
    },
    {
      "epoch": 292.64,
      "grad_norm": 9.734380722045898,
      "learning_rate": 2.4524524524524527e-07,
      "loss": 0.162,
      "step": 97450
    },
    {
      "epoch": 292.67,
      "grad_norm": 10.180914878845215,
      "learning_rate": 2.4424424424424424e-07,
      "loss": 0.1929,
      "step": 97460
    },
    {
      "epoch": 292.7,
      "grad_norm": 11.513069152832031,
      "learning_rate": 2.4324324324324326e-07,
      "loss": 0.1721,
      "step": 97470
    },
    {
      "epoch": 292.73,
      "grad_norm": 6.096655368804932,
      "learning_rate": 2.422422422422423e-07,
      "loss": 0.1659,
      "step": 97480
    },
    {
      "epoch": 292.76,
      "grad_norm": 9.24821949005127,
      "learning_rate": 2.4124124124124124e-07,
      "loss": 0.2022,
      "step": 97490
    },
    {
      "epoch": 292.79,
      "grad_norm": 37.50090026855469,
      "learning_rate": 2.4024024024024026e-07,
      "loss": 0.1418,
      "step": 97500
    },
    {
      "epoch": 292.82,
      "grad_norm": 35.785945892333984,
      "learning_rate": 2.392392392392393e-07,
      "loss": 0.178,
      "step": 97510
    },
    {
      "epoch": 292.85,
      "grad_norm": 5.393158912658691,
      "learning_rate": 2.3823823823823824e-07,
      "loss": 0.1409,
      "step": 97520
    },
    {
      "epoch": 292.88,
      "grad_norm": 7.336115837097168,
      "learning_rate": 2.3723723723723726e-07,
      "loss": 0.1487,
      "step": 97530
    },
    {
      "epoch": 292.91,
      "grad_norm": 6.544343948364258,
      "learning_rate": 2.3623623623623628e-07,
      "loss": 0.1754,
      "step": 97540
    },
    {
      "epoch": 292.94,
      "grad_norm": 7.259521484375,
      "learning_rate": 2.3523523523523527e-07,
      "loss": 0.2107,
      "step": 97550
    },
    {
      "epoch": 292.97,
      "grad_norm": 14.434693336486816,
      "learning_rate": 2.3423423423423426e-07,
      "loss": 0.2022,
      "step": 97560
    },
    {
      "epoch": 293.0,
      "eval_accuracy": 0.9229,
      "eval_loss": 0.35079532861709595,
      "eval_runtime": 29.8726,
      "eval_samples_per_second": 334.755,
      "eval_steps_per_second": 1.339,
      "step": 97569
    },
    {
      "epoch": 293.0,
      "grad_norm": 8.971282005310059,
      "learning_rate": 2.3323323323323325e-07,
      "loss": 0.1893,
      "step": 97570
    },
    {
      "epoch": 293.03,
      "grad_norm": 6.875269889831543,
      "learning_rate": 2.3223223223223225e-07,
      "loss": 0.1343,
      "step": 97580
    },
    {
      "epoch": 293.06,
      "grad_norm": 9.541326522827148,
      "learning_rate": 2.3123123123123124e-07,
      "loss": 0.1653,
      "step": 97590
    },
    {
      "epoch": 293.09,
      "grad_norm": 7.797970771789551,
      "learning_rate": 2.3023023023023026e-07,
      "loss": 0.2029,
      "step": 97600
    },
    {
      "epoch": 293.12,
      "grad_norm": 7.514450550079346,
      "learning_rate": 2.2922922922922925e-07,
      "loss": 0.1642,
      "step": 97610
    },
    {
      "epoch": 293.15,
      "grad_norm": 10.009089469909668,
      "learning_rate": 2.2822822822822824e-07,
      "loss": 0.1788,
      "step": 97620
    },
    {
      "epoch": 293.18,
      "grad_norm": 7.814962863922119,
      "learning_rate": 2.2722722722722726e-07,
      "loss": 0.2357,
      "step": 97630
    },
    {
      "epoch": 293.21,
      "grad_norm": 7.269006252288818,
      "learning_rate": 2.2622622622622625e-07,
      "loss": 0.1658,
      "step": 97640
    },
    {
      "epoch": 293.24,
      "grad_norm": 6.7736992835998535,
      "learning_rate": 2.2522522522522524e-07,
      "loss": 0.1757,
      "step": 97650
    },
    {
      "epoch": 293.27,
      "grad_norm": 18.101564407348633,
      "learning_rate": 2.2422422422422423e-07,
      "loss": 0.1984,
      "step": 97660
    },
    {
      "epoch": 293.3,
      "grad_norm": 12.212668418884277,
      "learning_rate": 2.2322322322322325e-07,
      "loss": 0.1828,
      "step": 97670
    },
    {
      "epoch": 293.33,
      "grad_norm": 13.234591484069824,
      "learning_rate": 2.2222222222222224e-07,
      "loss": 0.2032,
      "step": 97680
    },
    {
      "epoch": 293.36,
      "grad_norm": 7.669671535491943,
      "learning_rate": 2.2122122122122124e-07,
      "loss": 0.1664,
      "step": 97690
    },
    {
      "epoch": 293.39,
      "grad_norm": 8.292030334472656,
      "learning_rate": 2.2022022022022023e-07,
      "loss": 0.2137,
      "step": 97700
    },
    {
      "epoch": 293.42,
      "grad_norm": 6.4687628746032715,
      "learning_rate": 2.1921921921921922e-07,
      "loss": 0.1808,
      "step": 97710
    },
    {
      "epoch": 293.45,
      "grad_norm": 7.9614949226379395,
      "learning_rate": 2.182182182182182e-07,
      "loss": 0.2044,
      "step": 97720
    },
    {
      "epoch": 293.48,
      "grad_norm": 9.309539794921875,
      "learning_rate": 2.1721721721721726e-07,
      "loss": 0.2011,
      "step": 97730
    },
    {
      "epoch": 293.51,
      "grad_norm": 6.957324981689453,
      "learning_rate": 2.1621621621621625e-07,
      "loss": 0.1941,
      "step": 97740
    },
    {
      "epoch": 293.54,
      "grad_norm": 9.628689765930176,
      "learning_rate": 2.1521521521521524e-07,
      "loss": 0.1947,
      "step": 97750
    },
    {
      "epoch": 293.57,
      "grad_norm": 6.134148120880127,
      "learning_rate": 2.1421421421421423e-07,
      "loss": 0.1471,
      "step": 97760
    },
    {
      "epoch": 293.6,
      "grad_norm": 5.888867378234863,
      "learning_rate": 2.1321321321321322e-07,
      "loss": 0.143,
      "step": 97770
    },
    {
      "epoch": 293.63,
      "grad_norm": 5.631621360778809,
      "learning_rate": 2.1221221221221222e-07,
      "loss": 0.1745,
      "step": 97780
    },
    {
      "epoch": 293.66,
      "grad_norm": 6.445488929748535,
      "learning_rate": 2.112112112112112e-07,
      "loss": 0.1554,
      "step": 97790
    },
    {
      "epoch": 293.69,
      "grad_norm": 5.106081008911133,
      "learning_rate": 2.1021021021021025e-07,
      "loss": 0.1708,
      "step": 97800
    },
    {
      "epoch": 293.72,
      "grad_norm": 8.569723129272461,
      "learning_rate": 2.0920920920920924e-07,
      "loss": 0.1752,
      "step": 97810
    },
    {
      "epoch": 293.75,
      "grad_norm": 9.983181953430176,
      "learning_rate": 2.0820820820820824e-07,
      "loss": 0.1781,
      "step": 97820
    },
    {
      "epoch": 293.78,
      "grad_norm": 10.954108238220215,
      "learning_rate": 2.0720720720720723e-07,
      "loss": 0.2191,
      "step": 97830
    },
    {
      "epoch": 293.81,
      "grad_norm": 10.042722702026367,
      "learning_rate": 2.0620620620620622e-07,
      "loss": 0.1728,
      "step": 97840
    },
    {
      "epoch": 293.84,
      "grad_norm": 8.717355728149414,
      "learning_rate": 2.052052052052052e-07,
      "loss": 0.1968,
      "step": 97850
    },
    {
      "epoch": 293.87,
      "grad_norm": 5.418293476104736,
      "learning_rate": 2.042042042042042e-07,
      "loss": 0.1413,
      "step": 97860
    },
    {
      "epoch": 293.9,
      "grad_norm": 10.800609588623047,
      "learning_rate": 2.0320320320320322e-07,
      "loss": 0.1516,
      "step": 97870
    },
    {
      "epoch": 293.93,
      "grad_norm": 5.475895881652832,
      "learning_rate": 2.0220220220220224e-07,
      "loss": 0.1545,
      "step": 97880
    },
    {
      "epoch": 293.96,
      "grad_norm": 5.100027561187744,
      "learning_rate": 2.0120120120120123e-07,
      "loss": 0.1939,
      "step": 97890
    },
    {
      "epoch": 293.99,
      "grad_norm": 7.097851753234863,
      "learning_rate": 2.0020020020020022e-07,
      "loss": 0.1708,
      "step": 97900
    },
    {
      "epoch": 294.0,
      "eval_accuracy": 0.923,
      "eval_loss": 0.3504704236984253,
      "eval_runtime": 30.3658,
      "eval_samples_per_second": 329.317,
      "eval_steps_per_second": 1.317,
      "step": 97902
    },
    {
      "epoch": 294.02,
      "grad_norm": 6.2672200202941895,
      "learning_rate": 1.9919919919919922e-07,
      "loss": 0.1558,
      "step": 97910
    },
    {
      "epoch": 294.05,
      "grad_norm": 4.508206844329834,
      "learning_rate": 1.981981981981982e-07,
      "loss": 0.1759,
      "step": 97920
    },
    {
      "epoch": 294.08,
      "grad_norm": 8.369656562805176,
      "learning_rate": 1.971971971971972e-07,
      "loss": 0.1925,
      "step": 97930
    },
    {
      "epoch": 294.11,
      "grad_norm": 7.288471698760986,
      "learning_rate": 1.9619619619619622e-07,
      "loss": 0.1542,
      "step": 97940
    },
    {
      "epoch": 294.14,
      "grad_norm": 6.860898971557617,
      "learning_rate": 1.951951951951952e-07,
      "loss": 0.2128,
      "step": 97950
    },
    {
      "epoch": 294.17,
      "grad_norm": 7.634459495544434,
      "learning_rate": 1.941941941941942e-07,
      "loss": 0.1638,
      "step": 97960
    },
    {
      "epoch": 294.2,
      "grad_norm": 6.657994747161865,
      "learning_rate": 1.931931931931932e-07,
      "loss": 0.207,
      "step": 97970
    },
    {
      "epoch": 294.23,
      "grad_norm": 6.48544979095459,
      "learning_rate": 1.921921921921922e-07,
      "loss": 0.2003,
      "step": 97980
    },
    {
      "epoch": 294.26,
      "grad_norm": 9.943379402160645,
      "learning_rate": 1.911911911911912e-07,
      "loss": 0.1904,
      "step": 97990
    },
    {
      "epoch": 294.29,
      "grad_norm": 6.679036617279053,
      "learning_rate": 1.9019019019019022e-07,
      "loss": 0.1874,
      "step": 98000
    },
    {
      "epoch": 294.32,
      "grad_norm": 6.730116844177246,
      "learning_rate": 1.8918918918918921e-07,
      "loss": 0.1956,
      "step": 98010
    },
    {
      "epoch": 294.35,
      "grad_norm": 9.145742416381836,
      "learning_rate": 1.881881881881882e-07,
      "loss": 0.181,
      "step": 98020
    },
    {
      "epoch": 294.38,
      "grad_norm": 6.606112003326416,
      "learning_rate": 1.871871871871872e-07,
      "loss": 0.1723,
      "step": 98030
    },
    {
      "epoch": 294.41,
      "grad_norm": 36.06501007080078,
      "learning_rate": 1.861861861861862e-07,
      "loss": 0.2079,
      "step": 98040
    },
    {
      "epoch": 294.44,
      "grad_norm": 7.61054801940918,
      "learning_rate": 1.8518518518518518e-07,
      "loss": 0.1851,
      "step": 98050
    },
    {
      "epoch": 294.47,
      "grad_norm": 13.092728614807129,
      "learning_rate": 1.8418418418418417e-07,
      "loss": 0.1824,
      "step": 98060
    },
    {
      "epoch": 294.5,
      "grad_norm": 7.269385814666748,
      "learning_rate": 1.8318318318318322e-07,
      "loss": 0.167,
      "step": 98070
    },
    {
      "epoch": 294.53,
      "grad_norm": 7.955464839935303,
      "learning_rate": 1.821821821821822e-07,
      "loss": 0.1851,
      "step": 98080
    },
    {
      "epoch": 294.56,
      "grad_norm": 7.4256696701049805,
      "learning_rate": 1.811811811811812e-07,
      "loss": 0.1914,
      "step": 98090
    },
    {
      "epoch": 294.59,
      "grad_norm": 5.502331733703613,
      "learning_rate": 1.801801801801802e-07,
      "loss": 0.1585,
      "step": 98100
    },
    {
      "epoch": 294.62,
      "grad_norm": 9.890101432800293,
      "learning_rate": 1.7917917917917919e-07,
      "loss": 0.1973,
      "step": 98110
    },
    {
      "epoch": 294.65,
      "grad_norm": 10.878011703491211,
      "learning_rate": 1.7817817817817818e-07,
      "loss": 0.1793,
      "step": 98120
    },
    {
      "epoch": 294.68,
      "grad_norm": 13.217385292053223,
      "learning_rate": 1.7717717717717717e-07,
      "loss": 0.176,
      "step": 98130
    },
    {
      "epoch": 294.71,
      "grad_norm": 6.875507354736328,
      "learning_rate": 1.7617617617617621e-07,
      "loss": 0.2003,
      "step": 98140
    },
    {
      "epoch": 294.74,
      "grad_norm": 8.538390159606934,
      "learning_rate": 1.751751751751752e-07,
      "loss": 0.1934,
      "step": 98150
    },
    {
      "epoch": 294.77,
      "grad_norm": 13.891023635864258,
      "learning_rate": 1.741741741741742e-07,
      "loss": 0.1626,
      "step": 98160
    },
    {
      "epoch": 294.8,
      "grad_norm": 5.604267120361328,
      "learning_rate": 1.731731731731732e-07,
      "loss": 0.167,
      "step": 98170
    },
    {
      "epoch": 294.83,
      "grad_norm": 9.339226722717285,
      "learning_rate": 1.7217217217217218e-07,
      "loss": 0.2151,
      "step": 98180
    },
    {
      "epoch": 294.86,
      "grad_norm": 6.294442176818848,
      "learning_rate": 1.7117117117117117e-07,
      "loss": 0.1653,
      "step": 98190
    },
    {
      "epoch": 294.89,
      "grad_norm": 8.720332145690918,
      "learning_rate": 1.701701701701702e-07,
      "loss": 0.208,
      "step": 98200
    },
    {
      "epoch": 294.92,
      "grad_norm": 10.39927864074707,
      "learning_rate": 1.6916916916916918e-07,
      "loss": 0.2016,
      "step": 98210
    },
    {
      "epoch": 294.95,
      "grad_norm": 10.204832077026367,
      "learning_rate": 1.6816816816816818e-07,
      "loss": 0.1741,
      "step": 98220
    },
    {
      "epoch": 294.98,
      "grad_norm": 11.65427017211914,
      "learning_rate": 1.671671671671672e-07,
      "loss": 0.1924,
      "step": 98230
    },
    {
      "epoch": 295.0,
      "eval_accuracy": 0.9228,
      "eval_loss": 0.35080063343048096,
      "eval_runtime": 30.3794,
      "eval_samples_per_second": 329.171,
      "eval_steps_per_second": 1.317,
      "step": 98235
    },
    {
      "epoch": 295.02,
      "grad_norm": 12.273945808410645,
      "learning_rate": 1.6616616616616619e-07,
      "loss": 0.1823,
      "step": 98240
    },
    {
      "epoch": 295.05,
      "grad_norm": 4.834070205688477,
      "learning_rate": 1.6516516516516518e-07,
      "loss": 0.1581,
      "step": 98250
    },
    {
      "epoch": 295.08,
      "grad_norm": 10.939055442810059,
      "learning_rate": 1.6416416416416417e-07,
      "loss": 0.1913,
      "step": 98260
    },
    {
      "epoch": 295.11,
      "grad_norm": 5.226729393005371,
      "learning_rate": 1.631631631631632e-07,
      "loss": 0.2221,
      "step": 98270
    },
    {
      "epoch": 295.14,
      "grad_norm": 8.952728271484375,
      "learning_rate": 1.6216216216216218e-07,
      "loss": 0.1647,
      "step": 98280
    },
    {
      "epoch": 295.17,
      "grad_norm": 9.182406425476074,
      "learning_rate": 1.6116116116116117e-07,
      "loss": 0.2123,
      "step": 98290
    },
    {
      "epoch": 295.2,
      "grad_norm": 9.189105033874512,
      "learning_rate": 1.6016016016016016e-07,
      "loss": 0.1865,
      "step": 98300
    },
    {
      "epoch": 295.23,
      "grad_norm": 8.874194145202637,
      "learning_rate": 1.5915915915915916e-07,
      "loss": 0.1884,
      "step": 98310
    },
    {
      "epoch": 295.26,
      "grad_norm": 13.395391464233398,
      "learning_rate": 1.5815815815815815e-07,
      "loss": 0.1683,
      "step": 98320
    },
    {
      "epoch": 295.29,
      "grad_norm": 9.559453010559082,
      "learning_rate": 1.5715715715715717e-07,
      "loss": 0.1748,
      "step": 98330
    },
    {
      "epoch": 295.32,
      "grad_norm": 25.271223068237305,
      "learning_rate": 1.5615615615615616e-07,
      "loss": 0.1749,
      "step": 98340
    },
    {
      "epoch": 295.35,
      "grad_norm": 11.601675987243652,
      "learning_rate": 1.5515515515515518e-07,
      "loss": 0.2081,
      "step": 98350
    },
    {
      "epoch": 295.38,
      "grad_norm": 8.208320617675781,
      "learning_rate": 1.5415415415415417e-07,
      "loss": 0.1588,
      "step": 98360
    },
    {
      "epoch": 295.41,
      "grad_norm": 8.67174243927002,
      "learning_rate": 1.5315315315315316e-07,
      "loss": 0.1991,
      "step": 98370
    },
    {
      "epoch": 295.44,
      "grad_norm": 6.374279975891113,
      "learning_rate": 1.5215215215215215e-07,
      "loss": 0.1695,
      "step": 98380
    },
    {
      "epoch": 295.47,
      "grad_norm": 8.36281681060791,
      "learning_rate": 1.5115115115115117e-07,
      "loss": 0.1665,
      "step": 98390
    },
    {
      "epoch": 295.5,
      "grad_norm": 5.865175724029541,
      "learning_rate": 1.5015015015015016e-07,
      "loss": 0.2106,
      "step": 98400
    },
    {
      "epoch": 295.53,
      "grad_norm": 7.013520240783691,
      "learning_rate": 1.4914914914914915e-07,
      "loss": 0.225,
      "step": 98410
    },
    {
      "epoch": 295.56,
      "grad_norm": 8.761256217956543,
      "learning_rate": 1.4814814814814817e-07,
      "loss": 0.1917,
      "step": 98420
    },
    {
      "epoch": 295.59,
      "grad_norm": 13.013578414916992,
      "learning_rate": 1.4714714714714716e-07,
      "loss": 0.1899,
      "step": 98430
    },
    {
      "epoch": 295.62,
      "grad_norm": 11.213883399963379,
      "learning_rate": 1.4614614614614616e-07,
      "loss": 0.1497,
      "step": 98440
    },
    {
      "epoch": 295.65,
      "grad_norm": 7.90604305267334,
      "learning_rate": 1.4514514514514517e-07,
      "loss": 0.1771,
      "step": 98450
    },
    {
      "epoch": 295.68,
      "grad_norm": 5.583929061889648,
      "learning_rate": 1.4414414414414417e-07,
      "loss": 0.1495,
      "step": 98460
    },
    {
      "epoch": 295.71,
      "grad_norm": 5.479854106903076,
      "learning_rate": 1.4314314314314316e-07,
      "loss": 0.1834,
      "step": 98470
    },
    {
      "epoch": 295.74,
      "grad_norm": 5.300133228302002,
      "learning_rate": 1.4214214214214215e-07,
      "loss": 0.2148,
      "step": 98480
    },
    {
      "epoch": 295.77,
      "grad_norm": 10.122681617736816,
      "learning_rate": 1.4114114114114117e-07,
      "loss": 0.2199,
      "step": 98490
    },
    {
      "epoch": 295.8,
      "grad_norm": 7.309352397918701,
      "learning_rate": 1.4014014014014016e-07,
      "loss": 0.165,
      "step": 98500
    },
    {
      "epoch": 295.83,
      "grad_norm": 8.675699234008789,
      "learning_rate": 1.3913913913913915e-07,
      "loss": 0.2153,
      "step": 98510
    },
    {
      "epoch": 295.86,
      "grad_norm": 8.666951179504395,
      "learning_rate": 1.3813813813813814e-07,
      "loss": 0.1876,
      "step": 98520
    },
    {
      "epoch": 295.89,
      "grad_norm": 12.89278507232666,
      "learning_rate": 1.3713713713713716e-07,
      "loss": 0.1931,
      "step": 98530
    },
    {
      "epoch": 295.92,
      "grad_norm": 7.896059513092041,
      "learning_rate": 1.3613613613613615e-07,
      "loss": 0.1697,
      "step": 98540
    },
    {
      "epoch": 295.95,
      "grad_norm": 8.814970970153809,
      "learning_rate": 1.3513513513513515e-07,
      "loss": 0.1648,
      "step": 98550
    },
    {
      "epoch": 295.98,
      "grad_norm": 7.2696003913879395,
      "learning_rate": 1.3413413413413414e-07,
      "loss": 0.1775,
      "step": 98560
    },
    {
      "epoch": 296.0,
      "eval_accuracy": 0.9224,
      "eval_loss": 0.35109782218933105,
      "eval_runtime": 30.4731,
      "eval_samples_per_second": 328.158,
      "eval_steps_per_second": 1.313,
      "step": 98568
    },
    {
      "epoch": 296.01,
      "grad_norm": 3.892879009246826,
      "learning_rate": 1.3313313313313313e-07,
      "loss": 0.1199,
      "step": 98570
    },
    {
      "epoch": 296.04,
      "grad_norm": 11.924907684326172,
      "learning_rate": 1.3213213213213215e-07,
      "loss": 0.1555,
      "step": 98580
    },
    {
      "epoch": 296.07,
      "grad_norm": 5.635590076446533,
      "learning_rate": 1.3113113113113114e-07,
      "loss": 0.189,
      "step": 98590
    },
    {
      "epoch": 296.1,
      "grad_norm": 17.37589454650879,
      "learning_rate": 1.3013013013013013e-07,
      "loss": 0.2302,
      "step": 98600
    },
    {
      "epoch": 296.13,
      "grad_norm": 7.483663558959961,
      "learning_rate": 1.2912912912912912e-07,
      "loss": 0.1709,
      "step": 98610
    },
    {
      "epoch": 296.16,
      "grad_norm": 13.686518669128418,
      "learning_rate": 1.2812812812812814e-07,
      "loss": 0.1643,
      "step": 98620
    },
    {
      "epoch": 296.19,
      "grad_norm": 10.307132720947266,
      "learning_rate": 1.2712712712712713e-07,
      "loss": 0.2131,
      "step": 98630
    },
    {
      "epoch": 296.22,
      "grad_norm": 6.333912372589111,
      "learning_rate": 1.2612612612612613e-07,
      "loss": 0.1562,
      "step": 98640
    },
    {
      "epoch": 296.25,
      "grad_norm": 9.832839965820312,
      "learning_rate": 1.2512512512512514e-07,
      "loss": 0.2035,
      "step": 98650
    },
    {
      "epoch": 296.28,
      "grad_norm": 9.111720085144043,
      "learning_rate": 1.2412412412412414e-07,
      "loss": 0.1929,
      "step": 98660
    },
    {
      "epoch": 296.31,
      "grad_norm": 9.260516166687012,
      "learning_rate": 1.2312312312312313e-07,
      "loss": 0.1604,
      "step": 98670
    },
    {
      "epoch": 296.34,
      "grad_norm": 7.790345668792725,
      "learning_rate": 1.2212212212212212e-07,
      "loss": 0.2144,
      "step": 98680
    },
    {
      "epoch": 296.37,
      "grad_norm": 7.565441608428955,
      "learning_rate": 1.2112112112112114e-07,
      "loss": 0.1832,
      "step": 98690
    },
    {
      "epoch": 296.4,
      "grad_norm": 9.528629302978516,
      "learning_rate": 1.2012012012012013e-07,
      "loss": 0.1946,
      "step": 98700
    },
    {
      "epoch": 296.43,
      "grad_norm": 8.066445350646973,
      "learning_rate": 1.1911911911911912e-07,
      "loss": 0.1656,
      "step": 98710
    },
    {
      "epoch": 296.46,
      "grad_norm": 6.814592361450195,
      "learning_rate": 1.1811811811811814e-07,
      "loss": 0.1912,
      "step": 98720
    },
    {
      "epoch": 296.49,
      "grad_norm": 9.947332382202148,
      "learning_rate": 1.1711711711711713e-07,
      "loss": 0.192,
      "step": 98730
    },
    {
      "epoch": 296.52,
      "grad_norm": 6.69881534576416,
      "learning_rate": 1.1611611611611612e-07,
      "loss": 0.2026,
      "step": 98740
    },
    {
      "epoch": 296.55,
      "grad_norm": 8.709168434143066,
      "learning_rate": 1.1511511511511513e-07,
      "loss": 0.1932,
      "step": 98750
    },
    {
      "epoch": 296.58,
      "grad_norm": 8.136392593383789,
      "learning_rate": 1.1411411411411412e-07,
      "loss": 0.1902,
      "step": 98760
    },
    {
      "epoch": 296.61,
      "grad_norm": 10.547334671020508,
      "learning_rate": 1.1311311311311313e-07,
      "loss": 0.2135,
      "step": 98770
    },
    {
      "epoch": 296.64,
      "grad_norm": 9.078971862792969,
      "learning_rate": 1.1211211211211212e-07,
      "loss": 0.1708,
      "step": 98780
    },
    {
      "epoch": 296.67,
      "grad_norm": 5.5261006355285645,
      "learning_rate": 1.1111111111111112e-07,
      "loss": 0.195,
      "step": 98790
    },
    {
      "epoch": 296.7,
      "grad_norm": 8.106707572937012,
      "learning_rate": 1.1011011011011011e-07,
      "loss": 0.1722,
      "step": 98800
    },
    {
      "epoch": 296.73,
      "grad_norm": 6.8318867683410645,
      "learning_rate": 1.091091091091091e-07,
      "loss": 0.2205,
      "step": 98810
    },
    {
      "epoch": 296.76,
      "grad_norm": 13.725451469421387,
      "learning_rate": 1.0810810810810812e-07,
      "loss": 0.1891,
      "step": 98820
    },
    {
      "epoch": 296.79,
      "grad_norm": 9.009037971496582,
      "learning_rate": 1.0710710710710712e-07,
      "loss": 0.2022,
      "step": 98830
    },
    {
      "epoch": 296.82,
      "grad_norm": 9.115731239318848,
      "learning_rate": 1.0610610610610611e-07,
      "loss": 0.1803,
      "step": 98840
    },
    {
      "epoch": 296.85,
      "grad_norm": 7.076330661773682,
      "learning_rate": 1.0510510510510513e-07,
      "loss": 0.1738,
      "step": 98850
    },
    {
      "epoch": 296.88,
      "grad_norm": 9.310401916503906,
      "learning_rate": 1.0410410410410412e-07,
      "loss": 0.2016,
      "step": 98860
    },
    {
      "epoch": 296.91,
      "grad_norm": 11.235692977905273,
      "learning_rate": 1.0310310310310311e-07,
      "loss": 0.178,
      "step": 98870
    },
    {
      "epoch": 296.94,
      "grad_norm": 6.138362407684326,
      "learning_rate": 1.021021021021021e-07,
      "loss": 0.2022,
      "step": 98880
    },
    {
      "epoch": 296.97,
      "grad_norm": 7.528388500213623,
      "learning_rate": 1.0110110110110112e-07,
      "loss": 0.1637,
      "step": 98890
    },
    {
      "epoch": 297.0,
      "grad_norm": 5.268246650695801,
      "learning_rate": 1.0010010010010011e-07,
      "loss": 0.1681,
      "step": 98900
    },
    {
      "epoch": 297.0,
      "eval_accuracy": 0.9224,
      "eval_loss": 0.35087093710899353,
      "eval_runtime": 30.5029,
      "eval_samples_per_second": 327.838,
      "eval_steps_per_second": 1.311,
      "step": 98901
    },
    {
      "epoch": 297.03,
      "grad_norm": 10.673667907714844,
      "learning_rate": 9.90990990990991e-08,
      "loss": 0.1669,
      "step": 98910
    },
    {
      "epoch": 297.06,
      "grad_norm": 14.31691837310791,
      "learning_rate": 9.809809809809811e-08,
      "loss": 0.1577,
      "step": 98920
    },
    {
      "epoch": 297.09,
      "grad_norm": 19.130691528320312,
      "learning_rate": 9.70970970970971e-08,
      "loss": 0.183,
      "step": 98930
    },
    {
      "epoch": 297.12,
      "grad_norm": 13.734710693359375,
      "learning_rate": 9.60960960960961e-08,
      "loss": 0.201,
      "step": 98940
    },
    {
      "epoch": 297.15,
      "grad_norm": 8.000900268554688,
      "learning_rate": 9.509509509509511e-08,
      "loss": 0.2076,
      "step": 98950
    },
    {
      "epoch": 297.18,
      "grad_norm": 4.595077037811279,
      "learning_rate": 9.40940940940941e-08,
      "loss": 0.1291,
      "step": 98960
    },
    {
      "epoch": 297.21,
      "grad_norm": 8.093796730041504,
      "learning_rate": 9.30930930930931e-08,
      "loss": 0.2142,
      "step": 98970
    },
    {
      "epoch": 297.24,
      "grad_norm": 14.136248588562012,
      "learning_rate": 9.209209209209209e-08,
      "loss": 0.2075,
      "step": 98980
    },
    {
      "epoch": 297.27,
      "grad_norm": 9.155271530151367,
      "learning_rate": 9.10910910910911e-08,
      "loss": 0.2141,
      "step": 98990
    },
    {
      "epoch": 297.3,
      "grad_norm": 21.208436965942383,
      "learning_rate": 9.00900900900901e-08,
      "loss": 0.1961,
      "step": 99000
    },
    {
      "epoch": 297.33,
      "grad_norm": 9.15889835357666,
      "learning_rate": 8.908908908908909e-08,
      "loss": 0.2042,
      "step": 99010
    },
    {
      "epoch": 297.36,
      "grad_norm": 6.672938823699951,
      "learning_rate": 8.808808808808811e-08,
      "loss": 0.2047,
      "step": 99020
    },
    {
      "epoch": 297.39,
      "grad_norm": 8.8895263671875,
      "learning_rate": 8.70870870870871e-08,
      "loss": 0.1811,
      "step": 99030
    },
    {
      "epoch": 297.42,
      "grad_norm": 11.741869926452637,
      "learning_rate": 8.608608608608609e-08,
      "loss": 0.18,
      "step": 99040
    },
    {
      "epoch": 297.45,
      "grad_norm": 6.338979244232178,
      "learning_rate": 8.50850850850851e-08,
      "loss": 0.1576,
      "step": 99050
    },
    {
      "epoch": 297.48,
      "grad_norm": 7.054773330688477,
      "learning_rate": 8.408408408408409e-08,
      "loss": 0.2297,
      "step": 99060
    },
    {
      "epoch": 297.51,
      "grad_norm": 10.53976058959961,
      "learning_rate": 8.308308308308309e-08,
      "loss": 0.1948,
      "step": 99070
    },
    {
      "epoch": 297.54,
      "grad_norm": 9.608221054077148,
      "learning_rate": 8.208208208208208e-08,
      "loss": 0.2072,
      "step": 99080
    },
    {
      "epoch": 297.57,
      "grad_norm": 7.05378532409668,
      "learning_rate": 8.108108108108109e-08,
      "loss": 0.2631,
      "step": 99090
    },
    {
      "epoch": 297.6,
      "grad_norm": 5.904657363891602,
      "learning_rate": 8.008008008008008e-08,
      "loss": 0.1809,
      "step": 99100
    },
    {
      "epoch": 297.63,
      "grad_norm": 8.502961158752441,
      "learning_rate": 7.907907907907907e-08,
      "loss": 0.2032,
      "step": 99110
    },
    {
      "epoch": 297.66,
      "grad_norm": 9.886051177978516,
      "learning_rate": 7.807807807807808e-08,
      "loss": 0.1649,
      "step": 99120
    },
    {
      "epoch": 297.69,
      "grad_norm": 6.52326774597168,
      "learning_rate": 7.707707707707708e-08,
      "loss": 0.1575,
      "step": 99130
    },
    {
      "epoch": 297.72,
      "grad_norm": 6.63300085067749,
      "learning_rate": 7.607607607607608e-08,
      "loss": 0.1946,
      "step": 99140
    },
    {
      "epoch": 297.75,
      "grad_norm": 9.166041374206543,
      "learning_rate": 7.507507507507508e-08,
      "loss": 0.1885,
      "step": 99150
    },
    {
      "epoch": 297.78,
      "grad_norm": 9.458295822143555,
      "learning_rate": 7.407407407407409e-08,
      "loss": 0.1402,
      "step": 99160
    },
    {
      "epoch": 297.81,
      "grad_norm": 7.54589319229126,
      "learning_rate": 7.307307307307308e-08,
      "loss": 0.1835,
      "step": 99170
    },
    {
      "epoch": 297.84,
      "grad_norm": 8.338346481323242,
      "learning_rate": 7.207207207207208e-08,
      "loss": 0.1755,
      "step": 99180
    },
    {
      "epoch": 297.87,
      "grad_norm": 11.290776252746582,
      "learning_rate": 7.107107107107107e-08,
      "loss": 0.1801,
      "step": 99190
    },
    {
      "epoch": 297.9,
      "grad_norm": 7.982661247253418,
      "learning_rate": 7.007007007007008e-08,
      "loss": 0.2064,
      "step": 99200
    },
    {
      "epoch": 297.93,
      "grad_norm": 12.335919380187988,
      "learning_rate": 6.906906906906907e-08,
      "loss": 0.2286,
      "step": 99210
    },
    {
      "epoch": 297.96,
      "grad_norm": 3.7769718170166016,
      "learning_rate": 6.806806806806808e-08,
      "loss": 0.1911,
      "step": 99220
    },
    {
      "epoch": 297.99,
      "grad_norm": 6.068160533905029,
      "learning_rate": 6.706706706706707e-08,
      "loss": 0.1528,
      "step": 99230
    },
    {
      "epoch": 298.0,
      "eval_accuracy": 0.9225,
      "eval_loss": 0.35042741894721985,
      "eval_runtime": 30.0603,
      "eval_samples_per_second": 332.665,
      "eval_steps_per_second": 1.331,
      "step": 99234
    },
    {
      "epoch": 298.02,
      "grad_norm": 30.941457748413086,
      "learning_rate": 6.606606606606607e-08,
      "loss": 0.2084,
      "step": 99240
    },
    {
      "epoch": 298.05,
      "grad_norm": 4.703148365020752,
      "learning_rate": 6.506506506506507e-08,
      "loss": 0.1806,
      "step": 99250
    },
    {
      "epoch": 298.08,
      "grad_norm": 7.258764743804932,
      "learning_rate": 6.406406406406407e-08,
      "loss": 0.1656,
      "step": 99260
    },
    {
      "epoch": 298.11,
      "grad_norm": 7.957944393157959,
      "learning_rate": 6.306306306306306e-08,
      "loss": 0.1701,
      "step": 99270
    },
    {
      "epoch": 298.14,
      "grad_norm": 6.449383735656738,
      "learning_rate": 6.206206206206207e-08,
      "loss": 0.194,
      "step": 99280
    },
    {
      "epoch": 298.17,
      "grad_norm": 6.718631267547607,
      "learning_rate": 6.106106106106106e-08,
      "loss": 0.1702,
      "step": 99290
    },
    {
      "epoch": 298.2,
      "grad_norm": 13.296717643737793,
      "learning_rate": 6.006006006006006e-08,
      "loss": 0.1759,
      "step": 99300
    },
    {
      "epoch": 298.23,
      "grad_norm": 5.742093563079834,
      "learning_rate": 5.905905905905907e-08,
      "loss": 0.157,
      "step": 99310
    },
    {
      "epoch": 298.26,
      "grad_norm": 9.018274307250977,
      "learning_rate": 5.805805805805806e-08,
      "loss": 0.1771,
      "step": 99320
    },
    {
      "epoch": 298.29,
      "grad_norm": 11.336094856262207,
      "learning_rate": 5.705705705705706e-08,
      "loss": 0.1603,
      "step": 99330
    },
    {
      "epoch": 298.32,
      "grad_norm": 5.559664249420166,
      "learning_rate": 5.605605605605606e-08,
      "loss": 0.1518,
      "step": 99340
    },
    {
      "epoch": 298.35,
      "grad_norm": 10.356619834899902,
      "learning_rate": 5.505505505505506e-08,
      "loss": 0.2035,
      "step": 99350
    },
    {
      "epoch": 298.38,
      "grad_norm": 9.011303901672363,
      "learning_rate": 5.405405405405406e-08,
      "loss": 0.1515,
      "step": 99360
    },
    {
      "epoch": 298.41,
      "grad_norm": 4.541053771972656,
      "learning_rate": 5.3053053053053054e-08,
      "loss": 0.15,
      "step": 99370
    },
    {
      "epoch": 298.44,
      "grad_norm": 8.873855590820312,
      "learning_rate": 5.205205205205206e-08,
      "loss": 0.1747,
      "step": 99380
    },
    {
      "epoch": 298.47,
      "grad_norm": 10.648369789123535,
      "learning_rate": 5.105105105105105e-08,
      "loss": 0.1409,
      "step": 99390
    },
    {
      "epoch": 298.5,
      "grad_norm": 5.4094438552856445,
      "learning_rate": 5.0050050050050056e-08,
      "loss": 0.2026,
      "step": 99400
    },
    {
      "epoch": 298.53,
      "grad_norm": 7.279516696929932,
      "learning_rate": 4.9049049049049055e-08,
      "loss": 0.2083,
      "step": 99410
    },
    {
      "epoch": 298.56,
      "grad_norm": 6.239774703979492,
      "learning_rate": 4.804804804804805e-08,
      "loss": 0.1633,
      "step": 99420
    },
    {
      "epoch": 298.59,
      "grad_norm": 6.335087776184082,
      "learning_rate": 4.704704704704705e-08,
      "loss": 0.1384,
      "step": 99430
    },
    {
      "epoch": 298.62,
      "grad_norm": 3.842104434967041,
      "learning_rate": 4.6046046046046043e-08,
      "loss": 0.1659,
      "step": 99440
    },
    {
      "epoch": 298.65,
      "grad_norm": 6.795367240905762,
      "learning_rate": 4.504504504504505e-08,
      "loss": 0.2155,
      "step": 99450
    },
    {
      "epoch": 298.68,
      "grad_norm": 6.838936805725098,
      "learning_rate": 4.4044044044044054e-08,
      "loss": 0.1641,
      "step": 99460
    },
    {
      "epoch": 298.71,
      "grad_norm": 8.6826171875,
      "learning_rate": 4.3043043043043045e-08,
      "loss": 0.2105,
      "step": 99470
    },
    {
      "epoch": 298.74,
      "grad_norm": 8.692898750305176,
      "learning_rate": 4.2042042042042044e-08,
      "loss": 0.1531,
      "step": 99480
    },
    {
      "epoch": 298.77,
      "grad_norm": 6.979349136352539,
      "learning_rate": 4.104104104104104e-08,
      "loss": 0.1454,
      "step": 99490
    },
    {
      "epoch": 298.8,
      "grad_norm": 18.498456954956055,
      "learning_rate": 4.004004004004004e-08,
      "loss": 0.2167,
      "step": 99500
    },
    {
      "epoch": 298.83,
      "grad_norm": 19.582740783691406,
      "learning_rate": 3.903903903903904e-08,
      "loss": 0.2057,
      "step": 99510
    },
    {
      "epoch": 298.86,
      "grad_norm": 13.500566482543945,
      "learning_rate": 3.803803803803804e-08,
      "loss": 0.1909,
      "step": 99520
    },
    {
      "epoch": 298.89,
      "grad_norm": 7.102029323577881,
      "learning_rate": 3.703703703703704e-08,
      "loss": 0.1847,
      "step": 99530
    },
    {
      "epoch": 298.92,
      "grad_norm": 9.519046783447266,
      "learning_rate": 3.603603603603604e-08,
      "loss": 0.1545,
      "step": 99540
    },
    {
      "epoch": 298.95,
      "grad_norm": 7.066282272338867,
      "learning_rate": 3.503503503503504e-08,
      "loss": 0.1696,
      "step": 99550
    },
    {
      "epoch": 298.98,
      "grad_norm": 12.463377952575684,
      "learning_rate": 3.403403403403404e-08,
      "loss": 0.1978,
      "step": 99560
    },
    {
      "epoch": 299.0,
      "eval_accuracy": 0.9226,
      "eval_loss": 0.3503252863883972,
      "eval_runtime": 30.5611,
      "eval_samples_per_second": 327.214,
      "eval_steps_per_second": 1.309,
      "step": 99567
    },
    {
      "epoch": 299.01,
      "grad_norm": 14.752592086791992,
      "learning_rate": 3.303303303303304e-08,
      "loss": 0.166,
      "step": 99570
    },
    {
      "epoch": 299.04,
      "grad_norm": 57.44609451293945,
      "learning_rate": 3.2032032032032035e-08,
      "loss": 0.2107,
      "step": 99580
    },
    {
      "epoch": 299.07,
      "grad_norm": 11.633469581604004,
      "learning_rate": 3.1031031031031034e-08,
      "loss": 0.1597,
      "step": 99590
    },
    {
      "epoch": 299.1,
      "grad_norm": 4.120083332061768,
      "learning_rate": 3.003003003003003e-08,
      "loss": 0.1694,
      "step": 99600
    },
    {
      "epoch": 299.13,
      "grad_norm": 9.033282279968262,
      "learning_rate": 2.902902902902903e-08,
      "loss": 0.1701,
      "step": 99610
    },
    {
      "epoch": 299.16,
      "grad_norm": 5.067450523376465,
      "learning_rate": 2.802802802802803e-08,
      "loss": 0.1783,
      "step": 99620
    },
    {
      "epoch": 299.19,
      "grad_norm": 7.4798054695129395,
      "learning_rate": 2.702702702702703e-08,
      "loss": 0.1501,
      "step": 99630
    },
    {
      "epoch": 299.22,
      "grad_norm": 14.343533515930176,
      "learning_rate": 2.602602602602603e-08,
      "loss": 0.2198,
      "step": 99640
    },
    {
      "epoch": 299.25,
      "grad_norm": 5.197395324707031,
      "learning_rate": 2.5025025025025028e-08,
      "loss": 0.1807,
      "step": 99650
    },
    {
      "epoch": 299.28,
      "grad_norm": 8.971003532409668,
      "learning_rate": 2.4024024024024027e-08,
      "loss": 0.2002,
      "step": 99660
    },
    {
      "epoch": 299.31,
      "grad_norm": 7.071988105773926,
      "learning_rate": 2.3023023023023022e-08,
      "loss": 0.1827,
      "step": 99670
    },
    {
      "epoch": 299.34,
      "grad_norm": 5.75182580947876,
      "learning_rate": 2.2022022022022027e-08,
      "loss": 0.1986,
      "step": 99680
    },
    {
      "epoch": 299.37,
      "grad_norm": 6.502168655395508,
      "learning_rate": 2.1021021021021022e-08,
      "loss": 0.1889,
      "step": 99690
    },
    {
      "epoch": 299.4,
      "grad_norm": 5.676912307739258,
      "learning_rate": 2.002002002002002e-08,
      "loss": 0.146,
      "step": 99700
    },
    {
      "epoch": 299.43,
      "grad_norm": 9.79770278930664,
      "learning_rate": 1.901901901901902e-08,
      "loss": 0.1837,
      "step": 99710
    },
    {
      "epoch": 299.46,
      "grad_norm": 11.105896949768066,
      "learning_rate": 1.801801801801802e-08,
      "loss": 0.1748,
      "step": 99720
    },
    {
      "epoch": 299.49,
      "grad_norm": 7.112362384796143,
      "learning_rate": 1.701701701701702e-08,
      "loss": 0.1589,
      "step": 99730
    },
    {
      "epoch": 299.52,
      "grad_norm": 9.592845916748047,
      "learning_rate": 1.6016016016016018e-08,
      "loss": 0.2217,
      "step": 99740
    },
    {
      "epoch": 299.55,
      "grad_norm": 6.344649314880371,
      "learning_rate": 1.5015015015015016e-08,
      "loss": 0.1663,
      "step": 99750
    },
    {
      "epoch": 299.58,
      "grad_norm": 4.004574775695801,
      "learning_rate": 1.4014014014014015e-08,
      "loss": 0.1488,
      "step": 99760
    },
    {
      "epoch": 299.61,
      "grad_norm": 4.830495357513428,
      "learning_rate": 1.3013013013013015e-08,
      "loss": 0.167,
      "step": 99770
    },
    {
      "epoch": 299.64,
      "grad_norm": 8.940892219543457,
      "learning_rate": 1.2012012012012013e-08,
      "loss": 0.1532,
      "step": 99780
    },
    {
      "epoch": 299.67,
      "grad_norm": 13.181500434875488,
      "learning_rate": 1.1011011011011013e-08,
      "loss": 0.1777,
      "step": 99790
    },
    {
      "epoch": 299.7,
      "grad_norm": 10.163267135620117,
      "learning_rate": 1.001001001001001e-08,
      "loss": 0.1774,
      "step": 99800
    },
    {
      "epoch": 299.73,
      "grad_norm": 8.267355918884277,
      "learning_rate": 9.00900900900901e-09,
      "loss": 0.2018,
      "step": 99810
    },
    {
      "epoch": 299.76,
      "grad_norm": 5.464633941650391,
      "learning_rate": 8.008008008008009e-09,
      "loss": 0.1622,
      "step": 99820
    },
    {
      "epoch": 299.79,
      "grad_norm": 7.913015365600586,
      "learning_rate": 7.007007007007007e-09,
      "loss": 0.1954,
      "step": 99830
    },
    {
      "epoch": 299.82,
      "grad_norm": 8.670143127441406,
      "learning_rate": 6.006006006006007e-09,
      "loss": 0.1731,
      "step": 99840
    },
    {
      "epoch": 299.85,
      "grad_norm": 7.318467617034912,
      "learning_rate": 5.005005005005005e-09,
      "loss": 0.1991,
      "step": 99850
    },
    {
      "epoch": 299.88,
      "grad_norm": 8.884047508239746,
      "learning_rate": 4.0040040040040044e-09,
      "loss": 0.2221,
      "step": 99860
    },
    {
      "epoch": 299.91,
      "grad_norm": 13.504959106445312,
      "learning_rate": 3.0030030030030033e-09,
      "loss": 0.2383,
      "step": 99870
    },
    {
      "epoch": 299.94,
      "grad_norm": 7.2609477043151855,
      "learning_rate": 2.0020020020020022e-09,
      "loss": 0.1805,
      "step": 99880
    },
    {
      "epoch": 299.97,
      "grad_norm": 6.818796634674072,
      "learning_rate": 1.0010010010010011e-09,
      "loss": 0.2005,
      "step": 99890
    },
    {
      "epoch": 300.0,
      "grad_norm": 70.16392517089844,
      "learning_rate": 0.0,
      "loss": 0.1826,
      "step": 99900
    },
    {
      "epoch": 300.0,
      "eval_accuracy": 0.9226,
      "eval_loss": 0.35031357407569885,
      "eval_runtime": 30.443,
      "eval_samples_per_second": 328.483,
      "eval_steps_per_second": 1.314,
      "step": 99900
    },
    {
      "epoch": 300.0,
      "step": 99900,
      "total_flos": 1.000062555383815e+21,
      "train_loss": 0.2884208381307256,
      "train_runtime": 121518.863,
      "train_samples_per_second": 104.922,
      "train_steps_per_second": 0.822
    }
  ],
  "logging_steps": 10,
  "max_steps": 99900,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 300,
  "save_steps": 500,
  "total_flos": 1.000062555383815e+21,
  "train_batch_size": 128,
  "trial_name": null,
  "trial_params": null
}