{
  "best_metric": 0.0004703931335825473,
  "best_model_checkpoint": "./vit-base-beans-demo-v5\\checkpoint-10000",
  "epoch": 70.92198581560284,
  "global_step": 10000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.07,
      "learning_rate": 9.992907801418441e-05,
      "loss": 2.1762,
      "step": 10
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.98581560283688e-05,
      "loss": 2.1295,
      "step": 20
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.97872340425532e-05,
      "loss": 2.068,
      "step": 30
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.971631205673759e-05,
      "loss": 2.0647,
      "step": 40
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.9645390070922e-05,
      "loss": 1.9875,
      "step": 50
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.95744680851064e-05,
      "loss": 1.9116,
      "step": 60
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.950354609929079e-05,
      "loss": 1.7607,
      "step": 70
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.943262411347518e-05,
      "loss": 1.7694,
      "step": 80
    },
    {
      "epoch": 0.64,
      "learning_rate": 9.936170212765959e-05,
      "loss": 1.6876,
      "step": 90
    },
    {
      "epoch": 0.71,
      "learning_rate": 9.929078014184398e-05,
      "loss": 1.5852,
      "step": 100
    },
    {
      "epoch": 0.71,
      "eval_accuracy": 0.4691358024691358,
      "eval_loss": 1.652011752128601,
      "eval_runtime": 4.9705,
      "eval_samples_per_second": 114.072,
      "eval_steps_per_second": 14.284,
      "step": 100
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.921985815602838e-05,
      "loss": 1.5447,
      "step": 110
    },
    {
      "epoch": 0.85,
      "learning_rate": 9.914893617021277e-05,
      "loss": 1.4957,
      "step": 120
    },
    {
      "epoch": 0.92,
      "learning_rate": 9.907801418439718e-05,
      "loss": 1.4538,
      "step": 130
    },
    {
      "epoch": 0.99,
      "learning_rate": 9.900709219858157e-05,
      "loss": 1.4942,
      "step": 140
    },
    {
      "epoch": 1.06,
      "learning_rate": 9.893617021276596e-05,
      "loss": 1.3494,
      "step": 150
    },
    {
      "epoch": 1.13,
      "learning_rate": 9.886524822695036e-05,
      "loss": 1.3974,
      "step": 160
    },
    {
      "epoch": 1.21,
      "learning_rate": 9.879432624113476e-05,
      "loss": 1.455,
      "step": 170
    },
    {
      "epoch": 1.28,
      "learning_rate": 9.872340425531916e-05,
      "loss": 1.2304,
      "step": 180
    },
    {
      "epoch": 1.35,
      "learning_rate": 9.865248226950355e-05,
      "loss": 1.3776,
      "step": 190
    },
    {
      "epoch": 1.42,
      "learning_rate": 9.858156028368794e-05,
      "loss": 1.2438,
      "step": 200
    },
    {
      "epoch": 1.42,
      "eval_accuracy": 0.5767195767195767,
      "eval_loss": 1.2383878231048584,
      "eval_runtime": 5.082,
      "eval_samples_per_second": 111.569,
      "eval_steps_per_second": 13.971,
      "step": 200
    },
    {
      "epoch": 1.49,
      "learning_rate": 9.851063829787235e-05,
      "loss": 1.1531,
      "step": 210
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.843971631205675e-05,
      "loss": 1.2229,
      "step": 220
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.836879432624114e-05,
      "loss": 1.1538,
      "step": 230
    },
    {
      "epoch": 1.7,
      "learning_rate": 9.829787234042553e-05,
      "loss": 1.0427,
      "step": 240
    },
    {
      "epoch": 1.77,
      "learning_rate": 9.822695035460994e-05,
      "loss": 1.0046,
      "step": 250
    },
    {
      "epoch": 1.84,
      "learning_rate": 9.815602836879433e-05,
      "loss": 1.1033,
      "step": 260
    },
    {
      "epoch": 1.91,
      "learning_rate": 9.808510638297873e-05,
      "loss": 0.809,
      "step": 270
    },
    {
      "epoch": 1.99,
      "learning_rate": 9.801418439716312e-05,
      "loss": 0.8627,
      "step": 280
    },
    {
      "epoch": 2.06,
      "learning_rate": 9.794326241134753e-05,
      "loss": 0.8803,
      "step": 290
    },
    {
      "epoch": 2.13,
      "learning_rate": 9.787234042553192e-05,
      "loss": 1.0439,
      "step": 300
    },
    {
      "epoch": 2.13,
      "eval_accuracy": 0.7336860670194003,
      "eval_loss": 0.8531990647315979,
      "eval_runtime": 5.0605,
      "eval_samples_per_second": 112.043,
      "eval_steps_per_second": 14.03,
      "step": 300
    },
    {
      "epoch": 2.2,
      "learning_rate": 9.780141843971632e-05,
      "loss": 1.0166,
      "step": 310
    },
    {
      "epoch": 2.27,
      "learning_rate": 9.773049645390071e-05,
      "loss": 0.7573,
      "step": 320
    },
    {
      "epoch": 2.34,
      "learning_rate": 9.765957446808512e-05,
      "loss": 0.8116,
      "step": 330
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.758865248226951e-05,
      "loss": 0.7621,
      "step": 340
    },
    {
      "epoch": 2.48,
      "learning_rate": 9.75177304964539e-05,
      "loss": 0.7075,
      "step": 350
    },
    {
      "epoch": 2.55,
      "learning_rate": 9.74468085106383e-05,
      "loss": 0.6867,
      "step": 360
    },
    {
      "epoch": 2.62,
      "learning_rate": 9.73758865248227e-05,
      "loss": 0.7251,
      "step": 370
    },
    {
      "epoch": 2.7,
      "learning_rate": 9.73049645390071e-05,
      "loss": 0.9261,
      "step": 380
    },
    {
      "epoch": 2.77,
      "learning_rate": 9.723404255319149e-05,
      "loss": 0.8155,
      "step": 390
    },
    {
      "epoch": 2.84,
      "learning_rate": 9.716312056737588e-05,
      "loss": 0.695,
      "step": 400
    },
    {
      "epoch": 2.84,
      "eval_accuracy": 0.800705467372134,
      "eval_loss": 0.659205436706543,
      "eval_runtime": 5.1,
      "eval_samples_per_second": 111.176,
      "eval_steps_per_second": 13.921,
      "step": 400
    },
    {
      "epoch": 2.91,
      "learning_rate": 9.709219858156029e-05,
      "loss": 0.729,
      "step": 410
    },
    {
      "epoch": 2.98,
      "learning_rate": 9.702127659574469e-05,
      "loss": 0.658,
      "step": 420
    },
    {
      "epoch": 3.05,
      "learning_rate": 9.695035460992908e-05,
      "loss": 0.7035,
      "step": 430
    },
    {
      "epoch": 3.12,
      "learning_rate": 9.687943262411347e-05,
      "loss": 0.6528,
      "step": 440
    },
    {
      "epoch": 3.19,
      "learning_rate": 9.680851063829788e-05,
      "loss": 0.7444,
      "step": 450
    },
    {
      "epoch": 3.26,
      "learning_rate": 9.673758865248227e-05,
      "loss": 0.6091,
      "step": 460
    },
    {
      "epoch": 3.33,
      "learning_rate": 9.666666666666667e-05,
      "loss": 0.7101,
      "step": 470
    },
    {
      "epoch": 3.4,
      "learning_rate": 9.659574468085106e-05,
      "loss": 0.5602,
      "step": 480
    },
    {
      "epoch": 3.48,
      "learning_rate": 9.652482269503547e-05,
      "loss": 0.6144,
      "step": 490
    },
    {
      "epoch": 3.55,
      "learning_rate": 9.645390070921986e-05,
      "loss": 0.5892,
      "step": 500
    },
    {
      "epoch": 3.55,
      "eval_accuracy": 0.8059964726631393,
      "eval_loss": 0.6282046437263489,
      "eval_runtime": 5.29,
      "eval_samples_per_second": 107.182,
      "eval_steps_per_second": 13.421,
      "step": 500
    },
    {
      "epoch": 3.62,
      "learning_rate": 9.638297872340426e-05,
      "loss": 0.5771,
      "step": 510
    },
    {
      "epoch": 3.69,
      "learning_rate": 9.631205673758865e-05,
      "loss": 0.5012,
      "step": 520
    },
    {
      "epoch": 3.76,
      "learning_rate": 9.624113475177306e-05,
      "loss": 0.532,
      "step": 530
    },
    {
      "epoch": 3.83,
      "learning_rate": 9.617021276595745e-05,
      "loss": 0.5639,
      "step": 540
    },
    {
      "epoch": 3.9,
      "learning_rate": 9.609929078014184e-05,
      "loss": 0.6251,
      "step": 550
    },
    {
      "epoch": 3.97,
      "learning_rate": 9.602836879432625e-05,
      "loss": 0.4537,
      "step": 560
    },
    {
      "epoch": 4.04,
      "learning_rate": 9.595744680851064e-05,
      "loss": 0.4307,
      "step": 570
    },
    {
      "epoch": 4.11,
      "learning_rate": 9.588652482269505e-05,
      "loss": 0.446,
      "step": 580
    },
    {
      "epoch": 4.18,
      "learning_rate": 9.581560283687943e-05,
      "loss": 0.4428,
      "step": 590
    },
    {
      "epoch": 4.26,
      "learning_rate": 9.574468085106384e-05,
      "loss": 0.4763,
      "step": 600
    },
    {
      "epoch": 4.26,
      "eval_accuracy": 0.8359788359788359,
      "eval_loss": 0.5606168508529663,
      "eval_runtime": 5.2615,
      "eval_samples_per_second": 107.763,
      "eval_steps_per_second": 13.494,
      "step": 600
    },
    {
      "epoch": 4.33,
      "learning_rate": 9.567375886524823e-05,
      "loss": 0.4822,
      "step": 610
    },
    {
      "epoch": 4.4,
      "learning_rate": 9.560283687943264e-05,
      "loss": 0.5021,
      "step": 620
    },
    {
      "epoch": 4.47,
      "learning_rate": 9.553191489361702e-05,
      "loss": 0.4713,
      "step": 630
    },
    {
      "epoch": 4.54,
      "learning_rate": 9.546099290780143e-05,
      "loss": 0.5357,
      "step": 640
    },
    {
      "epoch": 4.61,
      "learning_rate": 9.539007092198582e-05,
      "loss": 0.5435,
      "step": 650
    },
    {
      "epoch": 4.68,
      "learning_rate": 9.531914893617023e-05,
      "loss": 0.3855,
      "step": 660
    },
    {
      "epoch": 4.75,
      "learning_rate": 9.524822695035461e-05,
      "loss": 0.4365,
      "step": 670
    },
    {
      "epoch": 4.82,
      "learning_rate": 9.517730496453901e-05,
      "loss": 0.4521,
      "step": 680
    },
    {
      "epoch": 4.89,
      "learning_rate": 9.510638297872341e-05,
      "loss": 0.4258,
      "step": 690
    },
    {
      "epoch": 4.96,
      "learning_rate": 9.503546099290782e-05,
      "loss": 0.4664,
      "step": 700
    },
    {
      "epoch": 4.96,
      "eval_accuracy": 0.8447971781305115,
      "eval_loss": 0.4649847149848938,
      "eval_runtime": 5.0845,
      "eval_samples_per_second": 111.514,
      "eval_steps_per_second": 13.964,
      "step": 700
    },
    {
      "epoch": 5.04,
      "learning_rate": 9.49645390070922e-05,
      "loss": 0.3395,
      "step": 710
    },
    {
      "epoch": 5.11,
      "learning_rate": 9.48936170212766e-05,
      "loss": 0.3529,
      "step": 720
    },
    {
      "epoch": 5.18,
      "learning_rate": 9.4822695035461e-05,
      "loss": 0.3969,
      "step": 730
    },
    {
      "epoch": 5.25,
      "learning_rate": 9.47517730496454e-05,
      "loss": 0.3368,
      "step": 740
    },
    {
      "epoch": 5.32,
      "learning_rate": 9.468085106382978e-05,
      "loss": 0.2199,
      "step": 750
    },
    {
      "epoch": 5.39,
      "learning_rate": 9.460992907801419e-05,
      "loss": 0.3191,
      "step": 760
    },
    {
      "epoch": 5.46,
      "learning_rate": 9.453900709219858e-05,
      "loss": 0.3199,
      "step": 770
    },
    {
      "epoch": 5.53,
      "learning_rate": 9.446808510638299e-05,
      "loss": 0.3181,
      "step": 780
    },
    {
      "epoch": 5.6,
      "learning_rate": 9.439716312056737e-05,
      "loss": 0.3619,
      "step": 790
    },
    {
      "epoch": 5.67,
      "learning_rate": 9.432624113475178e-05,
      "loss": 0.2486,
      "step": 800
    },
    {
      "epoch": 5.67,
      "eval_accuracy": 0.8800705467372134,
      "eval_loss": 0.3964751958847046,
      "eval_runtime": 4.927,
      "eval_samples_per_second": 115.079,
      "eval_steps_per_second": 14.41,
      "step": 800
    },
    {
      "epoch": 5.74,
      "learning_rate": 9.425531914893617e-05,
      "loss": 0.4254,
      "step": 810
    },
    {
      "epoch": 5.82,
      "learning_rate": 9.418439716312058e-05,
      "loss": 0.3612,
      "step": 820
    },
    {
      "epoch": 5.89,
      "learning_rate": 9.411347517730497e-05,
      "loss": 0.3508,
      "step": 830
    },
    {
      "epoch": 5.96,
      "learning_rate": 9.404255319148937e-05,
      "loss": 0.2778,
      "step": 840
    },
    {
      "epoch": 6.03,
      "learning_rate": 9.397163120567376e-05,
      "loss": 0.1774,
      "step": 850
    },
    {
      "epoch": 6.1,
      "learning_rate": 9.390070921985817e-05,
      "loss": 0.2778,
      "step": 860
    },
    {
      "epoch": 6.17,
      "learning_rate": 9.382978723404256e-05,
      "loss": 0.2184,
      "step": 870
    },
    {
      "epoch": 6.24,
      "learning_rate": 9.375886524822695e-05,
      "loss": 0.3215,
      "step": 880
    },
    {
      "epoch": 6.31,
      "learning_rate": 9.368794326241135e-05,
      "loss": 0.3755,
      "step": 890
    },
    {
      "epoch": 6.38,
      "learning_rate": 9.361702127659576e-05,
      "loss": 0.2726,
      "step": 900
    },
    {
      "epoch": 6.38,
      "eval_accuracy": 0.9047619047619048,
      "eval_loss": 0.34163007140159607,
      "eval_runtime": 4.777,
      "eval_samples_per_second": 118.693,
      "eval_steps_per_second": 14.863,
      "step": 900
    },
    {
      "epoch": 6.45,
      "learning_rate": 9.354609929078015e-05,
      "loss": 0.2767,
      "step": 910
    },
    {
      "epoch": 6.52,
      "learning_rate": 9.34822695035461e-05,
      "loss": 0.3285,
      "step": 920
    },
    {
      "epoch": 6.6,
      "learning_rate": 9.34113475177305e-05,
      "loss": 0.3168,
      "step": 930
    },
    {
      "epoch": 6.67,
      "learning_rate": 9.33404255319149e-05,
      "loss": 0.1633,
      "step": 940
    },
    {
      "epoch": 6.74,
      "learning_rate": 9.326950354609929e-05,
      "loss": 0.3204,
      "step": 950
    },
    {
      "epoch": 6.81,
      "learning_rate": 9.319858156028369e-05,
      "loss": 0.2753,
      "step": 960
    },
    {
      "epoch": 6.88,
      "learning_rate": 9.312765957446809e-05,
      "loss": 0.2775,
      "step": 970
    },
    {
      "epoch": 6.95,
      "learning_rate": 9.30567375886525e-05,
      "loss": 0.2352,
      "step": 980
    },
    {
      "epoch": 7.02,
      "learning_rate": 9.298581560283687e-05,
      "loss": 0.2665,
      "step": 990
    },
    {
      "epoch": 7.09,
      "learning_rate": 9.291489361702128e-05,
      "loss": 0.2587,
      "step": 1000
    },
    {
      "epoch": 7.09,
      "eval_accuracy": 0.9241622574955908,
      "eval_loss": 0.25383827090263367,
      "eval_runtime": 4.9285,
      "eval_samples_per_second": 115.044,
      "eval_steps_per_second": 14.406,
      "step": 1000
    },
    {
      "epoch": 7.16,
      "learning_rate": 9.284397163120567e-05,
      "loss": 0.3072,
      "step": 1010
    },
    {
      "epoch": 7.23,
      "learning_rate": 9.277304964539008e-05,
      "loss": 0.3872,
      "step": 1020
    },
    {
      "epoch": 7.3,
      "learning_rate": 9.270212765957446e-05,
      "loss": 0.2625,
      "step": 1030
    },
    {
      "epoch": 7.38,
      "learning_rate": 9.263120567375887e-05,
      "loss": 0.3353,
      "step": 1040
    },
    {
      "epoch": 7.45,
      "learning_rate": 9.256028368794328e-05,
      "loss": 0.1992,
      "step": 1050
    },
    {
      "epoch": 7.52,
      "learning_rate": 9.248936170212767e-05,
      "loss": 0.244,
      "step": 1060
    },
    {
      "epoch": 7.59,
      "learning_rate": 9.241843971631206e-05,
      "loss": 0.1355,
      "step": 1070
    },
    {
      "epoch": 7.66,
      "learning_rate": 9.234751773049646e-05,
      "loss": 0.1959,
      "step": 1080
    },
    {
      "epoch": 7.73,
      "learning_rate": 9.227659574468086e-05,
      "loss": 0.3748,
      "step": 1090
    },
    {
      "epoch": 7.8,
      "learning_rate": 9.220567375886526e-05,
      "loss": 0.2134,
      "step": 1100
    },
    {
      "epoch": 7.8,
      "eval_accuracy": 0.9347442680776014,
      "eval_loss": 0.21870087087154388,
      "eval_runtime": 4.96,
      "eval_samples_per_second": 114.314,
      "eval_steps_per_second": 14.314,
      "step": 1100
    },
    {
      "epoch": 7.87,
      "learning_rate": 9.213475177304965e-05,
      "loss": 0.2255,
      "step": 1110
    },
    {
      "epoch": 7.94,
      "learning_rate": 9.206382978723404e-05,
      "loss": 0.2098,
      "step": 1120
    },
    {
      "epoch": 8.01,
      "learning_rate": 9.199290780141845e-05,
      "loss": 0.1824,
      "step": 1130
    },
    {
      "epoch": 8.09,
      "learning_rate": 9.192198581560285e-05,
      "loss": 0.1689,
      "step": 1140
    },
    {
      "epoch": 8.16,
      "learning_rate": 9.185106382978724e-05,
      "loss": 0.1869,
      "step": 1150
    },
    {
      "epoch": 8.23,
      "learning_rate": 9.178014184397163e-05,
      "loss": 0.3218,
      "step": 1160
    },
    {
      "epoch": 8.3,
      "learning_rate": 9.170921985815604e-05,
      "loss": 0.2401,
      "step": 1170
    },
    {
      "epoch": 8.37,
      "learning_rate": 9.163829787234043e-05,
      "loss": 0.2202,
      "step": 1180
    },
    {
      "epoch": 8.44,
      "learning_rate": 9.156737588652483e-05,
      "loss": 0.1199,
      "step": 1190
    },
    {
      "epoch": 8.51,
      "learning_rate": 9.149645390070922e-05,
      "loss": 0.1366,
      "step": 1200
    },
    {
      "epoch": 8.51,
      "eval_accuracy": 0.9365079365079365,
      "eval_loss": 0.25872817635536194,
      "eval_runtime": 5.0905,
      "eval_samples_per_second": 111.383,
      "eval_steps_per_second": 13.947,
      "step": 1200
    },
    {
      "epoch": 8.58,
      "learning_rate": 9.142553191489363e-05,
      "loss": 0.2913,
      "step": 1210
    },
    {
      "epoch": 8.65,
      "learning_rate": 9.135460992907802e-05,
      "loss": 0.1504,
      "step": 1220
    },
    {
      "epoch": 8.72,
      "learning_rate": 9.128368794326241e-05,
      "loss": 0.1568,
      "step": 1230
    },
    {
      "epoch": 8.79,
      "learning_rate": 9.121276595744681e-05,
      "loss": 0.2478,
      "step": 1240
    },
    {
      "epoch": 8.87,
      "learning_rate": 9.114184397163122e-05,
      "loss": 0.1182,
      "step": 1250
    },
    {
      "epoch": 8.94,
      "learning_rate": 9.107092198581561e-05,
      "loss": 0.1879,
      "step": 1260
    },
    {
      "epoch": 9.01,
      "learning_rate": 9.1e-05,
      "loss": 0.0922,
      "step": 1270
    },
    {
      "epoch": 9.08,
      "learning_rate": 9.09290780141844e-05,
      "loss": 0.1083,
      "step": 1280
    },
    {
      "epoch": 9.15,
      "learning_rate": 9.08581560283688e-05,
      "loss": 0.1389,
      "step": 1290
    },
    {
      "epoch": 9.22,
      "learning_rate": 9.07872340425532e-05,
      "loss": 0.1062,
      "step": 1300
    },
    {
      "epoch": 9.22,
      "eval_accuracy": 0.9329805996472663,
      "eval_loss": 0.23322318494319916,
      "eval_runtime": 4.9,
      "eval_samples_per_second": 115.713,
      "eval_steps_per_second": 14.49,
      "step": 1300
    },
    {
      "epoch": 9.29,
      "learning_rate": 9.071631205673759e-05,
      "loss": 0.1666,
      "step": 1310
    },
    {
      "epoch": 9.36,
      "learning_rate": 9.064539007092198e-05,
      "loss": 0.143,
      "step": 1320
    },
    {
      "epoch": 9.43,
      "learning_rate": 9.057446808510639e-05,
      "loss": 0.1701,
      "step": 1330
    },
    {
      "epoch": 9.5,
      "learning_rate": 9.050354609929079e-05,
      "loss": 0.1869,
      "step": 1340
    },
    {
      "epoch": 9.57,
      "learning_rate": 9.043262411347518e-05,
      "loss": 0.2483,
      "step": 1350
    },
    {
      "epoch": 9.65,
      "learning_rate": 9.036170212765957e-05,
      "loss": 0.1374,
      "step": 1360
    },
    {
      "epoch": 9.72,
      "learning_rate": 9.029078014184398e-05,
      "loss": 0.2322,
      "step": 1370
    },
    {
      "epoch": 9.79,
      "learning_rate": 9.021985815602837e-05,
      "loss": 0.2282,
      "step": 1380
    },
    {
      "epoch": 9.86,
      "learning_rate": 9.014893617021277e-05,
      "loss": 0.1862,
      "step": 1390
    },
    {
      "epoch": 9.93,
      "learning_rate": 9.007801418439716e-05,
      "loss": 0.1873,
      "step": 1400
    },
    {
      "epoch": 9.93,
      "eval_accuracy": 0.9382716049382716,
      "eval_loss": 0.19933444261550903,
      "eval_runtime": 4.8755,
      "eval_samples_per_second": 116.295,
      "eval_steps_per_second": 14.562,
      "step": 1400
    },
    {
      "epoch": 10.0,
      "learning_rate": 9.000709219858157e-05,
      "loss": 0.0941,
      "step": 1410
    },
    {
      "epoch": 10.07,
      "learning_rate": 8.993617021276596e-05,
      "loss": 0.1053,
      "step": 1420
    },
    {
      "epoch": 10.14,
      "learning_rate": 8.986524822695035e-05,
      "loss": 0.1271,
      "step": 1430
    },
    {
      "epoch": 10.21,
      "learning_rate": 8.979432624113475e-05,
      "loss": 0.1664,
      "step": 1440
    },
    {
      "epoch": 10.28,
      "learning_rate": 8.972340425531916e-05,
      "loss": 0.123,
      "step": 1450
    },
    {
      "epoch": 10.35,
      "learning_rate": 8.965248226950355e-05,
      "loss": 0.2243,
      "step": 1460
    },
    {
      "epoch": 10.43,
      "learning_rate": 8.958156028368794e-05,
      "loss": 0.1105,
      "step": 1470
    },
    {
      "epoch": 10.5,
      "learning_rate": 8.951063829787234e-05,
      "loss": 0.128,
      "step": 1480
    },
    {
      "epoch": 10.57,
      "learning_rate": 8.943971631205674e-05,
      "loss": 0.177,
      "step": 1490
    },
    {
      "epoch": 10.64,
      "learning_rate": 8.936879432624114e-05,
      "loss": 0.0827,
      "step": 1500
    },
    {
      "epoch": 10.64,
      "eval_accuracy": 0.9435626102292769,
      "eval_loss": 0.2013310194015503,
      "eval_runtime": 4.8795,
      "eval_samples_per_second": 116.2,
      "eval_steps_per_second": 14.551,
      "step": 1500
    },
    {
      "epoch": 10.71,
      "learning_rate": 8.929787234042553e-05,
      "loss": 0.0708,
      "step": 1510
    },
    {
      "epoch": 10.78,
      "learning_rate": 8.922695035460992e-05,
      "loss": 0.0474,
      "step": 1520
    },
    {
      "epoch": 10.85,
      "learning_rate": 8.915602836879433e-05,
      "loss": 0.2452,
      "step": 1530
    },
    {
      "epoch": 10.92,
      "learning_rate": 8.908510638297873e-05,
      "loss": 0.1033,
      "step": 1540
    },
    {
      "epoch": 10.99,
      "learning_rate": 8.901418439716312e-05,
      "loss": 0.1123,
      "step": 1550
    },
    {
      "epoch": 11.06,
      "learning_rate": 8.894326241134751e-05,
      "loss": 0.0597,
      "step": 1560
    },
    {
      "epoch": 11.13,
      "learning_rate": 8.887234042553192e-05,
      "loss": 0.1401,
      "step": 1570
    },
    {
      "epoch": 11.21,
      "learning_rate": 8.880141843971633e-05,
      "loss": 0.1481,
      "step": 1580
    },
    {
      "epoch": 11.28,
      "learning_rate": 8.873049645390071e-05,
      "loss": 0.0823,
      "step": 1590
    },
    {
      "epoch": 11.35,
      "learning_rate": 8.865957446808511e-05,
      "loss": 0.2071,
      "step": 1600
    },
    {
      "epoch": 11.35,
      "eval_accuracy": 0.9188712522045855,
      "eval_loss": 0.2737657129764557,
      "eval_runtime": 4.986,
      "eval_samples_per_second": 113.717,
      "eval_steps_per_second": 14.24,
      "step": 1600
    },
    {
      "epoch": 11.42,
      "learning_rate": 8.858865248226951e-05,
      "loss": 0.1693,
      "step": 1610
    },
    {
      "epoch": 11.49,
      "learning_rate": 8.852482269503546e-05,
      "loss": 0.1344,
      "step": 1620
    },
    {
      "epoch": 11.56,
      "learning_rate": 8.845390070921986e-05,
      "loss": 0.0623,
      "step": 1630
    },
    {
      "epoch": 11.63,
      "learning_rate": 8.838297872340426e-05,
      "loss": 0.1087,
      "step": 1640
    },
    {
      "epoch": 11.7,
      "learning_rate": 8.831205673758866e-05,
      "loss": 0.0971,
      "step": 1650
    },
    {
      "epoch": 11.77,
      "learning_rate": 8.824113475177305e-05,
      "loss": 0.0733,
      "step": 1660
    },
    {
      "epoch": 11.84,
      "learning_rate": 8.817021276595744e-05,
      "loss": 0.0471,
      "step": 1670
    },
    {
      "epoch": 11.91,
      "learning_rate": 8.809929078014185e-05,
      "loss": 0.2112,
      "step": 1680
    },
    {
      "epoch": 11.99,
      "learning_rate": 8.802836879432625e-05,
      "loss": 0.1347,
      "step": 1690
    },
    {
      "epoch": 12.06,
      "learning_rate": 8.795744680851065e-05,
      "loss": 0.0477,
      "step": 1700
    },
    {
      "epoch": 12.06,
      "eval_accuracy": 0.9735449735449735,
      "eval_loss": 0.09033568948507309,
      "eval_runtime": 5.073,
      "eval_samples_per_second": 111.767,
      "eval_steps_per_second": 13.996,
      "step": 1700
    },
    {
      "epoch": 12.13,
      "learning_rate": 8.788652482269503e-05,
      "loss": 0.105,
      "step": 1710
    },
    {
      "epoch": 12.2,
      "learning_rate": 8.781560283687944e-05,
      "loss": 0.046,
      "step": 1720
    },
    {
      "epoch": 12.27,
      "learning_rate": 8.774468085106383e-05,
      "loss": 0.0763,
      "step": 1730
    },
    {
      "epoch": 12.34,
      "learning_rate": 8.767375886524824e-05,
      "loss": 0.1138,
      "step": 1740
    },
    {
      "epoch": 12.41,
      "learning_rate": 8.760283687943262e-05,
      "loss": 0.0569,
      "step": 1750
    },
    {
      "epoch": 12.48,
      "learning_rate": 8.753191489361703e-05,
      "loss": 0.1693,
      "step": 1760
    },
    {
      "epoch": 12.55,
      "learning_rate": 8.746099290780142e-05,
      "loss": 0.0377,
      "step": 1770
    },
    {
      "epoch": 12.62,
      "learning_rate": 8.739007092198583e-05,
      "loss": 0.0173,
      "step": 1780
    },
    {
      "epoch": 12.7,
      "learning_rate": 8.731914893617021e-05,
      "loss": 0.0383,
      "step": 1790
    },
    {
      "epoch": 12.77,
      "learning_rate": 8.724822695035462e-05,
      "loss": 0.081,
      "step": 1800
    },
    {
      "epoch": 12.77,
      "eval_accuracy": 0.9753086419753086,
      "eval_loss": 0.10142536461353302,
      "eval_runtime": 5.0755,
      "eval_samples_per_second": 111.712,
      "eval_steps_per_second": 13.989,
      "step": 1800
    },
    {
      "epoch": 12.84,
      "learning_rate": 8.717730496453901e-05,
      "loss": 0.011,
      "step": 1810
    },
    {
      "epoch": 12.91,
      "learning_rate": 8.710638297872342e-05,
      "loss": 0.0556,
      "step": 1820
    },
    {
      "epoch": 12.98,
      "learning_rate": 8.70354609929078e-05,
      "loss": 0.0798,
      "step": 1830
    },
    {
      "epoch": 13.05,
      "learning_rate": 8.69645390070922e-05,
      "loss": 0.0156,
      "step": 1840
    },
    {
      "epoch": 13.12,
      "learning_rate": 8.68936170212766e-05,
      "loss": 0.05,
      "step": 1850
    },
    {
      "epoch": 13.19,
      "learning_rate": 8.6822695035461e-05,
      "loss": 0.1144,
      "step": 1860
    },
    {
      "epoch": 13.26,
      "learning_rate": 8.675177304964538e-05,
      "loss": 0.1024,
      "step": 1870
    },
    {
      "epoch": 13.33,
      "learning_rate": 8.668085106382979e-05,
      "loss": 0.0993,
      "step": 1880
    },
    {
      "epoch": 13.4,
      "learning_rate": 8.660992907801419e-05,
      "loss": 0.0832,
      "step": 1890
    },
    {
      "epoch": 13.48,
      "learning_rate": 8.653900709219859e-05,
      "loss": 0.0191,
      "step": 1900
    },
    {
      "epoch": 13.48,
      "eval_accuracy": 0.9523809523809523,
      "eval_loss": 0.1780620813369751,
      "eval_runtime": 4.8325,
      "eval_samples_per_second": 117.33,
      "eval_steps_per_second": 14.692,
      "step": 1900
    },
    {
      "epoch": 13.55,
      "learning_rate": 8.646808510638297e-05,
      "loss": 0.0464,
      "step": 1910
    },
    {
      "epoch": 13.62,
      "learning_rate": 8.639716312056738e-05,
      "loss": 0.0368,
      "step": 1920
    },
    {
      "epoch": 13.69,
      "learning_rate": 8.632624113475177e-05,
      "loss": 0.0582,
      "step": 1930
    },
    {
      "epoch": 13.76,
      "learning_rate": 8.625531914893618e-05,
      "loss": 0.0272,
      "step": 1940
    },
    {
      "epoch": 13.83,
      "learning_rate": 8.618439716312056e-05,
      "loss": 0.0241,
      "step": 1950
    },
    {
      "epoch": 13.9,
      "learning_rate": 8.611347517730497e-05,
      "loss": 0.1776,
      "step": 1960
    },
    {
      "epoch": 13.97,
      "learning_rate": 8.604255319148936e-05,
      "loss": 0.108,
      "step": 1970
    },
    {
      "epoch": 14.04,
      "learning_rate": 8.597163120567377e-05,
      "loss": 0.1158,
      "step": 1980
    },
    {
      "epoch": 14.11,
      "learning_rate": 8.590070921985815e-05,
      "loss": 0.0953,
      "step": 1990
    },
    {
      "epoch": 14.18,
      "learning_rate": 8.582978723404256e-05,
      "loss": 0.1016,
      "step": 2000
    },
    {
      "epoch": 14.18,
      "eval_accuracy": 0.9365079365079365,
      "eval_loss": 0.23041455447673798,
      "eval_runtime": 5.125,
      "eval_samples_per_second": 110.633,
      "eval_steps_per_second": 13.854,
      "step": 2000
    },
    {
      "epoch": 14.26,
      "learning_rate": 8.575886524822695e-05,
      "loss": 0.0435,
      "step": 2010
    },
    {
      "epoch": 14.33,
      "learning_rate": 8.568794326241136e-05,
      "loss": 0.1088,
      "step": 2020
    },
    {
      "epoch": 14.4,
      "learning_rate": 8.561702127659574e-05,
      "loss": 0.0868,
      "step": 2030
    },
    {
      "epoch": 14.47,
      "learning_rate": 8.554609929078014e-05,
      "loss": 0.0725,
      "step": 2040
    },
    {
      "epoch": 14.54,
      "learning_rate": 8.547517730496454e-05,
      "loss": 0.1031,
      "step": 2050
    },
    {
      "epoch": 14.61,
      "learning_rate": 8.540425531914894e-05,
      "loss": 0.1031,
      "step": 2060
    },
    {
      "epoch": 14.68,
      "learning_rate": 8.533333333333334e-05,
      "loss": 0.0604,
      "step": 2070
    },
    {
      "epoch": 14.75,
      "learning_rate": 8.526241134751773e-05,
      "loss": 0.1943,
      "step": 2080
    },
    {
      "epoch": 14.82,
      "learning_rate": 8.519148936170214e-05,
      "loss": 0.106,
      "step": 2090
    },
    {
      "epoch": 14.89,
      "learning_rate": 8.512056737588653e-05,
      "loss": 0.0465,
      "step": 2100
    },
    {
      "epoch": 14.89,
      "eval_accuracy": 0.9594356261022927,
      "eval_loss": 0.14327698945999146,
      "eval_runtime": 5.158,
      "eval_samples_per_second": 109.925,
      "eval_steps_per_second": 13.765,
      "step": 2100
    },
    {
      "epoch": 14.96,
      "learning_rate": 8.504964539007093e-05,
      "loss": 0.0911,
      "step": 2110
    },
    {
      "epoch": 15.04,
      "learning_rate": 8.497872340425532e-05,
      "loss": 0.0861,
      "step": 2120
    },
    {
      "epoch": 15.11,
      "learning_rate": 8.490780141843973e-05,
      "loss": 0.0557,
      "step": 2130
    },
    {
      "epoch": 15.18,
      "learning_rate": 8.483687943262412e-05,
      "loss": 0.103,
      "step": 2140
    },
    {
      "epoch": 15.25,
      "learning_rate": 8.476595744680851e-05,
      "loss": 0.0555,
      "step": 2150
    },
    {
      "epoch": 15.32,
      "learning_rate": 8.469503546099291e-05,
      "loss": 0.0473,
      "step": 2160
    },
    {
      "epoch": 15.39,
      "learning_rate": 8.462411347517732e-05,
      "loss": 0.1456,
      "step": 2170
    },
    {
      "epoch": 15.46,
      "learning_rate": 8.455319148936171e-05,
      "loss": 0.0771,
      "step": 2180
    },
    {
      "epoch": 15.53,
      "learning_rate": 8.44822695035461e-05,
      "loss": 0.0529,
      "step": 2190
    },
    {
      "epoch": 15.6,
      "learning_rate": 8.44113475177305e-05,
      "loss": 0.0412,
      "step": 2200
    },
    {
      "epoch": 15.6,
      "eval_accuracy": 0.9735449735449735,
      "eval_loss": 0.12634754180908203,
      "eval_runtime": 4.8775,
      "eval_samples_per_second": 116.247,
      "eval_steps_per_second": 14.557,
      "step": 2200
    },
    {
      "epoch": 15.67,
      "learning_rate": 8.43404255319149e-05,
      "loss": 0.0639,
      "step": 2210
    },
    {
      "epoch": 15.74,
      "learning_rate": 8.42695035460993e-05,
      "loss": 0.0561,
      "step": 2220
    },
    {
      "epoch": 15.82,
      "learning_rate": 8.419858156028369e-05,
      "loss": 0.0537,
      "step": 2230
    },
    {
      "epoch": 15.89,
      "learning_rate": 8.412765957446808e-05,
      "loss": 0.0372,
      "step": 2240
    },
    {
      "epoch": 15.96,
      "learning_rate": 8.405673758865249e-05,
      "loss": 0.0293,
      "step": 2250
    },
    {
      "epoch": 16.03,
      "learning_rate": 8.398581560283688e-05,
      "loss": 0.0203,
      "step": 2260
    },
    {
      "epoch": 16.1,
      "learning_rate": 8.391489361702128e-05,
      "loss": 0.0838,
      "step": 2270
    },
    {
      "epoch": 16.17,
      "learning_rate": 8.384397163120567e-05,
      "loss": 0.015,
      "step": 2280
    },
    {
      "epoch": 16.24,
      "learning_rate": 8.377304964539008e-05,
      "loss": 0.0496,
      "step": 2290
    },
    {
      "epoch": 16.31,
      "learning_rate": 8.370212765957447e-05,
      "loss": 0.0449,
      "step": 2300
    },
    {
      "epoch": 16.31,
      "eval_accuracy": 0.9594356261022927,
      "eval_loss": 0.17587855458259583,
      "eval_runtime": 4.748,
      "eval_samples_per_second": 119.418,
      "eval_steps_per_second": 14.954,
      "step": 2300
    },
    {
      "epoch": 16.38,
      "learning_rate": 8.363120567375887e-05,
      "loss": 0.1219,
      "step": 2310
    },
    {
      "epoch": 16.45,
      "learning_rate": 8.356028368794326e-05,
      "loss": 0.0748,
      "step": 2320
    },
    {
      "epoch": 16.52,
      "learning_rate": 8.348936170212767e-05,
      "loss": 0.1141,
      "step": 2330
    },
    {
      "epoch": 16.6,
      "learning_rate": 8.341843971631206e-05,
      "loss": 0.0559,
      "step": 2340
    },
    {
      "epoch": 16.67,
      "learning_rate": 8.334751773049645e-05,
      "loss": 0.0261,
      "step": 2350
    },
    {
      "epoch": 16.74,
      "learning_rate": 8.327659574468085e-05,
      "loss": 0.0896,
      "step": 2360
    },
    {
      "epoch": 16.81,
      "learning_rate": 8.320567375886526e-05,
      "loss": 0.0891,
      "step": 2370
    },
    {
      "epoch": 16.88,
      "learning_rate": 8.313475177304965e-05,
      "loss": 0.0454,
      "step": 2380
    },
    {
      "epoch": 16.95,
      "learning_rate": 8.306382978723404e-05,
      "loss": 0.0323,
      "step": 2390
    },
    {
      "epoch": 17.02,
      "learning_rate": 8.299290780141844e-05,
      "loss": 0.0614,
      "step": 2400
    },
    {
      "epoch": 17.02,
      "eval_accuracy": 0.9682539682539683,
      "eval_loss": 0.11620796471834183,
      "eval_runtime": 5.1925,
      "eval_samples_per_second": 109.195,
      "eval_steps_per_second": 13.673,
      "step": 2400
    },
    {
      "epoch": 17.09,
      "learning_rate": 8.292198581560284e-05,
      "loss": 0.0402,
      "step": 2410
    },
    {
      "epoch": 17.16,
      "learning_rate": 8.285106382978724e-05,
      "loss": 0.0472,
      "step": 2420
    },
    {
      "epoch": 17.23,
      "learning_rate": 8.278014184397163e-05,
      "loss": 0.0331,
      "step": 2430
    },
    {
      "epoch": 17.3,
      "learning_rate": 8.270921985815602e-05,
      "loss": 0.1932,
      "step": 2440
    },
    {
      "epoch": 17.38,
      "learning_rate": 8.263829787234043e-05,
      "loss": 0.0786,
      "step": 2450
    },
    {
      "epoch": 17.45,
      "learning_rate": 8.256737588652482e-05,
      "loss": 0.1532,
      "step": 2460
    },
    {
      "epoch": 17.52,
      "learning_rate": 8.249645390070922e-05,
      "loss": 0.028,
      "step": 2470
    },
    {
      "epoch": 17.59,
      "learning_rate": 8.242553191489361e-05,
      "loss": 0.0448,
      "step": 2480
    },
    {
      "epoch": 17.66,
      "learning_rate": 8.235460992907802e-05,
      "loss": 0.0867,
      "step": 2490
    },
    {
      "epoch": 17.73,
      "learning_rate": 8.228368794326241e-05,
      "loss": 0.019,
      "step": 2500
    },
    {
      "epoch": 17.73,
      "eval_accuracy": 0.9841269841269841,
      "eval_loss": 0.057109106332063675,
      "eval_runtime": 4.9535,
      "eval_samples_per_second": 114.464,
      "eval_steps_per_second": 14.333,
      "step": 2500
    },
    {
      "epoch": 17.8,
      "learning_rate": 8.221276595744682e-05,
      "loss": 0.0047,
      "step": 2510
    },
    {
      "epoch": 17.87,
      "learning_rate": 8.21418439716312e-05,
      "loss": 0.0665,
      "step": 2520
    },
    {
      "epoch": 17.94,
      "learning_rate": 8.207092198581561e-05,
      "loss": 0.1127,
      "step": 2530
    },
    {
      "epoch": 18.01,
      "learning_rate": 8.2e-05,
      "loss": 0.073,
      "step": 2540
    },
    {
      "epoch": 18.09,
      "learning_rate": 8.192907801418441e-05,
      "loss": 0.0051,
      "step": 2550
    },
    {
      "epoch": 18.16,
      "learning_rate": 8.185815602836879e-05,
      "loss": 0.1039,
      "step": 2560
    },
    {
      "epoch": 18.23,
      "learning_rate": 8.17872340425532e-05,
      "loss": 0.117,
      "step": 2570
    },
    {
      "epoch": 18.3,
      "learning_rate": 8.171631205673759e-05,
      "loss": 0.0763,
      "step": 2580
    },
    {
      "epoch": 18.37,
      "learning_rate": 8.1645390070922e-05,
      "loss": 0.0678,
      "step": 2590
    },
    {
      "epoch": 18.44,
      "learning_rate": 8.157446808510638e-05,
      "loss": 0.1211,
      "step": 2600
    },
    {
      "epoch": 18.44,
      "eval_accuracy": 0.9541446208112875,
      "eval_loss": 0.21715059876441956,
      "eval_runtime": 5.0345,
      "eval_samples_per_second": 112.622,
      "eval_steps_per_second": 14.103,
      "step": 2600
    },
    {
      "epoch": 18.51,
      "learning_rate": 8.150354609929078e-05,
      "loss": 0.1104,
      "step": 2610
    },
    {
      "epoch": 18.58,
      "learning_rate": 8.143262411347519e-05,
      "loss": 0.0693,
      "step": 2620
    },
    {
      "epoch": 18.65,
      "learning_rate": 8.136170212765958e-05,
      "loss": 0.0279,
      "step": 2630
    },
    {
      "epoch": 18.72,
      "learning_rate": 8.129078014184398e-05,
      "loss": 0.0757,
      "step": 2640
    },
    {
      "epoch": 18.79,
      "learning_rate": 8.121985815602837e-05,
      "loss": 0.0996,
      "step": 2650
    },
    {
      "epoch": 18.87,
      "learning_rate": 8.114893617021278e-05,
      "loss": 0.0508,
      "step": 2660
    },
    {
      "epoch": 18.94,
      "learning_rate": 8.107801418439717e-05,
      "loss": 0.0114,
      "step": 2670
    },
    {
      "epoch": 19.01,
      "learning_rate": 8.100709219858157e-05,
      "loss": 0.0097,
      "step": 2680
    },
    {
      "epoch": 19.08,
      "learning_rate": 8.093617021276596e-05,
      "loss": 0.0695,
      "step": 2690
    },
    {
      "epoch": 19.15,
      "learning_rate": 8.086524822695037e-05,
      "loss": 0.12,
      "step": 2700
    },
    {
      "epoch": 19.15,
      "eval_accuracy": 0.9664902998236331,
      "eval_loss": 0.14609535038471222,
      "eval_runtime": 4.9415,
      "eval_samples_per_second": 114.741,
      "eval_steps_per_second": 14.368,
      "step": 2700
    },
    {
      "epoch": 19.22,
      "learning_rate": 8.079432624113476e-05,
      "loss": 0.0491,
      "step": 2710
    },
    {
      "epoch": 19.29,
      "learning_rate": 8.072340425531915e-05,
      "loss": 0.0297,
      "step": 2720
    },
    {
      "epoch": 19.36,
      "learning_rate": 8.065248226950355e-05,
      "loss": 0.0691,
      "step": 2730
    },
    {
      "epoch": 19.43,
      "learning_rate": 8.058156028368795e-05,
      "loss": 0.0482,
      "step": 2740
    },
    {
      "epoch": 19.5,
      "learning_rate": 8.051063829787235e-05,
      "loss": 0.0036,
      "step": 2750
    },
    {
      "epoch": 19.57,
      "learning_rate": 8.043971631205674e-05,
      "loss": 0.0894,
      "step": 2760
    },
    {
      "epoch": 19.65,
      "learning_rate": 8.036879432624114e-05,
      "loss": 0.0737,
      "step": 2770
    },
    {
      "epoch": 19.72,
      "learning_rate": 8.029787234042554e-05,
      "loss": 0.0783,
      "step": 2780
    },
    {
      "epoch": 19.79,
      "learning_rate": 8.022695035460994e-05,
      "loss": 0.069,
      "step": 2790
    },
    {
      "epoch": 19.86,
      "learning_rate": 8.015602836879433e-05,
      "loss": 0.0688,
      "step": 2800
    },
    {
      "epoch": 19.86,
      "eval_accuracy": 0.9488536155202821,
      "eval_loss": 0.2023719996213913,
      "eval_runtime": 5.025,
      "eval_samples_per_second": 112.835,
      "eval_steps_per_second": 14.129,
      "step": 2800
    },
    {
      "epoch": 19.93,
      "learning_rate": 8.008510638297872e-05,
      "loss": 0.2303,
      "step": 2810
    },
    {
      "epoch": 20.0,
      "learning_rate": 8.001418439716313e-05,
      "loss": 0.0436,
      "step": 2820
    },
    {
      "epoch": 20.07,
      "learning_rate": 7.994326241134752e-05,
      "loss": 0.0297,
      "step": 2830
    },
    {
      "epoch": 20.14,
      "learning_rate": 7.987234042553192e-05,
      "loss": 0.0584,
      "step": 2840
    },
    {
      "epoch": 20.21,
      "learning_rate": 7.980141843971631e-05,
      "loss": 0.0388,
      "step": 2850
    },
    {
      "epoch": 20.28,
      "learning_rate": 7.973049645390072e-05,
      "loss": 0.0745,
      "step": 2860
    },
    {
      "epoch": 20.35,
      "learning_rate": 7.965957446808511e-05,
      "loss": 0.0029,
      "step": 2870
    },
    {
      "epoch": 20.43,
      "learning_rate": 7.95886524822695e-05,
      "loss": 0.0432,
      "step": 2880
    },
    {
      "epoch": 20.5,
      "learning_rate": 7.95177304964539e-05,
      "loss": 0.0085,
      "step": 2890
    },
    {
      "epoch": 20.57,
      "learning_rate": 7.94468085106383e-05,
      "loss": 0.0462,
      "step": 2900
    },
    {
      "epoch": 20.57,
      "eval_accuracy": 0.982363315696649,
      "eval_loss": 0.04924876615405083,
      "eval_runtime": 4.8635,
      "eval_samples_per_second": 116.582,
      "eval_steps_per_second": 14.598,
      "step": 2900
    },
    {
      "epoch": 20.64,
      "learning_rate": 7.93758865248227e-05,
      "loss": 0.0076,
      "step": 2910
    },
    {
      "epoch": 20.71,
      "learning_rate": 7.93049645390071e-05,
      "loss": 0.0285,
      "step": 2920
    },
    {
      "epoch": 20.78,
      "learning_rate": 7.923404255319149e-05,
      "loss": 0.0394,
      "step": 2930
    },
    {
      "epoch": 20.85,
      "learning_rate": 7.91631205673759e-05,
      "loss": 0.2017,
      "step": 2940
    },
    {
      "epoch": 20.92,
      "learning_rate": 7.909219858156029e-05,
      "loss": 0.1279,
      "step": 2950
    },
    {
      "epoch": 20.99,
      "learning_rate": 7.902127659574468e-05,
      "loss": 0.1861,
      "step": 2960
    },
    {
      "epoch": 21.06,
      "learning_rate": 7.895035460992908e-05,
      "loss": 0.0247,
      "step": 2970
    },
    {
      "epoch": 21.13,
      "learning_rate": 7.887943262411348e-05,
      "loss": 0.0872,
      "step": 2980
    },
    {
      "epoch": 21.21,
      "learning_rate": 7.880851063829788e-05,
      "loss": 0.0511,
      "step": 2990
    },
    {
      "epoch": 21.28,
      "learning_rate": 7.873758865248227e-05,
      "loss": 0.0147,
      "step": 3000
    },
    {
      "epoch": 21.28,
      "eval_accuracy": 0.9788359788359788,
      "eval_loss": 0.08856220543384552,
      "eval_runtime": 4.799,
      "eval_samples_per_second": 118.149,
      "eval_steps_per_second": 14.795,
      "step": 3000
    },
    {
      "epoch": 21.35,
      "learning_rate": 7.866666666666666e-05,
      "loss": 0.0564,
      "step": 3010
    },
    {
      "epoch": 21.42,
      "learning_rate": 7.859574468085107e-05,
      "loss": 0.0178,
      "step": 3020
    },
    {
      "epoch": 21.49,
      "learning_rate": 7.852482269503546e-05,
      "loss": 0.0616,
      "step": 3030
    },
    {
      "epoch": 21.56,
      "learning_rate": 7.845390070921986e-05,
      "loss": 0.0285,
      "step": 3040
    },
    {
      "epoch": 21.63,
      "learning_rate": 7.838297872340425e-05,
      "loss": 0.0139,
      "step": 3050
    },
    {
      "epoch": 21.7,
      "learning_rate": 7.831205673758866e-05,
      "loss": 0.0465,
      "step": 3060
    },
    {
      "epoch": 21.77,
      "learning_rate": 7.824113475177305e-05,
      "loss": 0.0542,
      "step": 3070
    },
    {
      "epoch": 21.84,
      "learning_rate": 7.817021276595745e-05,
      "loss": 0.0835,
      "step": 3080
    },
    {
      "epoch": 21.91,
      "learning_rate": 7.809929078014184e-05,
      "loss": 0.0596,
      "step": 3090
    },
    {
      "epoch": 21.99,
      "learning_rate": 7.802836879432625e-05,
      "loss": 0.012,
      "step": 3100
    },
    {
      "epoch": 21.99,
      "eval_accuracy": 0.9894179894179894,
      "eval_loss": 0.03947828337550163,
      "eval_runtime": 4.8825,
      "eval_samples_per_second": 116.128,
      "eval_steps_per_second": 14.542,
      "step": 3100
    },
    {
      "epoch": 22.06,
      "learning_rate": 7.795744680851064e-05,
      "loss": 0.0741,
      "step": 3110
    },
    {
      "epoch": 22.13,
      "learning_rate": 7.788652482269503e-05,
      "loss": 0.0112,
      "step": 3120
    },
    {
      "epoch": 22.2,
      "learning_rate": 7.781560283687943e-05,
      "loss": 0.015,
      "step": 3130
    },
    {
      "epoch": 22.27,
      "learning_rate": 7.774468085106383e-05,
      "loss": 0.0037,
      "step": 3140
    },
    {
      "epoch": 22.34,
      "learning_rate": 7.767375886524824e-05,
      "loss": 0.0021,
      "step": 3150
    },
    {
      "epoch": 22.41,
      "learning_rate": 7.760283687943262e-05,
      "loss": 0.0854,
      "step": 3160
    },
    {
      "epoch": 22.48,
      "learning_rate": 7.753191489361703e-05,
      "loss": 0.0451,
      "step": 3170
    },
    {
      "epoch": 22.55,
      "learning_rate": 7.746099290780142e-05,
      "loss": 0.0276,
      "step": 3180
    },
    {
      "epoch": 22.62,
      "learning_rate": 7.739007092198583e-05,
      "loss": 0.0297,
      "step": 3190
    },
    {
      "epoch": 22.7,
      "learning_rate": 7.731914893617021e-05,
      "loss": 0.0059,
      "step": 3200
    },
    {
      "epoch": 22.7,
      "eval_accuracy": 0.982363315696649,
      "eval_loss": 0.0684400424361229,
      "eval_runtime": 4.958,
      "eval_samples_per_second": 114.36,
      "eval_steps_per_second": 14.32,
      "step": 3200
    },
    {
      "epoch": 22.77,
      "learning_rate": 7.724822695035462e-05,
      "loss": 0.038,
      "step": 3210
    },
    {
      "epoch": 22.84,
      "learning_rate": 7.717730496453901e-05,
      "loss": 0.0066,
      "step": 3220
    },
    {
      "epoch": 22.91,
      "learning_rate": 7.710638297872342e-05,
      "loss": 0.0023,
      "step": 3230
    },
    {
      "epoch": 22.98,
      "learning_rate": 7.70354609929078e-05,
      "loss": 0.0053,
      "step": 3240
    },
    {
      "epoch": 23.05,
      "learning_rate": 7.69645390070922e-05,
      "loss": 0.0833,
      "step": 3250
    },
    {
      "epoch": 23.12,
      "learning_rate": 7.68936170212766e-05,
      "loss": 0.0016,
      "step": 3260
    },
    {
      "epoch": 23.19,
      "learning_rate": 7.6822695035461e-05,
      "loss": 0.0563,
      "step": 3270
    },
    {
      "epoch": 23.26,
      "learning_rate": 7.675177304964539e-05,
      "loss": 0.0051,
      "step": 3280
    },
    {
      "epoch": 23.33,
      "learning_rate": 7.668085106382979e-05,
      "loss": 0.0029,
      "step": 3290
    },
    {
      "epoch": 23.4,
      "learning_rate": 7.660992907801419e-05,
      "loss": 0.0657,
      "step": 3300
    },
    {
      "epoch": 23.4,
      "eval_accuracy": 0.9735449735449735,
      "eval_loss": 0.10394405573606491,
      "eval_runtime": 4.9825,
      "eval_samples_per_second": 113.797,
      "eval_steps_per_second": 14.25,
      "step": 3300
    },
    {
      "epoch": 23.48,
      "learning_rate": 7.65390070921986e-05,
      "loss": 0.0159,
      "step": 3310
    },
    {
      "epoch": 23.55,
      "learning_rate": 7.646808510638299e-05,
      "loss": 0.0245,
      "step": 3320
    },
    {
      "epoch": 23.62,
      "learning_rate": 7.639716312056738e-05,
      "loss": 0.0262,
      "step": 3330
    },
    {
      "epoch": 23.69,
      "learning_rate": 7.632624113475177e-05,
      "loss": 0.0342,
      "step": 3340
    },
    {
      "epoch": 23.76,
      "learning_rate": 7.625531914893618e-05,
      "loss": 0.002,
      "step": 3350
    },
    {
      "epoch": 23.83,
      "learning_rate": 7.618439716312058e-05,
      "loss": 0.005,
      "step": 3360
    },
    {
      "epoch": 23.9,
      "learning_rate": 7.611347517730497e-05,
      "loss": 0.0269,
      "step": 3370
    },
    {
      "epoch": 23.97,
      "learning_rate": 7.604255319148936e-05,
      "loss": 0.0055,
      "step": 3380
    },
    {
      "epoch": 24.04,
      "learning_rate": 7.597163120567377e-05,
      "loss": 0.0762,
      "step": 3390
    },
    {
      "epoch": 24.11,
      "learning_rate": 7.590070921985816e-05,
      "loss": 0.0198,
      "step": 3400
    },
    {
      "epoch": 24.11,
      "eval_accuracy": 0.9805996472663139,
      "eval_loss": 0.09220729023218155,
      "eval_runtime": 4.826,
      "eval_samples_per_second": 117.488,
      "eval_steps_per_second": 14.712,
      "step": 3400
    },
    {
      "epoch": 24.18,
      "learning_rate": 7.582978723404256e-05,
      "loss": 0.0705,
      "step": 3410
    },
    {
      "epoch": 24.26,
      "learning_rate": 7.575886524822695e-05,
      "loss": 0.0074,
      "step": 3420
    },
    {
      "epoch": 24.33,
      "learning_rate": 7.568794326241136e-05,
      "loss": 0.0536,
      "step": 3430
    },
    {
      "epoch": 24.4,
      "learning_rate": 7.561702127659575e-05,
      "loss": 0.0031,
      "step": 3440
    },
    {
      "epoch": 24.47,
      "learning_rate": 7.554609929078014e-05,
      "loss": 0.036,
      "step": 3450
    },
    {
      "epoch": 24.54,
      "learning_rate": 7.547517730496454e-05,
      "loss": 0.0025,
      "step": 3460
    },
    {
      "epoch": 24.61,
      "learning_rate": 7.540425531914895e-05,
      "loss": 0.0025,
      "step": 3470
    },
    {
      "epoch": 24.68,
      "learning_rate": 7.533333333333334e-05,
      "loss": 0.0637,
      "step": 3480
    },
    {
      "epoch": 24.75,
      "learning_rate": 7.526241134751773e-05,
      "loss": 0.0017,
      "step": 3490
    },
    {
      "epoch": 24.82,
      "learning_rate": 7.519148936170213e-05,
      "loss": 0.0657,
      "step": 3500
    },
    {
      "epoch": 24.82,
      "eval_accuracy": 0.9858906525573192,
      "eval_loss": 0.048500701785087585,
      "eval_runtime": 5.0205,
      "eval_samples_per_second": 112.936,
      "eval_steps_per_second": 14.142,
      "step": 3500
    },
    {
      "epoch": 24.89,
      "learning_rate": 7.512056737588653e-05,
      "loss": 0.0545,
      "step": 3510
    },
    {
      "epoch": 24.96,
      "learning_rate": 7.504964539007093e-05,
      "loss": 0.0184,
      "step": 3520
    },
    {
      "epoch": 25.04,
      "learning_rate": 7.497872340425532e-05,
      "loss": 0.1491,
      "step": 3530
    },
    {
      "epoch": 25.11,
      "learning_rate": 7.490780141843971e-05,
      "loss": 0.0862,
      "step": 3540
    },
    {
      "epoch": 25.18,
      "learning_rate": 7.483687943262412e-05,
      "loss": 0.0532,
      "step": 3550
    },
    {
      "epoch": 25.25,
      "learning_rate": 7.476595744680852e-05,
      "loss": 0.0437,
      "step": 3560
    },
    {
      "epoch": 25.32,
      "learning_rate": 7.469503546099291e-05,
      "loss": 0.0189,
      "step": 3570
    },
    {
      "epoch": 25.39,
      "learning_rate": 7.46241134751773e-05,
      "loss": 0.013,
      "step": 3580
    },
    {
      "epoch": 25.46,
      "learning_rate": 7.455319148936171e-05,
      "loss": 0.0178,
      "step": 3590
    },
    {
      "epoch": 25.53,
      "learning_rate": 7.44822695035461e-05,
      "loss": 0.0749,
      "step": 3600
    },
    {
      "epoch": 25.53,
      "eval_accuracy": 0.9858906525573192,
      "eval_loss": 0.05182398855686188,
      "eval_runtime": 4.88,
      "eval_samples_per_second": 116.187,
      "eval_steps_per_second": 14.549,
      "step": 3600
    },
    {
      "epoch": 25.6,
      "learning_rate": 7.44113475177305e-05,
      "loss": 0.0469,
      "step": 3610
    },
    {
      "epoch": 25.67,
      "learning_rate": 7.434042553191489e-05,
      "loss": 0.0346,
      "step": 3620
    },
    {
      "epoch": 25.74,
      "learning_rate": 7.42695035460993e-05,
      "loss": 0.0324,
      "step": 3630
    },
    {
      "epoch": 25.82,
      "learning_rate": 7.419858156028369e-05,
      "loss": 0.0021,
      "step": 3640
    },
    {
      "epoch": 25.89,
      "learning_rate": 7.412765957446809e-05,
      "loss": 0.0149,
      "step": 3650
    },
    {
      "epoch": 25.96,
      "learning_rate": 7.405673758865248e-05,
      "loss": 0.115,
      "step": 3660
    },
    {
      "epoch": 26.03,
      "learning_rate": 7.398581560283689e-05,
      "loss": 0.0316,
      "step": 3670
    },
    {
      "epoch": 26.1,
      "learning_rate": 7.391489361702128e-05,
      "loss": 0.0417,
      "step": 3680
    },
    {
      "epoch": 26.17,
      "learning_rate": 7.384397163120567e-05,
      "loss": 0.0943,
      "step": 3690
    },
    {
      "epoch": 26.24,
      "learning_rate": 7.377304964539008e-05,
      "loss": 0.0909,
      "step": 3700
    },
    {
      "epoch": 26.24,
      "eval_accuracy": 0.9929453262786596,
      "eval_loss": 0.0322270467877388,
      "eval_runtime": 4.85,
      "eval_samples_per_second": 116.906,
      "eval_steps_per_second": 14.639,
      "step": 3700
    },
    {
      "epoch": 26.31,
      "learning_rate": 7.370212765957447e-05,
      "loss": 0.0028,
      "step": 3710
    },
    {
      "epoch": 26.38,
      "learning_rate": 7.363120567375887e-05,
      "loss": 0.0625,
      "step": 3720
    },
    {
      "epoch": 26.45,
      "learning_rate": 7.356028368794326e-05,
      "loss": 0.0526,
      "step": 3730
    },
    {
      "epoch": 26.52,
      "learning_rate": 7.348936170212767e-05,
      "loss": 0.0272,
      "step": 3740
    },
    {
      "epoch": 26.6,
      "learning_rate": 7.341843971631206e-05,
      "loss": 0.0791,
      "step": 3750
    },
    {
      "epoch": 26.67,
      "learning_rate": 7.334751773049647e-05,
      "loss": 0.0018,
      "step": 3760
    },
    {
      "epoch": 26.74,
      "learning_rate": 7.327659574468085e-05,
      "loss": 0.1004,
      "step": 3770
    },
    {
      "epoch": 26.81,
      "learning_rate": 7.320567375886526e-05,
      "loss": 0.0096,
      "step": 3780
    },
    {
      "epoch": 26.88,
      "learning_rate": 7.313475177304965e-05,
      "loss": 0.0094,
      "step": 3790
    },
    {
      "epoch": 26.95,
      "learning_rate": 7.306382978723406e-05,
      "loss": 0.0688,
      "step": 3800
    },
    {
      "epoch": 26.95,
      "eval_accuracy": 0.9876543209876543,
      "eval_loss": 0.0609063059091568,
      "eval_runtime": 4.94,
      "eval_samples_per_second": 114.776,
      "eval_steps_per_second": 14.372,
      "step": 3800
    },
    {
      "epoch": 27.02,
      "learning_rate": 7.299290780141844e-05,
      "loss": 0.0015,
      "step": 3810
    },
    {
      "epoch": 27.09,
      "learning_rate": 7.292198581560284e-05,
      "loss": 0.007,
      "step": 3820
    },
    {
      "epoch": 27.16,
      "learning_rate": 7.285106382978724e-05,
      "loss": 0.0021,
      "step": 3830
    },
    {
      "epoch": 27.23,
      "learning_rate": 7.278014184397164e-05,
      "loss": 0.0342,
      "step": 3840
    },
    {
      "epoch": 27.3,
      "learning_rate": 7.270921985815603e-05,
      "loss": 0.0551,
      "step": 3850
    },
    {
      "epoch": 27.38,
      "learning_rate": 7.263829787234043e-05,
      "loss": 0.0155,
      "step": 3860
    },
    {
      "epoch": 27.45,
      "learning_rate": 7.256737588652483e-05,
      "loss": 0.0038,
      "step": 3870
    },
    {
      "epoch": 27.52,
      "learning_rate": 7.249645390070923e-05,
      "loss": 0.0548,
      "step": 3880
    },
    {
      "epoch": 27.59,
      "learning_rate": 7.242553191489361e-05,
      "loss": 0.0674,
      "step": 3890
    },
    {
      "epoch": 27.66,
      "learning_rate": 7.235460992907802e-05,
      "loss": 0.0163,
      "step": 3900
    },
    {
      "epoch": 27.66,
      "eval_accuracy": 0.9700176366843033,
      "eval_loss": 0.12531960010528564,
      "eval_runtime": 4.8785,
      "eval_samples_per_second": 116.223,
      "eval_steps_per_second": 14.554,
      "step": 3900
    },
    {
      "epoch": 27.73,
      "learning_rate": 7.228368794326241e-05,
      "loss": 0.0656,
      "step": 3910
    },
    {
      "epoch": 27.8,
      "learning_rate": 7.221276595744682e-05,
      "loss": 0.0163,
      "step": 3920
    },
    {
      "epoch": 27.87,
      "learning_rate": 7.21418439716312e-05,
      "loss": 0.0037,
      "step": 3930
    },
    {
      "epoch": 27.94,
      "learning_rate": 7.207092198581561e-05,
      "loss": 0.0409,
      "step": 3940
    },
    {
      "epoch": 28.01,
      "learning_rate": 7.2e-05,
      "loss": 0.0033,
      "step": 3950
    },
    {
      "epoch": 28.09,
      "learning_rate": 7.192907801418441e-05,
      "loss": 0.026,
      "step": 3960
    },
    {
      "epoch": 28.16,
      "learning_rate": 7.185815602836879e-05,
      "loss": 0.0373,
      "step": 3970
    },
    {
      "epoch": 28.23,
      "learning_rate": 7.17872340425532e-05,
      "loss": 0.0323,
      "step": 3980
    },
    {
      "epoch": 28.3,
      "learning_rate": 7.171631205673759e-05,
      "loss": 0.0354,
      "step": 3990
    },
    {
      "epoch": 28.37,
      "learning_rate": 7.1645390070922e-05,
      "loss": 0.0458,
      "step": 4000
    },
    {
      "epoch": 28.37,
      "eval_accuracy": 0.9982363315696648,
      "eval_loss": 0.006850986275821924,
      "eval_runtime": 4.764,
      "eval_samples_per_second": 119.017,
      "eval_steps_per_second": 14.903,
      "step": 4000
    },
    {
      "epoch": 28.44,
      "learning_rate": 7.157446808510638e-05,
      "loss": 0.0027,
      "step": 4010
    },
    {
      "epoch": 28.51,
      "learning_rate": 7.150354609929078e-05,
      "loss": 0.1157,
      "step": 4020
    },
    {
      "epoch": 28.58,
      "learning_rate": 7.143262411347518e-05,
      "loss": 0.0919,
      "step": 4030
    },
    {
      "epoch": 28.65,
      "learning_rate": 7.136170212765958e-05,
      "loss": 0.0063,
      "step": 4040
    },
    {
      "epoch": 28.72,
      "learning_rate": 7.129078014184397e-05,
      "loss": 0.069,
      "step": 4050
    },
    {
      "epoch": 28.79,
      "learning_rate": 7.121985815602837e-05,
      "loss": 0.0034,
      "step": 4060
    },
    {
      "epoch": 28.87,
      "learning_rate": 7.114893617021277e-05,
      "loss": 0.0278,
      "step": 4070
    },
    {
      "epoch": 28.94,
      "learning_rate": 7.107801418439717e-05,
      "loss": 0.0084,
      "step": 4080
    },
    {
      "epoch": 29.01,
      "learning_rate": 7.100709219858155e-05,
      "loss": 0.012,
      "step": 4090
    },
    {
      "epoch": 29.08,
      "learning_rate": 7.093617021276596e-05,
      "loss": 0.0728,
      "step": 4100
    },
    {
      "epoch": 29.08,
      "eval_accuracy": 0.9894179894179894,
      "eval_loss": 0.03627217188477516,
      "eval_runtime": 4.911,
      "eval_samples_per_second": 115.454,
      "eval_steps_per_second": 14.457,
      "step": 4100
    },
    {
      "epoch": 29.15,
      "learning_rate": 7.086524822695035e-05,
      "loss": 0.0034,
      "step": 4110
    },
    {
      "epoch": 29.22,
      "learning_rate": 7.079432624113476e-05,
      "loss": 0.0015,
      "step": 4120
    },
    {
      "epoch": 29.29,
      "learning_rate": 7.072340425531914e-05,
      "loss": 0.0416,
      "step": 4130
    },
    {
      "epoch": 29.36,
      "learning_rate": 7.065248226950355e-05,
      "loss": 0.0031,
      "step": 4140
    },
    {
      "epoch": 29.43,
      "learning_rate": 7.058156028368794e-05,
      "loss": 0.0031,
      "step": 4150
    },
    {
      "epoch": 29.5,
      "learning_rate": 7.051063829787235e-05,
      "loss": 0.0256,
      "step": 4160
    },
    {
      "epoch": 29.57,
      "learning_rate": 7.043971631205674e-05,
      "loss": 0.001,
      "step": 4170
    },
    {
      "epoch": 29.65,
      "learning_rate": 7.036879432624114e-05,
      "loss": 0.0613,
      "step": 4180
    },
    {
      "epoch": 29.72,
      "learning_rate": 7.029787234042553e-05,
      "loss": 0.0031,
      "step": 4190
    },
    {
      "epoch": 29.79,
      "learning_rate": 7.022695035460994e-05,
      "loss": 0.1474,
      "step": 4200
    },
    {
      "epoch": 29.79,
      "eval_accuracy": 0.9894179894179894,
      "eval_loss": 0.08292630314826965,
      "eval_runtime": 4.8845,
      "eval_samples_per_second": 116.08,
      "eval_steps_per_second": 14.536,
      "step": 4200
    },
    {
      "epoch": 29.86,
      "learning_rate": 7.015602836879433e-05,
      "loss": 0.001,
      "step": 4210
    },
    {
      "epoch": 29.93,
      "learning_rate": 7.008510638297872e-05,
      "loss": 0.0717,
      "step": 4220
    },
    {
      "epoch": 30.0,
      "learning_rate": 7.001418439716312e-05,
      "loss": 0.0246,
      "step": 4230
    },
    {
      "epoch": 30.07,
      "learning_rate": 6.994326241134753e-05,
      "loss": 0.0333,
      "step": 4240
    },
    {
      "epoch": 30.14,
      "learning_rate": 6.987234042553192e-05,
      "loss": 0.0012,
      "step": 4250
    },
    {
      "epoch": 30.21,
      "learning_rate": 6.980141843971631e-05,
      "loss": 0.0737,
      "step": 4260
    },
    {
      "epoch": 30.28,
      "learning_rate": 6.973049645390072e-05,
      "loss": 0.001,
      "step": 4270
    },
    {
      "epoch": 30.35,
      "learning_rate": 6.965957446808511e-05,
      "loss": 0.0071,
      "step": 4280
    },
    {
      "epoch": 30.43,
      "learning_rate": 6.95886524822695e-05,
      "loss": 0.0252,
      "step": 4290
    },
    {
      "epoch": 30.5,
      "learning_rate": 6.95177304964539e-05,
      "loss": 0.0011,
      "step": 4300
    },
    {
      "epoch": 30.5,
      "eval_accuracy": 0.9858906525573192,
      "eval_loss": 0.07146803289651871,
      "eval_runtime": 4.945,
      "eval_samples_per_second": 114.66,
      "eval_steps_per_second": 14.358,
      "step": 4300
    },
    {
      "epoch": 30.57,
      "learning_rate": 6.944680851063831e-05,
      "loss": 0.0327,
      "step": 4310
    },
    {
      "epoch": 30.64,
      "learning_rate": 6.93758865248227e-05,
      "loss": 0.021,
      "step": 4320
    },
    {
      "epoch": 30.71,
      "learning_rate": 6.93049645390071e-05,
      "loss": 0.0422,
      "step": 4330
    },
    {
      "epoch": 30.78,
      "learning_rate": 6.923404255319149e-05,
      "loss": 0.1148,
      "step": 4340
    },
    {
      "epoch": 30.85,
      "learning_rate": 6.91631205673759e-05,
      "loss": 0.0802,
      "step": 4350
    },
    {
      "epoch": 30.92,
      "learning_rate": 6.909219858156029e-05,
      "loss": 0.0624,
      "step": 4360
    },
    {
      "epoch": 30.99,
      "learning_rate": 6.902127659574468e-05,
      "loss": 0.0206,
      "step": 4370
    },
    {
      "epoch": 31.06,
      "learning_rate": 6.895035460992908e-05,
      "loss": 0.0265,
      "step": 4380
    },
    {
      "epoch": 31.13,
      "learning_rate": 6.887943262411348e-05,
      "loss": 0.0684,
      "step": 4390
    },
    {
      "epoch": 31.21,
      "learning_rate": 6.880851063829788e-05,
      "loss": 0.1199,
      "step": 4400
    },
    {
      "epoch": 31.21,
      "eval_accuracy": 0.9735449735449735,
      "eval_loss": 0.14859944581985474,
      "eval_runtime": 4.6535,
      "eval_samples_per_second": 121.843,
      "eval_steps_per_second": 15.257,
      "step": 4400
    },
    {
      "epoch": 31.28,
      "learning_rate": 6.873758865248227e-05,
      "loss": 0.1303,
      "step": 4410
    },
    {
      "epoch": 31.35,
      "learning_rate": 6.866666666666666e-05,
      "loss": 0.0416,
      "step": 4420
    },
    {
      "epoch": 31.42,
      "learning_rate": 6.859574468085107e-05,
      "loss": 0.0012,
      "step": 4430
    },
    {
      "epoch": 31.49,
      "learning_rate": 6.852482269503547e-05,
      "loss": 0.0821,
      "step": 4440
    },
    {
      "epoch": 31.56,
      "learning_rate": 6.845390070921986e-05,
      "loss": 0.0271,
      "step": 4450
    },
    {
      "epoch": 31.63,
      "learning_rate": 6.838297872340425e-05,
      "loss": 0.0103,
      "step": 4460
    },
    {
      "epoch": 31.7,
      "learning_rate": 6.831205673758866e-05,
      "loss": 0.0016,
      "step": 4470
    },
    {
      "epoch": 31.77,
      "learning_rate": 6.824113475177305e-05,
      "loss": 0.036,
      "step": 4480
    },
    {
      "epoch": 31.84,
      "learning_rate": 6.817021276595745e-05,
      "loss": 0.0018,
      "step": 4490
    },
    {
      "epoch": 31.91,
      "learning_rate": 6.809929078014184e-05,
      "loss": 0.003,
      "step": 4500
    },
    {
      "epoch": 31.91,
      "eval_accuracy": 0.9841269841269841,
      "eval_loss": 0.08768551796674728,
      "eval_runtime": 4.81,
      "eval_samples_per_second": 117.879,
      "eval_steps_per_second": 14.761,
      "step": 4500
    },
    {
      "epoch": 31.99,
      "learning_rate": 6.802836879432625e-05,
      "loss": 0.0019,
      "step": 4510
    },
    {
      "epoch": 32.06,
      "learning_rate": 6.795744680851064e-05,
      "loss": 0.0646,
      "step": 4520
    },
    {
      "epoch": 32.13,
      "learning_rate": 6.788652482269503e-05,
      "loss": 0.0018,
      "step": 4530
    },
    {
      "epoch": 32.2,
      "learning_rate": 6.781560283687943e-05,
      "loss": 0.0381,
      "step": 4540
    },
    {
      "epoch": 32.27,
      "learning_rate": 6.774468085106384e-05,
      "loss": 0.0053,
      "step": 4550
    },
    {
      "epoch": 32.34,
      "learning_rate": 6.767375886524823e-05,
      "loss": 0.0357,
      "step": 4560
    },
    {
      "epoch": 32.41,
      "learning_rate": 6.760283687943262e-05,
      "loss": 0.008,
      "step": 4570
    },
    {
      "epoch": 32.48,
      "learning_rate": 6.753191489361702e-05,
      "loss": 0.0378,
      "step": 4580
    },
    {
      "epoch": 32.55,
      "learning_rate": 6.746099290780142e-05,
      "loss": 0.001,
      "step": 4590
    },
    {
      "epoch": 32.62,
      "learning_rate": 6.739007092198582e-05,
      "loss": 0.0098,
      "step": 4600
    },
    {
      "epoch": 32.62,
      "eval_accuracy": 0.9858906525573192,
      "eval_loss": 0.06747796386480331,
      "eval_runtime": 4.862,
      "eval_samples_per_second": 116.618,
      "eval_steps_per_second": 14.603,
      "step": 4600
    },
    {
      "epoch": 32.7,
      "learning_rate": 6.731914893617022e-05,
      "loss": 0.0011,
      "step": 4610
    },
    {
      "epoch": 32.77,
      "learning_rate": 6.72482269503546e-05,
      "loss": 0.0434,
      "step": 4620
    },
    {
      "epoch": 32.84,
      "learning_rate": 6.717730496453901e-05,
      "loss": 0.0545,
      "step": 4630
    },
    {
      "epoch": 32.91,
      "learning_rate": 6.71063829787234e-05,
      "loss": 0.0032,
      "step": 4640
    },
    {
      "epoch": 32.98,
      "learning_rate": 6.703546099290781e-05,
      "loss": 0.0259,
      "step": 4650
    },
    {
      "epoch": 33.05,
      "learning_rate": 6.696453900709219e-05,
      "loss": 0.0068,
      "step": 4660
    },
    {
      "epoch": 33.12,
      "learning_rate": 6.68936170212766e-05,
      "loss": 0.0493,
      "step": 4670
    },
    {
      "epoch": 33.19,
      "learning_rate": 6.682269503546099e-05,
      "loss": 0.0999,
      "step": 4680
    },
    {
      "epoch": 33.26,
      "learning_rate": 6.67517730496454e-05,
      "loss": 0.0049,
      "step": 4690
    },
    {
      "epoch": 33.33,
      "learning_rate": 6.668085106382978e-05,
      "loss": 0.0028,
      "step": 4700
    },
    {
      "epoch": 33.33,
      "eval_accuracy": 0.9876543209876543,
      "eval_loss": 0.08373798429965973,
      "eval_runtime": 4.809,
      "eval_samples_per_second": 117.903,
      "eval_steps_per_second": 14.764,
      "step": 4700
    },
    {
      "epoch": 33.4,
      "learning_rate": 6.660992907801419e-05,
      "loss": 0.0795,
      "step": 4710
    },
    {
      "epoch": 33.48,
      "learning_rate": 6.653900709219858e-05,
      "loss": 0.0605,
      "step": 4720
    },
    {
      "epoch": 33.55,
      "learning_rate": 6.646808510638299e-05,
      "loss": 0.0758,
      "step": 4730
    },
    {
      "epoch": 33.62,
      "learning_rate": 6.639716312056737e-05,
      "loss": 0.0798,
      "step": 4740
    },
    {
      "epoch": 33.69,
      "learning_rate": 6.632624113475178e-05,
      "loss": 0.0036,
      "step": 4750
    },
    {
      "epoch": 33.76,
      "learning_rate": 6.625531914893617e-05,
      "loss": 0.0013,
      "step": 4760
    },
    {
      "epoch": 33.83,
      "learning_rate": 6.618439716312058e-05,
      "loss": 0.0289,
      "step": 4770
    },
    {
      "epoch": 33.9,
      "learning_rate": 6.611347517730497e-05,
      "loss": 0.0408,
      "step": 4780
    },
    {
      "epoch": 33.97,
      "learning_rate": 6.604255319148936e-05,
      "loss": 0.0031,
      "step": 4790
    },
    {
      "epoch": 34.04,
      "learning_rate": 6.597163120567377e-05,
      "loss": 0.0269,
      "step": 4800
    },
    {
      "epoch": 34.04,
      "eval_accuracy": 0.9805996472663139,
      "eval_loss": 0.08315479755401611,
      "eval_runtime": 4.926,
      "eval_samples_per_second": 115.103,
      "eval_steps_per_second": 14.413,
      "step": 4800
    },
    {
      "epoch": 34.11,
      "learning_rate": 6.590070921985816e-05,
      "loss": 0.0023,
      "step": 4810
    },
    {
      "epoch": 34.18,
      "learning_rate": 6.582978723404256e-05,
      "loss": 0.0211,
      "step": 4820
    },
    {
      "epoch": 34.26,
      "learning_rate": 6.575886524822695e-05,
      "loss": 0.0012,
      "step": 4830
    },
    {
      "epoch": 34.33,
      "learning_rate": 6.568794326241136e-05,
      "loss": 0.0007,
      "step": 4840
    },
    {
      "epoch": 34.4,
      "learning_rate": 6.561702127659575e-05,
      "loss": 0.0007,
      "step": 4850
    },
    {
      "epoch": 34.47,
      "learning_rate": 6.554609929078015e-05,
      "loss": 0.0007,
      "step": 4860
    },
    {
      "epoch": 34.54,
      "learning_rate": 6.547517730496454e-05,
      "loss": 0.038,
      "step": 4870
    },
    {
      "epoch": 34.61,
      "learning_rate": 6.540425531914895e-05,
      "loss": 0.0072,
      "step": 4880
    },
    {
      "epoch": 34.68,
      "learning_rate": 6.533333333333334e-05,
      "loss": 0.0007,
      "step": 4890
    },
    {
      "epoch": 34.75,
      "learning_rate": 6.526241134751773e-05,
      "loss": 0.0028,
      "step": 4900
    },
    {
      "epoch": 34.75,
      "eval_accuracy": 0.9876543209876543,
      "eval_loss": 0.06866313517093658,
      "eval_runtime": 4.82,
      "eval_samples_per_second": 117.634,
      "eval_steps_per_second": 14.73,
      "step": 4900
    },
    {
      "epoch": 34.82,
      "learning_rate": 6.519148936170213e-05,
      "loss": 0.0202,
      "step": 4910
    },
    {
      "epoch": 34.89,
      "learning_rate": 6.512056737588653e-05,
      "loss": 0.0592,
      "step": 4920
    },
    {
      "epoch": 34.96,
      "learning_rate": 6.504964539007093e-05,
      "loss": 0.0196,
      "step": 4930
    },
    {
      "epoch": 35.04,
      "learning_rate": 6.497872340425532e-05,
      "loss": 0.0336,
      "step": 4940
    },
    {
      "epoch": 35.11,
      "learning_rate": 6.490780141843972e-05,
      "loss": 0.0212,
      "step": 4950
    },
    {
      "epoch": 35.18,
      "learning_rate": 6.483687943262412e-05,
      "loss": 0.0013,
      "step": 4960
    },
    {
      "epoch": 35.25,
      "learning_rate": 6.476595744680852e-05,
      "loss": 0.0024,
      "step": 4970
    },
    {
      "epoch": 35.32,
      "learning_rate": 6.469503546099291e-05,
      "loss": 0.0427,
      "step": 4980
    },
    {
      "epoch": 35.39,
      "learning_rate": 6.46241134751773e-05,
      "loss": 0.0008,
      "step": 4990
    },
    {
      "epoch": 35.46,
      "learning_rate": 6.455319148936171e-05,
      "loss": 0.0016,
      "step": 5000
    },
    {
      "epoch": 35.46,
      "eval_accuracy": 0.9929453262786596,
      "eval_loss": 0.0424884632229805,
      "eval_runtime": 4.8085,
      "eval_samples_per_second": 117.915,
      "eval_steps_per_second": 14.765,
      "step": 5000
    },
    {
      "epoch": 35.53,
      "learning_rate": 6.44822695035461e-05,
      "loss": 0.0147,
      "step": 5010
    },
    {
      "epoch": 35.6,
      "learning_rate": 6.44113475177305e-05,
      "loss": 0.0009,
      "step": 5020
    },
    {
      "epoch": 35.67,
      "learning_rate": 6.434042553191489e-05,
      "loss": 0.0012,
      "step": 5030
    },
    {
      "epoch": 35.74,
      "learning_rate": 6.42695035460993e-05,
      "loss": 0.0008,
      "step": 5040
    },
    {
      "epoch": 35.82,
      "learning_rate": 6.419858156028369e-05,
      "loss": 0.001,
      "step": 5050
    },
    {
      "epoch": 35.89,
      "learning_rate": 6.412765957446809e-05,
      "loss": 0.0712,
      "step": 5060
    },
    {
      "epoch": 35.96,
      "learning_rate": 6.405673758865248e-05,
      "loss": 0.076,
      "step": 5070
    },
    {
      "epoch": 36.03,
      "learning_rate": 6.398581560283689e-05,
      "loss": 0.0092,
      "step": 5080
    },
    {
      "epoch": 36.1,
      "learning_rate": 6.391489361702128e-05,
      "loss": 0.1197,
      "step": 5090
    },
    {
      "epoch": 36.17,
      "learning_rate": 6.384397163120567e-05,
      "loss": 0.0262,
      "step": 5100
    },
    {
      "epoch": 36.17,
      "eval_accuracy": 0.9841269841269841,
      "eval_loss": 0.08268722891807556,
      "eval_runtime": 4.865,
      "eval_samples_per_second": 116.546,
      "eval_steps_per_second": 14.594,
      "step": 5100
    },
    {
      "epoch": 36.24,
      "learning_rate": 6.377304964539007e-05,
      "loss": 0.0717,
      "step": 5110
    },
    {
      "epoch": 36.31,
      "learning_rate": 6.370212765957447e-05,
      "loss": 0.0486,
      "step": 5120
    },
    {
      "epoch": 36.38,
      "learning_rate": 6.363120567375887e-05,
      "loss": 0.049,
      "step": 5130
    },
    {
      "epoch": 36.45,
      "learning_rate": 6.356737588652482e-05,
      "loss": 0.0872,
      "step": 5140
    },
    {
      "epoch": 36.52,
      "learning_rate": 6.349645390070922e-05,
      "loss": 0.0394,
      "step": 5150
    },
    {
      "epoch": 36.6,
      "learning_rate": 6.342553191489362e-05,
      "loss": 0.0489,
      "step": 5160
    },
    {
      "epoch": 36.67,
      "learning_rate": 6.335460992907802e-05,
      "loss": 0.0121,
      "step": 5170
    },
    {
      "epoch": 36.74,
      "learning_rate": 6.328368794326241e-05,
      "loss": 0.0255,
      "step": 5180
    },
    {
      "epoch": 36.81,
      "learning_rate": 6.32127659574468e-05,
      "loss": 0.0584,
      "step": 5190
    },
    {
      "epoch": 36.88,
      "learning_rate": 6.314184397163121e-05,
      "loss": 0.0397,
      "step": 5200
    },
    {
      "epoch": 36.88,
      "eval_accuracy": 0.9753086419753086,
      "eval_loss": 0.15599842369556427,
      "eval_runtime": 4.811,
      "eval_samples_per_second": 117.854,
      "eval_steps_per_second": 14.758,
      "step": 5200
    },
    {
      "epoch": 36.95,
      "learning_rate": 6.30709219858156e-05,
      "loss": 0.0908,
      "step": 5210
    },
    {
      "epoch": 37.02,
      "learning_rate": 6.3e-05,
      "loss": 0.0488,
      "step": 5220
    },
    {
      "epoch": 37.09,
      "learning_rate": 6.29290780141844e-05,
      "loss": 0.0014,
      "step": 5230
    },
    {
      "epoch": 37.16,
      "learning_rate": 6.28581560283688e-05,
      "loss": 0.003,
      "step": 5240
    },
    {
      "epoch": 37.23,
      "learning_rate": 6.27872340425532e-05,
      "loss": 0.0039,
      "step": 5250
    },
    {
      "epoch": 37.3,
      "learning_rate": 6.271631205673759e-05,
      "loss": 0.0154,
      "step": 5260
    },
    {
      "epoch": 37.38,
      "learning_rate": 6.2645390070922e-05,
      "loss": 0.0412,
      "step": 5270
    },
    {
      "epoch": 37.45,
      "learning_rate": 6.257446808510639e-05,
      "loss": 0.0007,
      "step": 5280
    },
    {
      "epoch": 37.52,
      "learning_rate": 6.250354609929078e-05,
      "loss": 0.0178,
      "step": 5290
    },
    {
      "epoch": 37.59,
      "learning_rate": 6.243262411347518e-05,
      "loss": 0.0204,
      "step": 5300
    },
    {
      "epoch": 37.59,
      "eval_accuracy": 0.9876543209876543,
      "eval_loss": 0.07057899981737137,
      "eval_runtime": 4.7545,
      "eval_samples_per_second": 119.254,
      "eval_steps_per_second": 14.933,
      "step": 5300
    },
    {
      "epoch": 37.66,
      "learning_rate": 6.236170212765958e-05,
      "loss": 0.0513,
      "step": 5310
    },
    {
      "epoch": 37.73,
      "learning_rate": 6.229078014184398e-05,
      "loss": 0.0071,
      "step": 5320
    },
    {
      "epoch": 37.8,
      "learning_rate": 6.221985815602837e-05,
      "loss": 0.0691,
      "step": 5330
    },
    {
      "epoch": 37.87,
      "learning_rate": 6.214893617021276e-05,
      "loss": 0.0135,
      "step": 5340
    },
    {
      "epoch": 37.94,
      "learning_rate": 6.207801418439717e-05,
      "loss": 0.0033,
      "step": 5350
    },
    {
      "epoch": 38.01,
      "learning_rate": 6.200709219858156e-05,
      "loss": 0.0695,
      "step": 5360
    },
    {
      "epoch": 38.09,
      "learning_rate": 6.193617021276596e-05,
      "loss": 0.064,
      "step": 5370
    },
    {
      "epoch": 38.16,
      "learning_rate": 6.186524822695035e-05,
      "loss": 0.0013,
      "step": 5380
    },
    {
      "epoch": 38.23,
      "learning_rate": 6.179432624113476e-05,
      "loss": 0.0264,
      "step": 5390
    },
    {
      "epoch": 38.3,
      "learning_rate": 6.172340425531915e-05,
      "loss": 0.0146,
      "step": 5400
    },
    {
      "epoch": 38.3,
      "eval_accuracy": 0.9858906525573192,
      "eval_loss": 0.06010299175977707,
      "eval_runtime": 4.902,
      "eval_samples_per_second": 115.666,
      "eval_steps_per_second": 14.484,
      "step": 5400
    },
    {
      "epoch": 38.37,
      "learning_rate": 6.165248226950356e-05,
      "loss": 0.0185,
      "step": 5410
    },
    {
      "epoch": 38.44,
      "learning_rate": 6.158156028368794e-05,
      "loss": 0.0145,
      "step": 5420
    },
    {
      "epoch": 38.51,
      "learning_rate": 6.151063829787235e-05,
      "loss": 0.003,
      "step": 5430
    },
    {
      "epoch": 38.58,
      "learning_rate": 6.143971631205674e-05,
      "loss": 0.0207,
      "step": 5440
    },
    {
      "epoch": 38.65,
      "learning_rate": 6.136879432624115e-05,
      "loss": 0.1072,
      "step": 5450
    },
    {
      "epoch": 38.72,
      "learning_rate": 6.129787234042553e-05,
      "loss": 0.0283,
      "step": 5460
    },
    {
      "epoch": 38.79,
      "learning_rate": 6.122695035460994e-05,
      "loss": 0.0006,
      "step": 5470
    },
    {
      "epoch": 38.87,
      "learning_rate": 6.115602836879433e-05,
      "loss": 0.0007,
      "step": 5480
    },
    {
      "epoch": 38.94,
      "learning_rate": 6.108510638297874e-05,
      "loss": 0.0094,
      "step": 5490
    },
    {
      "epoch": 39.01,
      "learning_rate": 6.101418439716312e-05,
      "loss": 0.0288,
      "step": 5500
    },
    {
      "epoch": 39.01,
      "eval_accuracy": 0.9929453262786596,
      "eval_loss": 0.03392454981803894,
      "eval_runtime": 4.849,
      "eval_samples_per_second": 116.93,
      "eval_steps_per_second": 14.642,
      "step": 5500
    },
    {
      "epoch": 39.08,
      "learning_rate": 6.094326241134752e-05,
      "loss": 0.0007,
      "step": 5510
    },
    {
      "epoch": 39.15,
      "learning_rate": 6.087234042553192e-05,
      "loss": 0.001,
      "step": 5520
    },
    {
      "epoch": 39.22,
      "learning_rate": 6.080141843971632e-05,
      "loss": 0.0709,
      "step": 5530
    },
    {
      "epoch": 39.29,
      "learning_rate": 6.073049645390071e-05,
      "loss": 0.0006,
      "step": 5540
    },
    {
      "epoch": 39.36,
      "learning_rate": 6.065957446808511e-05,
      "loss": 0.0255,
      "step": 5550
    },
    {
      "epoch": 39.43,
      "learning_rate": 6.0588652482269505e-05,
      "loss": 0.0448,
      "step": 5560
    },
    {
      "epoch": 39.5,
      "learning_rate": 6.0517730496453905e-05,
      "loss": 0.0056,
      "step": 5570
    },
    {
      "epoch": 39.57,
      "learning_rate": 6.04468085106383e-05,
      "loss": 0.0006,
      "step": 5580
    },
    {
      "epoch": 39.65,
      "learning_rate": 6.03758865248227e-05,
      "loss": 0.0006,
      "step": 5590
    },
    {
      "epoch": 39.72,
      "learning_rate": 6.030496453900709e-05,
      "loss": 0.0009,
      "step": 5600
    },
    {
      "epoch": 39.72,
      "eval_accuracy": 0.9911816578483245,
      "eval_loss": 0.05390332266688347,
      "eval_runtime": 4.9225,
      "eval_samples_per_second": 115.184,
      "eval_steps_per_second": 14.423,
      "step": 5600
    },
    {
      "epoch": 39.79,
      "learning_rate": 6.023404255319149e-05,
      "loss": 0.0287,
      "step": 5610
    },
    {
      "epoch": 39.86,
      "learning_rate": 6.016312056737589e-05,
      "loss": 0.0005,
      "step": 5620
    },
    {
      "epoch": 39.93,
      "learning_rate": 6.009219858156029e-05,
      "loss": 0.0243,
      "step": 5630
    },
    {
      "epoch": 40.0,
      "learning_rate": 6.002127659574468e-05,
      "loss": 0.0007,
      "step": 5640
    },
    {
      "epoch": 40.07,
      "learning_rate": 5.995035460992908e-05,
      "loss": 0.0372,
      "step": 5650
    },
    {
      "epoch": 40.14,
      "learning_rate": 5.9879432624113475e-05,
      "loss": 0.0366,
      "step": 5660
    },
    {
      "epoch": 40.21,
      "learning_rate": 5.9808510638297875e-05,
      "loss": 0.0693,
      "step": 5670
    },
    {
      "epoch": 40.28,
      "learning_rate": 5.973758865248227e-05,
      "loss": 0.0025,
      "step": 5680
    },
    {
      "epoch": 40.35,
      "learning_rate": 5.966666666666667e-05,
      "loss": 0.0091,
      "step": 5690
    },
    {
      "epoch": 40.43,
      "learning_rate": 5.959574468085106e-05,
      "loss": 0.0492,
      "step": 5700
    },
    {
      "epoch": 40.43,
      "eval_accuracy": 0.9876543209876543,
      "eval_loss": 0.04930723458528519,
      "eval_runtime": 4.869,
      "eval_samples_per_second": 116.45,
      "eval_steps_per_second": 14.582,
      "step": 5700
    },
    {
      "epoch": 40.5,
      "learning_rate": 5.952482269503546e-05,
      "loss": 0.0008,
      "step": 5710
    },
    {
      "epoch": 40.57,
      "learning_rate": 5.945390070921986e-05,
      "loss": 0.0086,
      "step": 5720
    },
    {
      "epoch": 40.64,
      "learning_rate": 5.938297872340426e-05,
      "loss": 0.0328,
      "step": 5730
    },
    {
      "epoch": 40.71,
      "learning_rate": 5.931205673758865e-05,
      "loss": 0.0011,
      "step": 5740
    },
    {
      "epoch": 40.78,
      "learning_rate": 5.924113475177305e-05,
      "loss": 0.0077,
      "step": 5750
    },
    {
      "epoch": 40.85,
      "learning_rate": 5.9170212765957445e-05,
      "loss": 0.0641,
      "step": 5760
    },
    {
      "epoch": 40.92,
      "learning_rate": 5.9099290780141845e-05,
      "loss": 0.0005,
      "step": 5770
    },
    {
      "epoch": 40.99,
      "learning_rate": 5.902836879432624e-05,
      "loss": 0.0007,
      "step": 5780
    },
    {
      "epoch": 41.06,
      "learning_rate": 5.895744680851064e-05,
      "loss": 0.0722,
      "step": 5790
    },
    {
      "epoch": 41.13,
      "learning_rate": 5.888652482269503e-05,
      "loss": 0.0275,
      "step": 5800
    },
    {
      "epoch": 41.13,
      "eval_accuracy": 0.9929453262786596,
      "eval_loss": 0.03137438744306564,
      "eval_runtime": 4.7505,
      "eval_samples_per_second": 119.355,
      "eval_steps_per_second": 14.946,
      "step": 5800
    },
    {
      "epoch": 41.21,
      "learning_rate": 5.881560283687943e-05,
      "loss": 0.0332,
      "step": 5810
    },
    {
      "epoch": 41.28,
      "learning_rate": 5.874468085106384e-05,
      "loss": 0.0467,
      "step": 5820
    },
    {
      "epoch": 41.35,
      "learning_rate": 5.867375886524823e-05,
      "loss": 0.0108,
      "step": 5830
    },
    {
      "epoch": 41.42,
      "learning_rate": 5.8602836879432634e-05,
      "loss": 0.0548,
      "step": 5840
    },
    {
      "epoch": 41.49,
      "learning_rate": 5.853191489361702e-05,
      "loss": 0.0569,
      "step": 5850
    },
    {
      "epoch": 41.56,
      "learning_rate": 5.846099290780143e-05,
      "loss": 0.0009,
      "step": 5860
    },
    {
      "epoch": 41.63,
      "learning_rate": 5.8390070921985815e-05,
      "loss": 0.0383,
      "step": 5870
    },
    {
      "epoch": 41.7,
      "learning_rate": 5.831914893617022e-05,
      "loss": 0.0294,
      "step": 5880
    },
    {
      "epoch": 41.77,
      "learning_rate": 5.824822695035461e-05,
      "loss": 0.014,
      "step": 5890
    },
    {
      "epoch": 41.84,
      "learning_rate": 5.8177304964539016e-05,
      "loss": 0.0192,
      "step": 5900
    },
    {
      "epoch": 41.84,
      "eval_accuracy": 0.9805996472663139,
      "eval_loss": 0.1111595556139946,
      "eval_runtime": 4.7365,
      "eval_samples_per_second": 119.708,
      "eval_steps_per_second": 14.99,
      "step": 5900
    },
    {
      "epoch": 41.91,
      "learning_rate": 5.81063829787234e-05,
      "loss": 0.0577,
      "step": 5910
    },
    {
      "epoch": 41.99,
      "learning_rate": 5.803546099290781e-05,
      "loss": 0.0911,
      "step": 5920
    },
    {
      "epoch": 42.06,
      "learning_rate": 5.79645390070922e-05,
      "loss": 0.0007,
      "step": 5930
    },
    {
      "epoch": 42.13,
      "learning_rate": 5.7893617021276604e-05,
      "loss": 0.0281,
      "step": 5940
    },
    {
      "epoch": 42.2,
      "learning_rate": 5.782269503546099e-05,
      "loss": 0.0429,
      "step": 5950
    },
    {
      "epoch": 42.27,
      "learning_rate": 5.77517730496454e-05,
      "loss": 0.0224,
      "step": 5960
    },
    {
      "epoch": 42.34,
      "learning_rate": 5.7680851063829785e-05,
      "loss": 0.0216,
      "step": 5970
    },
    {
      "epoch": 42.41,
      "learning_rate": 5.760992907801419e-05,
      "loss": 0.0015,
      "step": 5980
    },
    {
      "epoch": 42.48,
      "learning_rate": 5.753900709219858e-05,
      "loss": 0.0052,
      "step": 5990
    },
    {
      "epoch": 42.55,
      "learning_rate": 5.7468085106382986e-05,
      "loss": 0.0013,
      "step": 6000
    },
    {
      "epoch": 42.55,
      "eval_accuracy": 0.9894179894179894,
      "eval_loss": 0.06109577417373657,
      "eval_runtime": 4.7055,
      "eval_samples_per_second": 120.496,
      "eval_steps_per_second": 15.089,
      "step": 6000
    },
    {
      "epoch": 42.62,
      "learning_rate": 5.739716312056738e-05,
      "loss": 0.0009,
      "step": 6010
    },
    {
      "epoch": 42.7,
      "learning_rate": 5.732624113475178e-05,
      "loss": 0.0016,
      "step": 6020
    },
    {
      "epoch": 42.77,
      "learning_rate": 5.7255319148936174e-05,
      "loss": 0.0014,
      "step": 6030
    },
    {
      "epoch": 42.84,
      "learning_rate": 5.7184397163120574e-05,
      "loss": 0.0007,
      "step": 6040
    },
    {
      "epoch": 42.91,
      "learning_rate": 5.711347517730497e-05,
      "loss": 0.0062,
      "step": 6050
    },
    {
      "epoch": 42.98,
      "learning_rate": 5.704255319148937e-05,
      "loss": 0.0006,
      "step": 6060
    },
    {
      "epoch": 43.05,
      "learning_rate": 5.697163120567376e-05,
      "loss": 0.0023,
      "step": 6070
    },
    {
      "epoch": 43.12,
      "learning_rate": 5.690070921985816e-05,
      "loss": 0.0004,
      "step": 6080
    },
    {
      "epoch": 43.19,
      "learning_rate": 5.6829787234042556e-05,
      "loss": 0.0005,
      "step": 6090
    },
    {
      "epoch": 43.26,
      "learning_rate": 5.6758865248226956e-05,
      "loss": 0.0208,
      "step": 6100
    },
    {
      "epoch": 43.26,
      "eval_accuracy": 0.9947089947089947,
      "eval_loss": 0.013094047084450722,
      "eval_runtime": 4.8125,
      "eval_samples_per_second": 117.817,
      "eval_steps_per_second": 14.753,
      "step": 6100
    },
    {
      "epoch": 43.33,
      "learning_rate": 5.668794326241135e-05,
      "loss": 0.0005,
      "step": 6110
    },
    {
      "epoch": 43.4,
      "learning_rate": 5.661702127659575e-05,
      "loss": 0.0007,
      "step": 6120
    },
    {
      "epoch": 43.48,
      "learning_rate": 5.6546099290780144e-05,
      "loss": 0.0018,
      "step": 6130
    },
    {
      "epoch": 43.55,
      "learning_rate": 5.6475177304964544e-05,
      "loss": 0.0005,
      "step": 6140
    },
    {
      "epoch": 43.62,
      "learning_rate": 5.640425531914894e-05,
      "loss": 0.0405,
      "step": 6150
    },
    {
      "epoch": 43.69,
      "learning_rate": 5.633333333333334e-05,
      "loss": 0.0089,
      "step": 6160
    },
    {
      "epoch": 43.76,
      "learning_rate": 5.626241134751773e-05,
      "loss": 0.0127,
      "step": 6170
    },
    {
      "epoch": 43.83,
      "learning_rate": 5.619148936170213e-05,
      "loss": 0.0506,
      "step": 6180
    },
    {
      "epoch": 43.9,
      "learning_rate": 5.6120567375886526e-05,
      "loss": 0.0835,
      "step": 6190
    },
    {
      "epoch": 43.97,
      "learning_rate": 5.6049645390070926e-05,
      "loss": 0.0018,
      "step": 6200
    },
    {
      "epoch": 43.97,
      "eval_accuracy": 0.9805996472663139,
      "eval_loss": 0.09741748869419098,
      "eval_runtime": 4.787,
      "eval_samples_per_second": 118.445,
      "eval_steps_per_second": 14.832,
      "step": 6200
    },
    {
      "epoch": 44.04,
      "learning_rate": 5.597872340425532e-05,
      "loss": 0.031,
      "step": 6210
    },
    {
      "epoch": 44.11,
      "learning_rate": 5.590780141843972e-05,
      "loss": 0.0296,
      "step": 6220
    },
    {
      "epoch": 44.18,
      "learning_rate": 5.5836879432624114e-05,
      "loss": 0.0049,
      "step": 6230
    },
    {
      "epoch": 44.26,
      "learning_rate": 5.5765957446808514e-05,
      "loss": 0.0814,
      "step": 6240
    },
    {
      "epoch": 44.33,
      "learning_rate": 5.569503546099291e-05,
      "loss": 0.0165,
      "step": 6250
    },
    {
      "epoch": 44.4,
      "learning_rate": 5.562411347517731e-05,
      "loss": 0.0414,
      "step": 6260
    },
    {
      "epoch": 44.47,
      "learning_rate": 5.55531914893617e-05,
      "loss": 0.0137,
      "step": 6270
    },
    {
      "epoch": 44.54,
      "learning_rate": 5.54822695035461e-05,
      "loss": 0.0005,
      "step": 6280
    },
    {
      "epoch": 44.61,
      "learning_rate": 5.5411347517730496e-05,
      "loss": 0.011,
      "step": 6290
    },
    {
      "epoch": 44.68,
      "learning_rate": 5.5340425531914896e-05,
      "loss": 0.0285,
      "step": 6300
    },
    {
      "epoch": 44.68,
      "eval_accuracy": 0.9929453262786596,
      "eval_loss": 0.008973962627351284,
      "eval_runtime": 4.7045,
      "eval_samples_per_second": 120.522,
      "eval_steps_per_second": 15.092,
      "step": 6300
    },
    {
      "epoch": 44.75,
      "learning_rate": 5.526950354609929e-05,
      "loss": 0.0006,
      "step": 6310
    },
    {
      "epoch": 44.82,
      "learning_rate": 5.519858156028369e-05,
      "loss": 0.0011,
      "step": 6320
    },
    {
      "epoch": 44.89,
      "learning_rate": 5.5127659574468084e-05,
      "loss": 0.0005,
      "step": 6330
    },
    {
      "epoch": 44.96,
      "learning_rate": 5.5056737588652484e-05,
      "loss": 0.0027,
      "step": 6340
    },
    {
      "epoch": 45.04,
      "learning_rate": 5.4985815602836885e-05,
      "loss": 0.0155,
      "step": 6350
    },
    {
      "epoch": 45.11,
      "learning_rate": 5.491489361702128e-05,
      "loss": 0.0139,
      "step": 6360
    },
    {
      "epoch": 45.18,
      "learning_rate": 5.484397163120568e-05,
      "loss": 0.0019,
      "step": 6370
    },
    {
      "epoch": 45.25,
      "learning_rate": 5.477304964539007e-05,
      "loss": 0.084,
      "step": 6380
    },
    {
      "epoch": 45.32,
      "learning_rate": 5.470212765957447e-05,
      "loss": 0.0038,
      "step": 6390
    },
    {
      "epoch": 45.39,
      "learning_rate": 5.4631205673758866e-05,
      "loss": 0.0365,
      "step": 6400
    },
    {
      "epoch": 45.39,
      "eval_accuracy": 0.9876543209876543,
      "eval_loss": 0.040918827056884766,
      "eval_runtime": 4.769,
      "eval_samples_per_second": 118.892,
      "eval_steps_per_second": 14.888,
      "step": 6400
    },
    {
      "epoch": 45.46,
      "learning_rate": 5.456028368794327e-05,
      "loss": 0.0014,
      "step": 6410
    },
    {
      "epoch": 45.53,
      "learning_rate": 5.448936170212766e-05,
      "loss": 0.0566,
      "step": 6420
    },
    {
      "epoch": 45.6,
      "learning_rate": 5.441843971631206e-05,
      "loss": 0.0243,
      "step": 6430
    },
    {
      "epoch": 45.67,
      "learning_rate": 5.4347517730496454e-05,
      "loss": 0.0018,
      "step": 6440
    },
    {
      "epoch": 45.74,
      "learning_rate": 5.427659574468086e-05,
      "loss": 0.1125,
      "step": 6450
    },
    {
      "epoch": 45.82,
      "learning_rate": 5.420567375886525e-05,
      "loss": 0.0053,
      "step": 6460
    },
    {
      "epoch": 45.89,
      "learning_rate": 5.4134751773049656e-05,
      "loss": 0.0143,
      "step": 6470
    },
    {
      "epoch": 45.96,
      "learning_rate": 5.406382978723404e-05,
      "loss": 0.0023,
      "step": 6480
    },
    {
      "epoch": 46.03,
      "learning_rate": 5.399290780141845e-05,
      "loss": 0.0005,
      "step": 6490
    },
    {
      "epoch": 46.1,
      "learning_rate": 5.3921985815602836e-05,
      "loss": 0.0007,
      "step": 6500
    },
    {
      "epoch": 46.1,
      "eval_accuracy": 0.9929453262786596,
      "eval_loss": 0.04250045865774155,
      "eval_runtime": 4.579,
      "eval_samples_per_second": 123.825,
      "eval_steps_per_second": 15.505,
      "step": 6500
    },
    {
      "epoch": 46.17,
      "learning_rate": 5.3851063829787244e-05,
      "loss": 0.0431,
      "step": 6510
    },
    {
      "epoch": 46.24,
      "learning_rate": 5.378014184397163e-05,
      "loss": 0.0006,
      "step": 6520
    },
    {
      "epoch": 46.31,
      "learning_rate": 5.370921985815604e-05,
      "loss": 0.0006,
      "step": 6530
    },
    {
      "epoch": 46.38,
      "learning_rate": 5.3638297872340424e-05,
      "loss": 0.0007,
      "step": 6540
    },
    {
      "epoch": 46.45,
      "learning_rate": 5.356737588652483e-05,
      "loss": 0.041,
      "step": 6550
    },
    {
      "epoch": 46.52,
      "learning_rate": 5.349645390070922e-05,
      "loss": 0.0242,
      "step": 6560
    },
    {
      "epoch": 46.6,
      "learning_rate": 5.3425531914893626e-05,
      "loss": 0.041,
      "step": 6570
    },
    {
      "epoch": 46.67,
      "learning_rate": 5.335460992907801e-05,
      "loss": 0.0233,
      "step": 6580
    },
    {
      "epoch": 46.74,
      "learning_rate": 5.328368794326242e-05,
      "loss": 0.0155,
      "step": 6590
    },
    {
      "epoch": 46.81,
      "learning_rate": 5.3212765957446806e-05,
      "loss": 0.0094,
      "step": 6600
    },
    {
      "epoch": 46.81,
      "eval_accuracy": 0.9929453262786596,
      "eval_loss": 0.04256556183099747,
      "eval_runtime": 4.876,
      "eval_samples_per_second": 116.283,
      "eval_steps_per_second": 14.561,
      "step": 6600
    },
    {
      "epoch": 46.88,
      "learning_rate": 5.3141843971631214e-05,
      "loss": 0.0216,
      "step": 6610
    },
    {
      "epoch": 46.95,
      "learning_rate": 5.30709219858156e-05,
      "loss": 0.0003,
      "step": 6620
    },
    {
      "epoch": 47.02,
      "learning_rate": 5.300000000000001e-05,
      "loss": 0.0089,
      "step": 6630
    },
    {
      "epoch": 47.09,
      "learning_rate": 5.2929078014184394e-05,
      "loss": 0.0327,
      "step": 6640
    },
    {
      "epoch": 47.16,
      "learning_rate": 5.28581560283688e-05,
      "loss": 0.0008,
      "step": 6650
    },
    {
      "epoch": 47.23,
      "learning_rate": 5.278723404255319e-05,
      "loss": 0.1177,
      "step": 6660
    },
    {
      "epoch": 47.3,
      "learning_rate": 5.2716312056737596e-05,
      "loss": 0.0005,
      "step": 6670
    },
    {
      "epoch": 47.38,
      "learning_rate": 5.264539007092198e-05,
      "loss": 0.0582,
      "step": 6680
    },
    {
      "epoch": 47.45,
      "learning_rate": 5.257446808510639e-05,
      "loss": 0.0611,
      "step": 6690
    },
    {
      "epoch": 47.52,
      "learning_rate": 5.2503546099290776e-05,
      "loss": 0.0278,
      "step": 6700
    },
    {
      "epoch": 47.52,
      "eval_accuracy": 0.9929453262786596,
      "eval_loss": 0.02249726839363575,
      "eval_runtime": 4.833,
      "eval_samples_per_second": 117.317,
      "eval_steps_per_second": 14.691,
      "step": 6700
    },
    {
      "epoch": 47.59,
      "learning_rate": 5.2432624113475184e-05,
      "loss": 0.0208,
      "step": 6710
    },
    {
      "epoch": 47.66,
      "learning_rate": 5.236170212765957e-05,
      "loss": 0.0485,
      "step": 6720
    },
    {
      "epoch": 47.73,
      "learning_rate": 5.229078014184398e-05,
      "loss": 0.0007,
      "step": 6730
    },
    {
      "epoch": 47.8,
      "learning_rate": 5.2219858156028364e-05,
      "loss": 0.0082,
      "step": 6740
    },
    {
      "epoch": 47.87,
      "learning_rate": 5.214893617021277e-05,
      "loss": 0.0151,
      "step": 6750
    },
    {
      "epoch": 47.94,
      "learning_rate": 5.207801418439716e-05,
      "loss": 0.0006,
      "step": 6760
    },
    {
      "epoch": 48.01,
      "learning_rate": 5.2007092198581566e-05,
      "loss": 0.0007,
      "step": 6770
    },
    {
      "epoch": 48.09,
      "learning_rate": 5.193617021276595e-05,
      "loss": 0.0211,
      "step": 6780
    },
    {
      "epoch": 48.16,
      "learning_rate": 5.186524822695036e-05,
      "loss": 0.0202,
      "step": 6790
    },
    {
      "epoch": 48.23,
      "learning_rate": 5.1794326241134746e-05,
      "loss": 0.0062,
      "step": 6800
    },
    {
      "epoch": 48.23,
      "eval_accuracy": 0.9947089947089947,
      "eval_loss": 0.031184891238808632,
      "eval_runtime": 4.7515,
      "eval_samples_per_second": 119.33,
      "eval_steps_per_second": 14.943,
      "step": 6800
    },
    {
      "epoch": 48.3,
      "learning_rate": 5.1723404255319154e-05,
      "loss": 0.0005,
      "step": 6810
    },
    {
      "epoch": 48.37,
      "learning_rate": 5.165248226950354e-05,
      "loss": 0.0538,
      "step": 6820
    },
    {
      "epoch": 48.44,
      "learning_rate": 5.158156028368795e-05,
      "loss": 0.0006,
      "step": 6830
    },
    {
      "epoch": 48.51,
      "learning_rate": 5.151063829787234e-05,
      "loss": 0.023,
      "step": 6840
    },
    {
      "epoch": 48.58,
      "learning_rate": 5.143971631205674e-05,
      "loss": 0.0009,
      "step": 6850
    },
    {
      "epoch": 48.65,
      "learning_rate": 5.1368794326241135e-05,
      "loss": 0.0005,
      "step": 6860
    },
    {
      "epoch": 48.72,
      "learning_rate": 5.1297872340425536e-05,
      "loss": 0.0025,
      "step": 6870
    },
    {
      "epoch": 48.79,
      "learning_rate": 5.122695035460993e-05,
      "loss": 0.0353,
      "step": 6880
    },
    {
      "epoch": 48.87,
      "learning_rate": 5.115602836879433e-05,
      "loss": 0.0091,
      "step": 6890
    },
    {
      "epoch": 48.94,
      "learning_rate": 5.108510638297873e-05,
      "loss": 0.0128,
      "step": 6900
    },
    {
      "epoch": 48.94,
      "eval_accuracy": 0.9911816578483245,
      "eval_loss": 0.029220430180430412,
      "eval_runtime": 4.8855,
      "eval_samples_per_second": 116.057,
      "eval_steps_per_second": 14.533,
      "step": 6900
    },
    {
      "epoch": 49.01,
      "learning_rate": 5.1014184397163124e-05,
      "loss": 0.0011,
      "step": 6910
    },
    {
      "epoch": 49.08,
      "learning_rate": 5.0943262411347524e-05,
      "loss": 0.0494,
      "step": 6920
    },
    {
      "epoch": 49.15,
      "learning_rate": 5.087234042553192e-05,
      "loss": 0.0007,
      "step": 6930
    },
    {
      "epoch": 49.22,
      "learning_rate": 5.080141843971632e-05,
      "loss": 0.0325,
      "step": 6940
    },
    {
      "epoch": 49.29,
      "learning_rate": 5.073049645390071e-05,
      "loss": 0.0009,
      "step": 6950
    },
    {
      "epoch": 49.36,
      "learning_rate": 5.065957446808511e-05,
      "loss": 0.0564,
      "step": 6960
    },
    {
      "epoch": 49.43,
      "learning_rate": 5.0588652482269506e-05,
      "loss": 0.0004,
      "step": 6970
    },
    {
      "epoch": 49.5,
      "learning_rate": 5.0517730496453906e-05,
      "loss": 0.0344,
      "step": 6980
    },
    {
      "epoch": 49.57,
      "learning_rate": 5.04468085106383e-05,
      "loss": 0.0244,
      "step": 6990
    },
    {
      "epoch": 49.65,
      "learning_rate": 5.03758865248227e-05,
      "loss": 0.0559,
      "step": 7000
    },
    {
      "epoch": 49.65,
      "eval_accuracy": 0.9911816578483245,
      "eval_loss": 0.04575134441256523,
      "eval_runtime": 4.7935,
      "eval_samples_per_second": 118.284,
      "eval_steps_per_second": 14.812,
      "step": 7000
    },
    {
      "epoch": 49.72,
      "learning_rate": 5.0304964539007094e-05,
      "loss": 0.0502,
      "step": 7010
    },
    {
      "epoch": 49.79,
      "learning_rate": 5.0234042553191494e-05,
      "loss": 0.0074,
      "step": 7020
    },
    {
      "epoch": 49.86,
      "learning_rate": 5.016312056737589e-05,
      "loss": 0.0005,
      "step": 7030
    },
    {
      "epoch": 49.93,
      "learning_rate": 5.009219858156029e-05,
      "loss": 0.001,
      "step": 7040
    },
    {
      "epoch": 50.0,
      "learning_rate": 5.002127659574468e-05,
      "loss": 0.0151,
      "step": 7050
    },
    {
      "epoch": 50.07,
      "learning_rate": 4.995035460992908e-05,
      "loss": 0.0018,
      "step": 7060
    },
    {
      "epoch": 50.14,
      "learning_rate": 4.9879432624113476e-05,
      "loss": 0.0278,
      "step": 7070
    },
    {
      "epoch": 50.21,
      "learning_rate": 4.9808510638297876e-05,
      "loss": 0.0346,
      "step": 7080
    },
    {
      "epoch": 50.28,
      "learning_rate": 4.973758865248227e-05,
      "loss": 0.0004,
      "step": 7090
    },
    {
      "epoch": 50.35,
      "learning_rate": 4.966666666666667e-05,
      "loss": 0.0042,
      "step": 7100
    },
    {
      "epoch": 50.35,
      "eval_accuracy": 0.982363315696649,
      "eval_loss": 0.063356913626194,
      "eval_runtime": 4.855,
      "eval_samples_per_second": 116.786,
      "eval_steps_per_second": 14.624,
      "step": 7100
    },
    {
      "epoch": 50.43,
      "learning_rate": 4.9595744680851064e-05,
      "loss": 0.0006,
      "step": 7110
    },
    {
      "epoch": 50.5,
      "learning_rate": 4.9524822695035464e-05,
      "loss": 0.0005,
      "step": 7120
    },
    {
      "epoch": 50.57,
      "learning_rate": 4.945390070921986e-05,
      "loss": 0.0004,
      "step": 7130
    },
    {
      "epoch": 50.64,
      "learning_rate": 4.938297872340426e-05,
      "loss": 0.0555,
      "step": 7140
    },
    {
      "epoch": 50.71,
      "learning_rate": 4.931205673758865e-05,
      "loss": 0.0236,
      "step": 7150
    },
    {
      "epoch": 50.78,
      "learning_rate": 4.924113475177305e-05,
      "loss": 0.0131,
      "step": 7160
    },
    {
      "epoch": 50.85,
      "learning_rate": 4.917021276595745e-05,
      "loss": 0.0021,
      "step": 7170
    },
    {
      "epoch": 50.92,
      "learning_rate": 4.9099290780141846e-05,
      "loss": 0.0013,
      "step": 7180
    },
    {
      "epoch": 50.99,
      "learning_rate": 4.9028368794326246e-05,
      "loss": 0.0007,
      "step": 7190
    },
    {
      "epoch": 51.06,
      "learning_rate": 4.895744680851064e-05,
      "loss": 0.0005,
      "step": 7200
    },
    {
      "epoch": 51.06,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0010965272085741162,
      "eval_runtime": 4.691,
      "eval_samples_per_second": 120.869,
      "eval_steps_per_second": 15.135,
      "step": 7200
    },
    {
      "epoch": 51.13,
      "learning_rate": 4.888652482269504e-05,
      "loss": 0.0005,
      "step": 7210
    },
    {
      "epoch": 51.21,
      "learning_rate": 4.8815602836879434e-05,
      "loss": 0.0003,
      "step": 7220
    },
    {
      "epoch": 51.28,
      "learning_rate": 4.8744680851063834e-05,
      "loss": 0.0003,
      "step": 7230
    },
    {
      "epoch": 51.35,
      "learning_rate": 4.867375886524823e-05,
      "loss": 0.0004,
      "step": 7240
    },
    {
      "epoch": 51.42,
      "learning_rate": 4.860283687943263e-05,
      "loss": 0.0111,
      "step": 7250
    },
    {
      "epoch": 51.49,
      "learning_rate": 4.853191489361702e-05,
      "loss": 0.0003,
      "step": 7260
    },
    {
      "epoch": 51.56,
      "learning_rate": 4.846099290780142e-05,
      "loss": 0.0562,
      "step": 7270
    },
    {
      "epoch": 51.63,
      "learning_rate": 4.839007092198582e-05,
      "loss": 0.0004,
      "step": 7280
    },
    {
      "epoch": 51.7,
      "learning_rate": 4.8319148936170216e-05,
      "loss": 0.0003,
      "step": 7290
    },
    {
      "epoch": 51.77,
      "learning_rate": 4.824822695035462e-05,
      "loss": 0.0003,
      "step": 7300
    },
    {
      "epoch": 51.77,
      "eval_accuracy": 0.9947089947089947,
      "eval_loss": 0.030498141422867775,
      "eval_runtime": 4.635,
      "eval_samples_per_second": 122.329,
      "eval_steps_per_second": 15.318,
      "step": 7300
    },
    {
      "epoch": 51.84,
      "learning_rate": 4.817730496453901e-05,
      "loss": 0.0049,
      "step": 7310
    },
    {
      "epoch": 51.91,
      "learning_rate": 4.810638297872341e-05,
      "loss": 0.0071,
      "step": 7320
    },
    {
      "epoch": 51.99,
      "learning_rate": 4.8035460992907804e-05,
      "loss": 0.0004,
      "step": 7330
    },
    {
      "epoch": 52.06,
      "learning_rate": 4.7964539007092205e-05,
      "loss": 0.0004,
      "step": 7340
    },
    {
      "epoch": 52.13,
      "learning_rate": 4.78936170212766e-05,
      "loss": 0.0237,
      "step": 7350
    },
    {
      "epoch": 52.2,
      "learning_rate": 4.7822695035461e-05,
      "loss": 0.0005,
      "step": 7360
    },
    {
      "epoch": 52.27,
      "learning_rate": 4.775177304964539e-05,
      "loss": 0.0036,
      "step": 7370
    },
    {
      "epoch": 52.34,
      "learning_rate": 4.768085106382979e-05,
      "loss": 0.0502,
      "step": 7380
    },
    {
      "epoch": 52.41,
      "learning_rate": 4.7609929078014186e-05,
      "loss": 0.0133,
      "step": 7390
    },
    {
      "epoch": 52.48,
      "learning_rate": 4.753900709219859e-05,
      "loss": 0.0003,
      "step": 7400
    },
    {
      "epoch": 52.48,
      "eval_accuracy": 0.9929453262786596,
      "eval_loss": 0.044626634567976,
      "eval_runtime": 4.8305,
      "eval_samples_per_second": 117.378,
      "eval_steps_per_second": 14.698,
      "step": 7400
    },
    {
      "epoch": 52.55,
      "learning_rate": 4.746808510638298e-05,
      "loss": 0.0003,
      "step": 7410
    },
    {
      "epoch": 52.62,
      "learning_rate": 4.739716312056738e-05,
      "loss": 0.0175,
      "step": 7420
    },
    {
      "epoch": 52.7,
      "learning_rate": 4.7326241134751774e-05,
      "loss": 0.0013,
      "step": 7430
    },
    {
      "epoch": 52.77,
      "learning_rate": 4.7255319148936175e-05,
      "loss": 0.0581,
      "step": 7440
    },
    {
      "epoch": 52.84,
      "learning_rate": 4.718439716312057e-05,
      "loss": 0.0116,
      "step": 7450
    },
    {
      "epoch": 52.91,
      "learning_rate": 4.711347517730497e-05,
      "loss": 0.0003,
      "step": 7460
    },
    {
      "epoch": 52.98,
      "learning_rate": 4.704255319148936e-05,
      "loss": 0.0054,
      "step": 7470
    },
    {
      "epoch": 53.05,
      "learning_rate": 4.697163120567376e-05,
      "loss": 0.0429,
      "step": 7480
    },
    {
      "epoch": 53.12,
      "learning_rate": 4.6900709219858156e-05,
      "loss": 0.0294,
      "step": 7490
    },
    {
      "epoch": 53.19,
      "learning_rate": 4.682978723404256e-05,
      "loss": 0.0032,
      "step": 7500
    },
    {
      "epoch": 53.19,
      "eval_accuracy": 0.9929453262786596,
      "eval_loss": 0.046005308628082275,
      "eval_runtime": 4.868,
      "eval_samples_per_second": 116.474,
      "eval_steps_per_second": 14.585,
      "step": 7500
    },
    {
      "epoch": 53.26,
      "learning_rate": 4.675886524822695e-05,
      "loss": 0.0003,
      "step": 7510
    },
    {
      "epoch": 53.33,
      "learning_rate": 4.668794326241135e-05,
      "loss": 0.0505,
      "step": 7520
    },
    {
      "epoch": 53.4,
      "learning_rate": 4.6617021276595744e-05,
      "loss": 0.0034,
      "step": 7530
    },
    {
      "epoch": 53.48,
      "learning_rate": 4.6546099290780145e-05,
      "loss": 0.0102,
      "step": 7540
    },
    {
      "epoch": 53.55,
      "learning_rate": 4.647517730496454e-05,
      "loss": 0.0003,
      "step": 7550
    },
    {
      "epoch": 53.62,
      "learning_rate": 4.640425531914894e-05,
      "loss": 0.0101,
      "step": 7560
    },
    {
      "epoch": 53.69,
      "learning_rate": 4.633333333333333e-05,
      "loss": 0.0705,
      "step": 7570
    },
    {
      "epoch": 53.76,
      "learning_rate": 4.626241134751773e-05,
      "loss": 0.0007,
      "step": 7580
    },
    {
      "epoch": 53.83,
      "learning_rate": 4.6191489361702126e-05,
      "loss": 0.0875,
      "step": 7590
    },
    {
      "epoch": 53.9,
      "learning_rate": 4.612056737588653e-05,
      "loss": 0.0893,
      "step": 7600
    },
    {
      "epoch": 53.9,
      "eval_accuracy": 0.9929453262786596,
      "eval_loss": 0.03069092519581318,
      "eval_runtime": 4.635,
      "eval_samples_per_second": 122.329,
      "eval_steps_per_second": 15.318,
      "step": 7600
    },
    {
      "epoch": 53.97,
      "learning_rate": 4.604964539007092e-05,
      "loss": 0.0006,
      "step": 7610
    },
    {
      "epoch": 54.04,
      "learning_rate": 4.597872340425532e-05,
      "loss": 0.0604,
      "step": 7620
    },
    {
      "epoch": 54.11,
      "learning_rate": 4.5907801418439714e-05,
      "loss": 0.0007,
      "step": 7630
    },
    {
      "epoch": 54.18,
      "learning_rate": 4.5836879432624115e-05,
      "loss": 0.0064,
      "step": 7640
    },
    {
      "epoch": 54.26,
      "learning_rate": 4.576595744680851e-05,
      "loss": 0.0005,
      "step": 7650
    },
    {
      "epoch": 54.33,
      "learning_rate": 4.569503546099291e-05,
      "loss": 0.0195,
      "step": 7660
    },
    {
      "epoch": 54.4,
      "learning_rate": 4.56241134751773e-05,
      "loss": 0.0255,
      "step": 7670
    },
    {
      "epoch": 54.47,
      "learning_rate": 4.55531914893617e-05,
      "loss": 0.0018,
      "step": 7680
    },
    {
      "epoch": 54.54,
      "learning_rate": 4.5482269503546096e-05,
      "loss": 0.0747,
      "step": 7690
    },
    {
      "epoch": 54.61,
      "learning_rate": 4.54113475177305e-05,
      "loss": 0.0506,
      "step": 7700
    },
    {
      "epoch": 54.61,
      "eval_accuracy": 0.9858906525573192,
      "eval_loss": 0.08691811561584473,
      "eval_runtime": 4.666,
      "eval_samples_per_second": 121.516,
      "eval_steps_per_second": 15.216,
      "step": 7700
    },
    {
      "epoch": 54.68,
      "learning_rate": 4.53404255319149e-05,
      "loss": 0.0006,
      "step": 7710
    },
    {
      "epoch": 54.75,
      "learning_rate": 4.52695035460993e-05,
      "loss": 0.0301,
      "step": 7720
    },
    {
      "epoch": 54.82,
      "learning_rate": 4.519858156028369e-05,
      "loss": 0.0571,
      "step": 7730
    },
    {
      "epoch": 54.89,
      "learning_rate": 4.512765957446809e-05,
      "loss": 0.0067,
      "step": 7740
    },
    {
      "epoch": 54.96,
      "learning_rate": 4.5056737588652485e-05,
      "loss": 0.0014,
      "step": 7750
    },
    {
      "epoch": 55.04,
      "learning_rate": 4.4985815602836886e-05,
      "loss": 0.0408,
      "step": 7760
    },
    {
      "epoch": 55.11,
      "learning_rate": 4.491489361702128e-05,
      "loss": 0.0127,
      "step": 7770
    },
    {
      "epoch": 55.18,
      "learning_rate": 4.4851063829787235e-05,
      "loss": 0.0532,
      "step": 7780
    },
    {
      "epoch": 55.25,
      "learning_rate": 4.4780141843971635e-05,
      "loss": 0.0005,
      "step": 7790
    },
    {
      "epoch": 55.32,
      "learning_rate": 4.470921985815603e-05,
      "loss": 0.0321,
      "step": 7800
    },
    {
      "epoch": 55.32,
      "eval_accuracy": 0.9894179894179894,
      "eval_loss": 0.06382487714290619,
      "eval_runtime": 4.709,
      "eval_samples_per_second": 120.407,
      "eval_steps_per_second": 15.077,
      "step": 7800
    },
    {
      "epoch": 55.39,
      "learning_rate": 4.463829787234043e-05,
      "loss": 0.0004,
      "step": 7810
    },
    {
      "epoch": 55.46,
      "learning_rate": 4.456737588652482e-05,
      "loss": 0.0036,
      "step": 7820
    },
    {
      "epoch": 55.53,
      "learning_rate": 4.449645390070922e-05,
      "loss": 0.0005,
      "step": 7830
    },
    {
      "epoch": 55.6,
      "learning_rate": 4.442553191489362e-05,
      "loss": 0.0004,
      "step": 7840
    },
    {
      "epoch": 55.67,
      "learning_rate": 4.435460992907802e-05,
      "loss": 0.0549,
      "step": 7850
    },
    {
      "epoch": 55.74,
      "learning_rate": 4.428368794326241e-05,
      "loss": 0.0272,
      "step": 7860
    },
    {
      "epoch": 55.82,
      "learning_rate": 4.421276595744681e-05,
      "loss": 0.0036,
      "step": 7870
    },
    {
      "epoch": 55.89,
      "learning_rate": 4.4141843971631205e-05,
      "loss": 0.0113,
      "step": 7880
    },
    {
      "epoch": 55.96,
      "learning_rate": 4.4070921985815605e-05,
      "loss": 0.0232,
      "step": 7890
    },
    {
      "epoch": 56.03,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.0004,
      "step": 7900
    },
    {
      "epoch": 56.03,
      "eval_accuracy": 0.9894179894179894,
      "eval_loss": 0.06995675712823868,
      "eval_runtime": 4.5765,
      "eval_samples_per_second": 123.893,
      "eval_steps_per_second": 15.514,
      "step": 7900
    },
    {
      "epoch": 56.1,
      "learning_rate": 4.39290780141844e-05,
      "loss": 0.0058,
      "step": 7910
    },
    {
      "epoch": 56.17,
      "learning_rate": 4.38581560283688e-05,
      "loss": 0.0234,
      "step": 7920
    },
    {
      "epoch": 56.24,
      "learning_rate": 4.378723404255319e-05,
      "loss": 0.0524,
      "step": 7930
    },
    {
      "epoch": 56.31,
      "learning_rate": 4.3716312056737594e-05,
      "loss": 0.038,
      "step": 7940
    },
    {
      "epoch": 56.38,
      "learning_rate": 4.364539007092199e-05,
      "loss": 0.0047,
      "step": 7950
    },
    {
      "epoch": 56.45,
      "learning_rate": 4.357446808510639e-05,
      "loss": 0.0114,
      "step": 7960
    },
    {
      "epoch": 56.52,
      "learning_rate": 4.350354609929078e-05,
      "loss": 0.0129,
      "step": 7970
    },
    {
      "epoch": 56.6,
      "learning_rate": 4.343262411347518e-05,
      "loss": 0.0081,
      "step": 7980
    },
    {
      "epoch": 56.67,
      "learning_rate": 4.3361702127659575e-05,
      "loss": 0.0011,
      "step": 7990
    },
    {
      "epoch": 56.74,
      "learning_rate": 4.3290780141843976e-05,
      "loss": 0.0162,
      "step": 8000
    },
    {
      "epoch": 56.74,
      "eval_accuracy": 0.9858906525573192,
      "eval_loss": 0.05421828478574753,
      "eval_runtime": 4.7415,
      "eval_samples_per_second": 119.581,
      "eval_steps_per_second": 14.974,
      "step": 8000
    },
    {
      "epoch": 56.81,
      "learning_rate": 4.321985815602837e-05,
      "loss": 0.0244,
      "step": 8010
    },
    {
      "epoch": 56.88,
      "learning_rate": 4.314893617021277e-05,
      "loss": 0.007,
      "step": 8020
    },
    {
      "epoch": 56.95,
      "learning_rate": 4.307801418439716e-05,
      "loss": 0.0306,
      "step": 8030
    },
    {
      "epoch": 57.02,
      "learning_rate": 4.3007092198581564e-05,
      "loss": 0.0004,
      "step": 8040
    },
    {
      "epoch": 57.09,
      "learning_rate": 4.293617021276596e-05,
      "loss": 0.0215,
      "step": 8050
    },
    {
      "epoch": 57.16,
      "learning_rate": 4.286524822695036e-05,
      "loss": 0.0607,
      "step": 8060
    },
    {
      "epoch": 57.23,
      "learning_rate": 4.279432624113475e-05,
      "loss": 0.0018,
      "step": 8070
    },
    {
      "epoch": 57.3,
      "learning_rate": 4.272340425531915e-05,
      "loss": 0.0304,
      "step": 8080
    },
    {
      "epoch": 57.38,
      "learning_rate": 4.2652482269503545e-05,
      "loss": 0.0389,
      "step": 8090
    },
    {
      "epoch": 57.45,
      "learning_rate": 4.2581560283687946e-05,
      "loss": 0.0093,
      "step": 8100
    },
    {
      "epoch": 57.45,
      "eval_accuracy": 0.9929453262786596,
      "eval_loss": 0.04040713235735893,
      "eval_runtime": 4.616,
      "eval_samples_per_second": 122.833,
      "eval_steps_per_second": 15.381,
      "step": 8100
    },
    {
      "epoch": 57.52,
      "learning_rate": 4.251063829787234e-05,
      "loss": 0.0174,
      "step": 8110
    },
    {
      "epoch": 57.59,
      "learning_rate": 4.243971631205674e-05,
      "loss": 0.0023,
      "step": 8120
    },
    {
      "epoch": 57.66,
      "learning_rate": 4.236879432624113e-05,
      "loss": 0.0448,
      "step": 8130
    },
    {
      "epoch": 57.73,
      "learning_rate": 4.2297872340425534e-05,
      "loss": 0.0004,
      "step": 8140
    },
    {
      "epoch": 57.8,
      "learning_rate": 4.222695035460993e-05,
      "loss": 0.0004,
      "step": 8150
    },
    {
      "epoch": 57.87,
      "learning_rate": 4.215602836879433e-05,
      "loss": 0.0842,
      "step": 8160
    },
    {
      "epoch": 57.94,
      "learning_rate": 4.208510638297872e-05,
      "loss": 0.0196,
      "step": 8170
    },
    {
      "epoch": 58.01,
      "learning_rate": 4.201418439716312e-05,
      "loss": 0.0234,
      "step": 8180
    },
    {
      "epoch": 58.09,
      "learning_rate": 4.1943262411347515e-05,
      "loss": 0.0214,
      "step": 8190
    },
    {
      "epoch": 58.16,
      "learning_rate": 4.187234042553192e-05,
      "loss": 0.0004,
      "step": 8200
    },
    {
      "epoch": 58.16,
      "eval_accuracy": 0.9929453262786596,
      "eval_loss": 0.02249576337635517,
      "eval_runtime": 4.65,
      "eval_samples_per_second": 121.934,
      "eval_steps_per_second": 15.269,
      "step": 8200
    },
    {
      "epoch": 58.23,
      "learning_rate": 4.1801418439716316e-05,
      "loss": 0.0003,
      "step": 8210
    },
    {
      "epoch": 58.3,
      "learning_rate": 4.1730496453900716e-05,
      "loss": 0.0003,
      "step": 8220
    },
    {
      "epoch": 58.37,
      "learning_rate": 4.165957446808511e-05,
      "loss": 0.0006,
      "step": 8230
    },
    {
      "epoch": 58.44,
      "learning_rate": 4.158865248226951e-05,
      "loss": 0.0003,
      "step": 8240
    },
    {
      "epoch": 58.51,
      "learning_rate": 4.1517730496453904e-05,
      "loss": 0.0003,
      "step": 8250
    },
    {
      "epoch": 58.58,
      "learning_rate": 4.1446808510638304e-05,
      "loss": 0.0003,
      "step": 8260
    },
    {
      "epoch": 58.65,
      "learning_rate": 4.13758865248227e-05,
      "loss": 0.0003,
      "step": 8270
    },
    {
      "epoch": 58.72,
      "learning_rate": 4.13049645390071e-05,
      "loss": 0.0003,
      "step": 8280
    },
    {
      "epoch": 58.79,
      "learning_rate": 4.123404255319149e-05,
      "loss": 0.0004,
      "step": 8290
    },
    {
      "epoch": 58.87,
      "learning_rate": 4.116312056737589e-05,
      "loss": 0.0023,
      "step": 8300
    },
    {
      "epoch": 58.87,
      "eval_accuracy": 0.9947089947089947,
      "eval_loss": 0.023073412477970123,
      "eval_runtime": 4.6315,
      "eval_samples_per_second": 122.422,
      "eval_steps_per_second": 15.33,
      "step": 8300
    },
    {
      "epoch": 58.94,
      "learning_rate": 4.1092198581560286e-05,
      "loss": 0.0004,
      "step": 8310
    },
    {
      "epoch": 59.01,
      "learning_rate": 4.1021276595744686e-05,
      "loss": 0.0023,
      "step": 8320
    },
    {
      "epoch": 59.08,
      "learning_rate": 4.095035460992908e-05,
      "loss": 0.0002,
      "step": 8330
    },
    {
      "epoch": 59.15,
      "learning_rate": 4.087943262411348e-05,
      "loss": 0.0004,
      "step": 8340
    },
    {
      "epoch": 59.22,
      "learning_rate": 4.0808510638297874e-05,
      "loss": 0.0424,
      "step": 8350
    },
    {
      "epoch": 59.29,
      "learning_rate": 4.0737588652482274e-05,
      "loss": 0.0003,
      "step": 8360
    },
    {
      "epoch": 59.36,
      "learning_rate": 4.066666666666667e-05,
      "loss": 0.0008,
      "step": 8370
    },
    {
      "epoch": 59.43,
      "learning_rate": 4.059574468085107e-05,
      "loss": 0.0003,
      "step": 8380
    },
    {
      "epoch": 59.5,
      "learning_rate": 4.052482269503546e-05,
      "loss": 0.0029,
      "step": 8390
    },
    {
      "epoch": 59.57,
      "learning_rate": 4.045390070921986e-05,
      "loss": 0.0017,
      "step": 8400
    },
    {
      "epoch": 59.57,
      "eval_accuracy": 0.9929453262786596,
      "eval_loss": 0.03826703876256943,
      "eval_runtime": 4.6215,
      "eval_samples_per_second": 122.686,
      "eval_steps_per_second": 15.363,
      "step": 8400
    },
    {
      "epoch": 59.65,
      "learning_rate": 4.0382978723404256e-05,
      "loss": 0.0002,
      "step": 8410
    },
    {
      "epoch": 59.72,
      "learning_rate": 4.0312056737588656e-05,
      "loss": 0.0007,
      "step": 8420
    },
    {
      "epoch": 59.79,
      "learning_rate": 4.024113475177305e-05,
      "loss": 0.0187,
      "step": 8430
    },
    {
      "epoch": 59.86,
      "learning_rate": 4.017021276595745e-05,
      "loss": 0.0002,
      "step": 8440
    },
    {
      "epoch": 59.93,
      "learning_rate": 4.0099290780141844e-05,
      "loss": 0.0003,
      "step": 8450
    },
    {
      "epoch": 60.0,
      "learning_rate": 4.0028368794326244e-05,
      "loss": 0.0003,
      "step": 8460
    },
    {
      "epoch": 60.07,
      "learning_rate": 3.995744680851064e-05,
      "loss": 0.0023,
      "step": 8470
    },
    {
      "epoch": 60.14,
      "learning_rate": 3.988652482269504e-05,
      "loss": 0.0003,
      "step": 8480
    },
    {
      "epoch": 60.21,
      "learning_rate": 3.981560283687943e-05,
      "loss": 0.0003,
      "step": 8490
    },
    {
      "epoch": 60.28,
      "learning_rate": 3.974468085106383e-05,
      "loss": 0.0002,
      "step": 8500
    },
    {
      "epoch": 60.28,
      "eval_accuracy": 0.9894179894179894,
      "eval_loss": 0.0411054752767086,
      "eval_runtime": 4.6025,
      "eval_samples_per_second": 123.193,
      "eval_steps_per_second": 15.426,
      "step": 8500
    },
    {
      "epoch": 60.35,
      "learning_rate": 3.9673758865248226e-05,
      "loss": 0.0007,
      "step": 8510
    },
    {
      "epoch": 60.43,
      "learning_rate": 3.9602836879432626e-05,
      "loss": 0.0002,
      "step": 8520
    },
    {
      "epoch": 60.5,
      "learning_rate": 3.953191489361702e-05,
      "loss": 0.0002,
      "step": 8530
    },
    {
      "epoch": 60.57,
      "learning_rate": 3.946099290780142e-05,
      "loss": 0.0002,
      "step": 8540
    },
    {
      "epoch": 60.64,
      "learning_rate": 3.9390070921985814e-05,
      "loss": 0.0008,
      "step": 8550
    },
    {
      "epoch": 60.71,
      "learning_rate": 3.9319148936170214e-05,
      "loss": 0.0092,
      "step": 8560
    },
    {
      "epoch": 60.78,
      "learning_rate": 3.924822695035461e-05,
      "loss": 0.0023,
      "step": 8570
    },
    {
      "epoch": 60.85,
      "learning_rate": 3.917730496453901e-05,
      "loss": 0.0106,
      "step": 8580
    },
    {
      "epoch": 60.92,
      "learning_rate": 3.91063829787234e-05,
      "loss": 0.0004,
      "step": 8590
    },
    {
      "epoch": 60.99,
      "learning_rate": 3.90354609929078e-05,
      "loss": 0.0002,
      "step": 8600
    },
    {
      "epoch": 60.99,
      "eval_accuracy": 0.9929453262786596,
      "eval_loss": 0.0303778238594532,
      "eval_runtime": 4.6325,
      "eval_samples_per_second": 122.395,
      "eval_steps_per_second": 15.326,
      "step": 8600
    },
    {
      "epoch": 61.06,
      "learning_rate": 3.8964539007092196e-05,
      "loss": 0.0362,
      "step": 8610
    },
    {
      "epoch": 61.13,
      "learning_rate": 3.8893617021276596e-05,
      "loss": 0.0026,
      "step": 8620
    },
    {
      "epoch": 61.21,
      "learning_rate": 3.882269503546099e-05,
      "loss": 0.0186,
      "step": 8630
    },
    {
      "epoch": 61.28,
      "learning_rate": 3.875177304964539e-05,
      "loss": 0.0002,
      "step": 8640
    },
    {
      "epoch": 61.35,
      "learning_rate": 3.8680851063829784e-05,
      "loss": 0.0002,
      "step": 8650
    },
    {
      "epoch": 61.42,
      "learning_rate": 3.8609929078014184e-05,
      "loss": 0.0002,
      "step": 8660
    },
    {
      "epoch": 61.49,
      "learning_rate": 3.853900709219858e-05,
      "loss": 0.0003,
      "step": 8670
    },
    {
      "epoch": 61.56,
      "learning_rate": 3.846808510638298e-05,
      "loss": 0.0002,
      "step": 8680
    },
    {
      "epoch": 61.63,
      "learning_rate": 3.839716312056737e-05,
      "loss": 0.0002,
      "step": 8690
    },
    {
      "epoch": 61.7,
      "learning_rate": 3.832624113475177e-05,
      "loss": 0.0002,
      "step": 8700
    },
    {
      "epoch": 61.7,
      "eval_accuracy": 0.9964726631393298,
      "eval_loss": 0.01797201670706272,
      "eval_runtime": 4.6115,
      "eval_samples_per_second": 122.953,
      "eval_steps_per_second": 15.396,
      "step": 8700
    },
    {
      "epoch": 61.77,
      "learning_rate": 3.825531914893617e-05,
      "loss": 0.0507,
      "step": 8710
    },
    {
      "epoch": 61.84,
      "learning_rate": 3.8184397163120567e-05,
      "loss": 0.0002,
      "step": 8720
    },
    {
      "epoch": 61.91,
      "learning_rate": 3.811347517730497e-05,
      "loss": 0.0002,
      "step": 8730
    },
    {
      "epoch": 61.99,
      "learning_rate": 3.804255319148937e-05,
      "loss": 0.0004,
      "step": 8740
    },
    {
      "epoch": 62.06,
      "learning_rate": 3.797163120567376e-05,
      "loss": 0.0002,
      "step": 8750
    },
    {
      "epoch": 62.13,
      "learning_rate": 3.790070921985816e-05,
      "loss": 0.0002,
      "step": 8760
    },
    {
      "epoch": 62.2,
      "learning_rate": 3.7829787234042555e-05,
      "loss": 0.0184,
      "step": 8770
    },
    {
      "epoch": 62.27,
      "learning_rate": 3.7758865248226955e-05,
      "loss": 0.0002,
      "step": 8780
    },
    {
      "epoch": 62.34,
      "learning_rate": 3.768794326241135e-05,
      "loss": 0.0003,
      "step": 8790
    },
    {
      "epoch": 62.41,
      "learning_rate": 3.761702127659575e-05,
      "loss": 0.0002,
      "step": 8800
    },
    {
      "epoch": 62.41,
      "eval_accuracy": 0.9911816578483245,
      "eval_loss": 0.06262991577386856,
      "eval_runtime": 4.583,
      "eval_samples_per_second": 123.717,
      "eval_steps_per_second": 15.492,
      "step": 8800
    },
    {
      "epoch": 62.48,
      "learning_rate": 3.754609929078014e-05,
      "loss": 0.0002,
      "step": 8810
    },
    {
      "epoch": 62.55,
      "learning_rate": 3.747517730496454e-05,
      "loss": 0.0002,
      "step": 8820
    },
    {
      "epoch": 62.62,
      "learning_rate": 3.740425531914894e-05,
      "loss": 0.0002,
      "step": 8830
    },
    {
      "epoch": 62.7,
      "learning_rate": 3.733333333333334e-05,
      "loss": 0.0002,
      "step": 8840
    },
    {
      "epoch": 62.77,
      "learning_rate": 3.726241134751773e-05,
      "loss": 0.0002,
      "step": 8850
    },
    {
      "epoch": 62.84,
      "learning_rate": 3.719148936170213e-05,
      "loss": 0.0547,
      "step": 8860
    },
    {
      "epoch": 62.91,
      "learning_rate": 3.7120567375886525e-05,
      "loss": 0.0002,
      "step": 8870
    },
    {
      "epoch": 62.98,
      "learning_rate": 3.7049645390070925e-05,
      "loss": 0.0058,
      "step": 8880
    },
    {
      "epoch": 63.05,
      "learning_rate": 3.697872340425532e-05,
      "loss": 0.0002,
      "step": 8890
    },
    {
      "epoch": 63.12,
      "learning_rate": 3.690780141843972e-05,
      "loss": 0.0002,
      "step": 8900
    },
    {
      "epoch": 63.12,
      "eval_accuracy": 0.9964726631393298,
      "eval_loss": 0.034689877182245255,
      "eval_runtime": 4.55,
      "eval_samples_per_second": 124.614,
      "eval_steps_per_second": 15.604,
      "step": 8900
    },
    {
      "epoch": 63.19,
      "learning_rate": 3.683687943262411e-05,
      "loss": 0.0002,
      "step": 8910
    },
    {
      "epoch": 63.26,
      "learning_rate": 3.676595744680851e-05,
      "loss": 0.0003,
      "step": 8920
    },
    {
      "epoch": 63.33,
      "learning_rate": 3.6695035460992914e-05,
      "loss": 0.0002,
      "step": 8930
    },
    {
      "epoch": 63.4,
      "learning_rate": 3.662411347517731e-05,
      "loss": 0.0002,
      "step": 8940
    },
    {
      "epoch": 63.48,
      "learning_rate": 3.655319148936171e-05,
      "loss": 0.0002,
      "step": 8950
    },
    {
      "epoch": 63.55,
      "learning_rate": 3.64822695035461e-05,
      "loss": 0.0002,
      "step": 8960
    },
    {
      "epoch": 63.62,
      "learning_rate": 3.64113475177305e-05,
      "loss": 0.0002,
      "step": 8970
    },
    {
      "epoch": 63.69,
      "learning_rate": 3.6340425531914895e-05,
      "loss": 0.0005,
      "step": 8980
    },
    {
      "epoch": 63.76,
      "learning_rate": 3.6269503546099296e-05,
      "loss": 0.0002,
      "step": 8990
    },
    {
      "epoch": 63.83,
      "learning_rate": 3.619858156028369e-05,
      "loss": 0.0021,
      "step": 9000
    },
    {
      "epoch": 63.83,
      "eval_accuracy": 0.9947089947089947,
      "eval_loss": 0.040755413472652435,
      "eval_runtime": 4.651,
      "eval_samples_per_second": 121.908,
      "eval_steps_per_second": 15.265,
      "step": 9000
    },
    {
      "epoch": 63.9,
      "learning_rate": 3.612765957446809e-05,
      "loss": 0.0002,
      "step": 9010
    },
    {
      "epoch": 63.97,
      "learning_rate": 3.605673758865248e-05,
      "loss": 0.0002,
      "step": 9020
    },
    {
      "epoch": 64.04,
      "learning_rate": 3.5985815602836884e-05,
      "loss": 0.0679,
      "step": 9030
    },
    {
      "epoch": 64.11,
      "learning_rate": 3.591489361702128e-05,
      "loss": 0.0064,
      "step": 9040
    },
    {
      "epoch": 64.18,
      "learning_rate": 3.584397163120568e-05,
      "loss": 0.0003,
      "step": 9050
    },
    {
      "epoch": 64.26,
      "learning_rate": 3.577304964539007e-05,
      "loss": 0.0044,
      "step": 9060
    },
    {
      "epoch": 64.33,
      "learning_rate": 3.570212765957447e-05,
      "loss": 0.0296,
      "step": 9070
    },
    {
      "epoch": 64.4,
      "learning_rate": 3.5631205673758865e-05,
      "loss": 0.0498,
      "step": 9080
    },
    {
      "epoch": 64.47,
      "learning_rate": 3.5560283687943266e-05,
      "loss": 0.0003,
      "step": 9090
    },
    {
      "epoch": 64.54,
      "learning_rate": 3.548936170212766e-05,
      "loss": 0.0002,
      "step": 9100
    },
    {
      "epoch": 64.54,
      "eval_accuracy": 0.9947089947089947,
      "eval_loss": 0.022945033386349678,
      "eval_runtime": 4.681,
      "eval_samples_per_second": 121.127,
      "eval_steps_per_second": 15.168,
      "step": 9100
    },
    {
      "epoch": 64.61,
      "learning_rate": 3.541843971631206e-05,
      "loss": 0.0002,
      "step": 9110
    },
    {
      "epoch": 64.68,
      "learning_rate": 3.534751773049645e-05,
      "loss": 0.0234,
      "step": 9120
    },
    {
      "epoch": 64.75,
      "learning_rate": 3.5276595744680854e-05,
      "loss": 0.0002,
      "step": 9130
    },
    {
      "epoch": 64.82,
      "learning_rate": 3.520567375886525e-05,
      "loss": 0.0122,
      "step": 9140
    },
    {
      "epoch": 64.89,
      "learning_rate": 3.513475177304965e-05,
      "loss": 0.0002,
      "step": 9150
    },
    {
      "epoch": 64.96,
      "learning_rate": 3.506382978723404e-05,
      "loss": 0.0431,
      "step": 9160
    },
    {
      "epoch": 65.04,
      "learning_rate": 3.499290780141844e-05,
      "loss": 0.0003,
      "step": 9170
    },
    {
      "epoch": 65.11,
      "learning_rate": 3.4921985815602835e-05,
      "loss": 0.0629,
      "step": 9180
    },
    {
      "epoch": 65.18,
      "learning_rate": 3.4851063829787236e-05,
      "loss": 0.0004,
      "step": 9190
    },
    {
      "epoch": 65.25,
      "learning_rate": 3.478014184397163e-05,
      "loss": 0.0003,
      "step": 9200
    },
    {
      "epoch": 65.25,
      "eval_accuracy": 0.9911816578483245,
      "eval_loss": 0.02761363796889782,
      "eval_runtime": 4.5695,
      "eval_samples_per_second": 124.083,
      "eval_steps_per_second": 15.538,
      "step": 9200
    },
    {
      "epoch": 65.32,
      "learning_rate": 3.470921985815603e-05,
      "loss": 0.0011,
      "step": 9210
    },
    {
      "epoch": 65.39,
      "learning_rate": 3.463829787234042e-05,
      "loss": 0.0002,
      "step": 9220
    },
    {
      "epoch": 65.46,
      "learning_rate": 3.4567375886524824e-05,
      "loss": 0.0011,
      "step": 9230
    },
    {
      "epoch": 65.53,
      "learning_rate": 3.449645390070922e-05,
      "loss": 0.0005,
      "step": 9240
    },
    {
      "epoch": 65.6,
      "learning_rate": 3.442553191489362e-05,
      "loss": 0.0745,
      "step": 9250
    },
    {
      "epoch": 65.67,
      "learning_rate": 3.435460992907801e-05,
      "loss": 0.0002,
      "step": 9260
    },
    {
      "epoch": 65.74,
      "learning_rate": 3.428368794326241e-05,
      "loss": 0.0002,
      "step": 9270
    },
    {
      "epoch": 65.82,
      "learning_rate": 3.421276595744681e-05,
      "loss": 0.0002,
      "step": 9280
    },
    {
      "epoch": 65.89,
      "learning_rate": 3.414184397163121e-05,
      "loss": 0.0431,
      "step": 9290
    },
    {
      "epoch": 65.96,
      "learning_rate": 3.4070921985815606e-05,
      "loss": 0.0004,
      "step": 9300
    },
    {
      "epoch": 65.96,
      "eval_accuracy": 0.9947089947089947,
      "eval_loss": 0.03187382593750954,
      "eval_runtime": 4.623,
      "eval_samples_per_second": 122.647,
      "eval_steps_per_second": 15.358,
      "step": 9300
    },
    {
      "epoch": 66.03,
      "learning_rate": 3.4000000000000007e-05,
      "loss": 0.0336,
      "step": 9310
    },
    {
      "epoch": 66.1,
      "learning_rate": 3.39290780141844e-05,
      "loss": 0.0002,
      "step": 9320
    },
    {
      "epoch": 66.17,
      "learning_rate": 3.38581560283688e-05,
      "loss": 0.0259,
      "step": 9330
    },
    {
      "epoch": 66.24,
      "learning_rate": 3.3787234042553194e-05,
      "loss": 0.0002,
      "step": 9340
    },
    {
      "epoch": 66.31,
      "learning_rate": 3.372340425531915e-05,
      "loss": 0.0056,
      "step": 9350
    },
    {
      "epoch": 66.38,
      "learning_rate": 3.365248226950355e-05,
      "loss": 0.0313,
      "step": 9360
    },
    {
      "epoch": 66.45,
      "learning_rate": 3.3581560283687944e-05,
      "loss": 0.0003,
      "step": 9370
    },
    {
      "epoch": 66.52,
      "learning_rate": 3.3510638297872344e-05,
      "loss": 0.0002,
      "step": 9380
    },
    {
      "epoch": 66.6,
      "learning_rate": 3.343971631205674e-05,
      "loss": 0.0004,
      "step": 9390
    },
    {
      "epoch": 66.67,
      "learning_rate": 3.336879432624114e-05,
      "loss": 0.0002,
      "step": 9400
    },
    {
      "epoch": 66.67,
      "eval_accuracy": 0.9964726631393298,
      "eval_loss": 0.016740096732974052,
      "eval_runtime": 4.8155,
      "eval_samples_per_second": 117.744,
      "eval_steps_per_second": 14.744,
      "step": 9400
    },
    {
      "epoch": 66.74,
      "learning_rate": 3.329787234042553e-05,
      "loss": 0.0015,
      "step": 9410
    },
    {
      "epoch": 66.81,
      "learning_rate": 3.322695035460993e-05,
      "loss": 0.0002,
      "step": 9420
    },
    {
      "epoch": 66.88,
      "learning_rate": 3.3156028368794326e-05,
      "loss": 0.0002,
      "step": 9430
    },
    {
      "epoch": 66.95,
      "learning_rate": 3.3085106382978726e-05,
      "loss": 0.0002,
      "step": 9440
    },
    {
      "epoch": 67.02,
      "learning_rate": 3.301418439716312e-05,
      "loss": 0.0002,
      "step": 9450
    },
    {
      "epoch": 67.09,
      "learning_rate": 3.294326241134752e-05,
      "loss": 0.0002,
      "step": 9460
    },
    {
      "epoch": 67.16,
      "learning_rate": 3.2872340425531914e-05,
      "loss": 0.0002,
      "step": 9470
    },
    {
      "epoch": 67.23,
      "learning_rate": 3.2801418439716314e-05,
      "loss": 0.0003,
      "step": 9480
    },
    {
      "epoch": 67.3,
      "learning_rate": 3.273049645390071e-05,
      "loss": 0.0002,
      "step": 9490
    },
    {
      "epoch": 67.38,
      "learning_rate": 3.265957446808511e-05,
      "loss": 0.0036,
      "step": 9500
    },
    {
      "epoch": 67.38,
      "eval_accuracy": 0.9964726631393298,
      "eval_loss": 0.01307621132582426,
      "eval_runtime": 4.6715,
      "eval_samples_per_second": 121.373,
      "eval_steps_per_second": 15.198,
      "step": 9500
    },
    {
      "epoch": 67.45,
      "learning_rate": 3.25886524822695e-05,
      "loss": 0.0002,
      "step": 9510
    },
    {
      "epoch": 67.52,
      "learning_rate": 3.25177304964539e-05,
      "loss": 0.0002,
      "step": 9520
    },
    {
      "epoch": 67.59,
      "learning_rate": 3.2446808510638296e-05,
      "loss": 0.0002,
      "step": 9530
    },
    {
      "epoch": 67.66,
      "learning_rate": 3.2375886524822696e-05,
      "loss": 0.0004,
      "step": 9540
    },
    {
      "epoch": 67.73,
      "learning_rate": 3.2304964539007097e-05,
      "loss": 0.0002,
      "step": 9550
    },
    {
      "epoch": 67.8,
      "learning_rate": 3.223404255319149e-05,
      "loss": 0.0002,
      "step": 9560
    },
    {
      "epoch": 67.87,
      "learning_rate": 3.216312056737589e-05,
      "loss": 0.0002,
      "step": 9570
    },
    {
      "epoch": 67.94,
      "learning_rate": 3.2092198581560284e-05,
      "loss": 0.0002,
      "step": 9580
    },
    {
      "epoch": 68.01,
      "learning_rate": 3.2021276595744685e-05,
      "loss": 0.0004,
      "step": 9590
    },
    {
      "epoch": 68.09,
      "learning_rate": 3.195035460992908e-05,
      "loss": 0.0007,
      "step": 9600
    },
    {
      "epoch": 68.09,
      "eval_accuracy": 0.9929453262786596,
      "eval_loss": 0.0501413494348526,
      "eval_runtime": 4.575,
      "eval_samples_per_second": 123.933,
      "eval_steps_per_second": 15.519,
      "step": 9600
    },
    {
      "epoch": 68.16,
      "learning_rate": 3.187943262411348e-05,
      "loss": 0.0158,
      "step": 9610
    },
    {
      "epoch": 68.23,
      "learning_rate": 3.180851063829787e-05,
      "loss": 0.0002,
      "step": 9620
    },
    {
      "epoch": 68.3,
      "learning_rate": 3.173758865248227e-05,
      "loss": 0.0007,
      "step": 9630
    },
    {
      "epoch": 68.37,
      "learning_rate": 3.1666666666666666e-05,
      "loss": 0.0002,
      "step": 9640
    },
    {
      "epoch": 68.44,
      "learning_rate": 3.1595744680851067e-05,
      "loss": 0.0002,
      "step": 9650
    },
    {
      "epoch": 68.51,
      "learning_rate": 3.152482269503546e-05,
      "loss": 0.0003,
      "step": 9660
    },
    {
      "epoch": 68.58,
      "learning_rate": 3.145390070921986e-05,
      "loss": 0.1066,
      "step": 9670
    },
    {
      "epoch": 68.65,
      "learning_rate": 3.1382978723404254e-05,
      "loss": 0.0002,
      "step": 9680
    },
    {
      "epoch": 68.72,
      "learning_rate": 3.1312056737588655e-05,
      "loss": 0.0003,
      "step": 9690
    },
    {
      "epoch": 68.79,
      "learning_rate": 3.124113475177305e-05,
      "loss": 0.0002,
      "step": 9700
    },
    {
      "epoch": 68.79,
      "eval_accuracy": 0.9982363315696648,
      "eval_loss": 0.003756000893190503,
      "eval_runtime": 4.627,
      "eval_samples_per_second": 122.541,
      "eval_steps_per_second": 15.345,
      "step": 9700
    },
    {
      "epoch": 68.87,
      "learning_rate": 3.117021276595745e-05,
      "loss": 0.002,
      "step": 9710
    },
    {
      "epoch": 68.94,
      "learning_rate": 3.109929078014184e-05,
      "loss": 0.0002,
      "step": 9720
    },
    {
      "epoch": 69.01,
      "learning_rate": 3.102836879432624e-05,
      "loss": 0.0181,
      "step": 9730
    },
    {
      "epoch": 69.08,
      "learning_rate": 3.0957446808510636e-05,
      "loss": 0.0084,
      "step": 9740
    },
    {
      "epoch": 69.15,
      "learning_rate": 3.0886524822695037e-05,
      "loss": 0.0223,
      "step": 9750
    },
    {
      "epoch": 69.22,
      "learning_rate": 3.081560283687943e-05,
      "loss": 0.0009,
      "step": 9760
    },
    {
      "epoch": 69.29,
      "learning_rate": 3.074468085106384e-05,
      "loss": 0.0573,
      "step": 9770
    },
    {
      "epoch": 69.36,
      "learning_rate": 3.067375886524823e-05,
      "loss": 0.0084,
      "step": 9780
    },
    {
      "epoch": 69.43,
      "learning_rate": 3.060283687943263e-05,
      "loss": 0.0039,
      "step": 9790
    },
    {
      "epoch": 69.5,
      "learning_rate": 3.0531914893617025e-05,
      "loss": 0.0009,
      "step": 9800
    },
    {
      "epoch": 69.5,
      "eval_accuracy": 0.9982363315696648,
      "eval_loss": 0.012646459974348545,
      "eval_runtime": 4.6265,
      "eval_samples_per_second": 122.554,
      "eval_steps_per_second": 15.346,
      "step": 9800
    },
    {
      "epoch": 69.57,
      "learning_rate": 3.0460992907801422e-05,
      "loss": 0.0002,
      "step": 9810
    },
    {
      "epoch": 69.65,
      "learning_rate": 3.039007092198582e-05,
      "loss": 0.0002,
      "step": 9820
    },
    {
      "epoch": 69.72,
      "learning_rate": 3.0319148936170216e-05,
      "loss": 0.047,
      "step": 9830
    },
    {
      "epoch": 69.79,
      "learning_rate": 3.0248226950354613e-05,
      "loss": 0.0007,
      "step": 9840
    },
    {
      "epoch": 69.86,
      "learning_rate": 3.017730496453901e-05,
      "loss": 0.0007,
      "step": 9850
    },
    {
      "epoch": 69.93,
      "learning_rate": 3.0106382978723407e-05,
      "loss": 0.0038,
      "step": 9860
    },
    {
      "epoch": 70.0,
      "learning_rate": 3.0035460992907804e-05,
      "loss": 0.0004,
      "step": 9870
    },
    {
      "epoch": 70.07,
      "learning_rate": 2.99645390070922e-05,
      "loss": 0.0008,
      "step": 9880
    },
    {
      "epoch": 70.14,
      "learning_rate": 2.9893617021276598e-05,
      "loss": 0.0003,
      "step": 9890
    },
    {
      "epoch": 70.21,
      "learning_rate": 2.9822695035460995e-05,
      "loss": 0.0002,
      "step": 9900
    },
    {
      "epoch": 70.21,
      "eval_accuracy": 0.9964726631393298,
      "eval_loss": 0.016881294548511505,
      "eval_runtime": 4.685,
      "eval_samples_per_second": 121.024,
      "eval_steps_per_second": 15.155,
      "step": 9900
    },
    {
      "epoch": 70.28,
      "learning_rate": 2.9751773049645392e-05,
      "loss": 0.0001,
      "step": 9910
    },
    {
      "epoch": 70.35,
      "learning_rate": 2.968085106382979e-05,
      "loss": 0.0002,
      "step": 9920
    },
    {
      "epoch": 70.43,
      "learning_rate": 2.9609929078014186e-05,
      "loss": 0.0002,
      "step": 9930
    },
    {
      "epoch": 70.5,
      "learning_rate": 2.9539007092198583e-05,
      "loss": 0.0005,
      "step": 9940
    },
    {
      "epoch": 70.57,
      "learning_rate": 2.946808510638298e-05,
      "loss": 0.0003,
      "step": 9950
    },
    {
      "epoch": 70.64,
      "learning_rate": 2.9397163120567377e-05,
      "loss": 0.0002,
      "step": 9960
    },
    {
      "epoch": 70.71,
      "learning_rate": 2.9326241134751774e-05,
      "loss": 0.0001,
      "step": 9970
    },
    {
      "epoch": 70.78,
      "learning_rate": 2.925531914893617e-05,
      "loss": 0.0003,
      "step": 9980
    },
    {
      "epoch": 70.85,
      "learning_rate": 2.9184397163120568e-05,
      "loss": 0.0002,
      "step": 9990
    },
    {
      "epoch": 70.92,
      "learning_rate": 2.9113475177304965e-05,
      "loss": 0.0002,
      "step": 10000
    },
    {
      "epoch": 70.92,
      "eval_accuracy": 1.0,
      "eval_loss": 0.0004703931335825473,
      "eval_runtime": 4.594,
      "eval_samples_per_second": 123.421,
      "eval_steps_per_second": 15.455,
      "step": 10000
    }
  ],
  "max_steps": 14100,
  "num_train_epochs": 100,
  "total_flos": 1.2366947541295227e+19,
  "trial_name": null,
  "trial_params": null
}