{
  "best_metric": 0.9238578680203046,
  "best_model_checkpoint": "convnextv2-large-22k-384-skin7/checkpoint-2212",
  "epoch": 19.9774011299435,
  "eval_steps": 500,
  "global_step": 4420,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.05,
      "learning_rate": 2.2624434389140273e-07,
      "loss": 1.5421,
      "step": 10
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.5248868778280546e-07,
      "loss": 1.5264,
      "step": 20
    },
    {
      "epoch": 0.14,
      "learning_rate": 6.787330316742082e-07,
      "loss": 1.5224,
      "step": 30
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.049773755656109e-07,
      "loss": 1.4078,
      "step": 40
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.1312217194570136e-06,
      "loss": 1.3928,
      "step": 50
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.3574660633484164e-06,
      "loss": 1.3119,
      "step": 60
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.583710407239819e-06,
      "loss": 1.2454,
      "step": 70
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.8099547511312218e-06,
      "loss": 1.1405,
      "step": 80
    },
    {
      "epoch": 0.41,
      "learning_rate": 2.0361990950226245e-06,
      "loss": 1.1869,
      "step": 90
    },
    {
      "epoch": 0.45,
      "learning_rate": 2.2624434389140273e-06,
      "loss": 1.0729,
      "step": 100
    },
    {
      "epoch": 0.5,
      "learning_rate": 2.48868778280543e-06,
      "loss": 0.9751,
      "step": 110
    },
    {
      "epoch": 0.54,
      "learning_rate": 2.7149321266968327e-06,
      "loss": 0.9227,
      "step": 120
    },
    {
      "epoch": 0.59,
      "learning_rate": 2.9411764705882355e-06,
      "loss": 0.9586,
      "step": 130
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.167420814479638e-06,
      "loss": 0.9994,
      "step": 140
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.3936651583710413e-06,
      "loss": 0.8943,
      "step": 150
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.6199095022624436e-06,
      "loss": 0.962,
      "step": 160
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.846153846153847e-06,
      "loss": 0.7883,
      "step": 170
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.072398190045249e-06,
      "loss": 1.0049,
      "step": 180
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.298642533936652e-06,
      "loss": 0.7962,
      "step": 190
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.5248868778280546e-06,
      "loss": 0.8276,
      "step": 200
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.751131221719457e-06,
      "loss": 0.7951,
      "step": 210
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.97737556561086e-06,
      "loss": 0.8282,
      "step": 220
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7106598984771574,
      "eval_loss": 0.75575852394104,
      "eval_runtime": 11.6178,
      "eval_samples_per_second": 16.957,
      "eval_steps_per_second": 8.521,
      "step": 221
    },
    {
      "epoch": 1.04,
      "learning_rate": 5.203619909502263e-06,
      "loss": 0.6385,
      "step": 230
    },
    {
      "epoch": 1.08,
      "learning_rate": 5.4298642533936655e-06,
      "loss": 0.6934,
      "step": 240
    },
    {
      "epoch": 1.13,
      "learning_rate": 5.656108597285069e-06,
      "loss": 0.5725,
      "step": 250
    },
    {
      "epoch": 1.18,
      "learning_rate": 5.882352941176471e-06,
      "loss": 0.6576,
      "step": 260
    },
    {
      "epoch": 1.22,
      "learning_rate": 6.108597285067874e-06,
      "loss": 0.5355,
      "step": 270
    },
    {
      "epoch": 1.27,
      "learning_rate": 6.334841628959276e-06,
      "loss": 0.5946,
      "step": 280
    },
    {
      "epoch": 1.31,
      "learning_rate": 6.5610859728506795e-06,
      "loss": 0.7324,
      "step": 290
    },
    {
      "epoch": 1.36,
      "learning_rate": 6.787330316742083e-06,
      "loss": 0.5147,
      "step": 300
    },
    {
      "epoch": 1.4,
      "learning_rate": 7.013574660633484e-06,
      "loss": 0.5388,
      "step": 310
    },
    {
      "epoch": 1.45,
      "learning_rate": 7.239819004524887e-06,
      "loss": 0.6137,
      "step": 320
    },
    {
      "epoch": 1.49,
      "learning_rate": 7.4660633484162904e-06,
      "loss": 0.5181,
      "step": 330
    },
    {
      "epoch": 1.54,
      "learning_rate": 7.692307692307694e-06,
      "loss": 0.4731,
      "step": 340
    },
    {
      "epoch": 1.58,
      "learning_rate": 7.918552036199096e-06,
      "loss": 0.4595,
      "step": 350
    },
    {
      "epoch": 1.63,
      "learning_rate": 8.144796380090498e-06,
      "loss": 0.4189,
      "step": 360
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.371040723981902e-06,
      "loss": 0.3487,
      "step": 370
    },
    {
      "epoch": 1.72,
      "learning_rate": 8.597285067873304e-06,
      "loss": 0.4765,
      "step": 380
    },
    {
      "epoch": 1.76,
      "learning_rate": 8.823529411764707e-06,
      "loss": 0.5676,
      "step": 390
    },
    {
      "epoch": 1.81,
      "learning_rate": 9.049773755656109e-06,
      "loss": 0.5559,
      "step": 400
    },
    {
      "epoch": 1.85,
      "learning_rate": 9.276018099547513e-06,
      "loss": 0.428,
      "step": 410
    },
    {
      "epoch": 1.9,
      "learning_rate": 9.502262443438914e-06,
      "loss": 0.5264,
      "step": 420
    },
    {
      "epoch": 1.94,
      "learning_rate": 9.728506787330318e-06,
      "loss": 0.5958,
      "step": 430
    },
    {
      "epoch": 1.99,
      "learning_rate": 9.95475113122172e-06,
      "loss": 0.2558,
      "step": 440
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7563451776649747,
      "eval_loss": 0.6533816456794739,
      "eval_runtime": 11.5555,
      "eval_samples_per_second": 17.048,
      "eval_steps_per_second": 8.567,
      "step": 442
    },
    {
      "epoch": 2.03,
      "learning_rate": 9.979889391654098e-06,
      "loss": 0.2877,
      "step": 450
    },
    {
      "epoch": 2.08,
      "learning_rate": 9.95475113122172e-06,
      "loss": 0.3569,
      "step": 460
    },
    {
      "epoch": 2.12,
      "learning_rate": 9.929612870789342e-06,
      "loss": 0.4063,
      "step": 470
    },
    {
      "epoch": 2.17,
      "learning_rate": 9.904474610356965e-06,
      "loss": 0.2305,
      "step": 480
    },
    {
      "epoch": 2.21,
      "learning_rate": 9.879336349924586e-06,
      "loss": 0.4066,
      "step": 490
    },
    {
      "epoch": 2.26,
      "learning_rate": 9.854198089492208e-06,
      "loss": 0.3364,
      "step": 500
    },
    {
      "epoch": 2.31,
      "learning_rate": 9.82905982905983e-06,
      "loss": 0.2868,
      "step": 510
    },
    {
      "epoch": 2.35,
      "learning_rate": 9.803921568627451e-06,
      "loss": 0.2592,
      "step": 520
    },
    {
      "epoch": 2.4,
      "learning_rate": 9.778783308195073e-06,
      "loss": 0.3144,
      "step": 530
    },
    {
      "epoch": 2.44,
      "learning_rate": 9.753645047762696e-06,
      "loss": 0.3704,
      "step": 540
    },
    {
      "epoch": 2.49,
      "learning_rate": 9.728506787330318e-06,
      "loss": 0.3673,
      "step": 550
    },
    {
      "epoch": 2.53,
      "learning_rate": 9.70336852689794e-06,
      "loss": 0.2999,
      "step": 560
    },
    {
      "epoch": 2.58,
      "learning_rate": 9.678230266465561e-06,
      "loss": 0.2797,
      "step": 570
    },
    {
      "epoch": 2.62,
      "learning_rate": 9.653092006033184e-06,
      "loss": 0.3679,
      "step": 580
    },
    {
      "epoch": 2.67,
      "learning_rate": 9.627953745600806e-06,
      "loss": 0.3574,
      "step": 590
    },
    {
      "epoch": 2.71,
      "learning_rate": 9.602815485168427e-06,
      "loss": 0.4179,
      "step": 600
    },
    {
      "epoch": 2.76,
      "learning_rate": 9.577677224736049e-06,
      "loss": 0.3687,
      "step": 610
    },
    {
      "epoch": 2.8,
      "learning_rate": 9.552538964303672e-06,
      "loss": 0.3177,
      "step": 620
    },
    {
      "epoch": 2.85,
      "learning_rate": 9.527400703871292e-06,
      "loss": 0.4708,
      "step": 630
    },
    {
      "epoch": 2.89,
      "learning_rate": 9.502262443438914e-06,
      "loss": 0.4286,
      "step": 640
    },
    {
      "epoch": 2.94,
      "learning_rate": 9.477124183006537e-06,
      "loss": 0.3106,
      "step": 650
    },
    {
      "epoch": 2.98,
      "learning_rate": 9.451985922574159e-06,
      "loss": 0.4331,
      "step": 660
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.7614213197969543,
      "eval_loss": 0.5474016666412354,
      "eval_runtime": 11.4119,
      "eval_samples_per_second": 17.263,
      "eval_steps_per_second": 8.675,
      "step": 663
    },
    {
      "epoch": 3.03,
      "learning_rate": 9.42684766214178e-06,
      "loss": 0.3012,
      "step": 670
    },
    {
      "epoch": 3.07,
      "learning_rate": 9.401709401709402e-06,
      "loss": 0.2867,
      "step": 680
    },
    {
      "epoch": 3.12,
      "learning_rate": 9.376571141277025e-06,
      "loss": 0.3876,
      "step": 690
    },
    {
      "epoch": 3.16,
      "learning_rate": 9.351432880844647e-06,
      "loss": 0.2319,
      "step": 700
    },
    {
      "epoch": 3.21,
      "learning_rate": 9.326294620412268e-06,
      "loss": 0.1813,
      "step": 710
    },
    {
      "epoch": 3.25,
      "learning_rate": 9.30115635997989e-06,
      "loss": 0.2617,
      "step": 720
    },
    {
      "epoch": 3.3,
      "learning_rate": 9.276018099547513e-06,
      "loss": 0.3725,
      "step": 730
    },
    {
      "epoch": 3.34,
      "learning_rate": 9.250879839115135e-06,
      "loss": 0.2051,
      "step": 740
    },
    {
      "epoch": 3.39,
      "learning_rate": 9.225741578682756e-06,
      "loss": 0.2708,
      "step": 750
    },
    {
      "epoch": 3.44,
      "learning_rate": 9.200603318250378e-06,
      "loss": 0.2074,
      "step": 760
    },
    {
      "epoch": 3.48,
      "learning_rate": 9.175465057818e-06,
      "loss": 0.2486,
      "step": 770
    },
    {
      "epoch": 3.53,
      "learning_rate": 9.150326797385621e-06,
      "loss": 0.625,
      "step": 780
    },
    {
      "epoch": 3.57,
      "learning_rate": 9.125188536953243e-06,
      "loss": 0.3169,
      "step": 790
    },
    {
      "epoch": 3.62,
      "learning_rate": 9.100050276520866e-06,
      "loss": 0.2199,
      "step": 800
    },
    {
      "epoch": 3.66,
      "learning_rate": 9.074912016088488e-06,
      "loss": 0.3553,
      "step": 810
    },
    {
      "epoch": 3.71,
      "learning_rate": 9.049773755656109e-06,
      "loss": 0.3351,
      "step": 820
    },
    {
      "epoch": 3.75,
      "learning_rate": 9.02463549522373e-06,
      "loss": 0.2544,
      "step": 830
    },
    {
      "epoch": 3.8,
      "learning_rate": 8.999497234791354e-06,
      "loss": 0.292,
      "step": 840
    },
    {
      "epoch": 3.84,
      "learning_rate": 8.974358974358976e-06,
      "loss": 0.2352,
      "step": 850
    },
    {
      "epoch": 3.89,
      "learning_rate": 8.949220713926597e-06,
      "loss": 0.3726,
      "step": 860
    },
    {
      "epoch": 3.93,
      "learning_rate": 8.924082453494219e-06,
      "loss": 0.2584,
      "step": 870
    },
    {
      "epoch": 3.98,
      "learning_rate": 8.89894419306184e-06,
      "loss": 0.3389,
      "step": 880
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.8477157360406091,
      "eval_loss": 0.48645320534706116,
      "eval_runtime": 11.4481,
      "eval_samples_per_second": 17.208,
      "eval_steps_per_second": 8.648,
      "step": 885
    },
    {
      "epoch": 4.02,
      "learning_rate": 8.873805932629462e-06,
      "loss": 0.2025,
      "step": 890
    },
    {
      "epoch": 4.07,
      "learning_rate": 8.848667672197084e-06,
      "loss": 0.3162,
      "step": 900
    },
    {
      "epoch": 4.11,
      "learning_rate": 8.823529411764707e-06,
      "loss": 0.3351,
      "step": 910
    },
    {
      "epoch": 4.16,
      "learning_rate": 8.798391151332328e-06,
      "loss": 0.1491,
      "step": 920
    },
    {
      "epoch": 4.2,
      "learning_rate": 8.77325289089995e-06,
      "loss": 0.2215,
      "step": 930
    },
    {
      "epoch": 4.25,
      "learning_rate": 8.748114630467572e-06,
      "loss": 0.2601,
      "step": 940
    },
    {
      "epoch": 4.29,
      "learning_rate": 8.722976370035195e-06,
      "loss": 0.1189,
      "step": 950
    },
    {
      "epoch": 4.34,
      "learning_rate": 8.697838109602816e-06,
      "loss": 0.1602,
      "step": 960
    },
    {
      "epoch": 4.38,
      "learning_rate": 8.672699849170438e-06,
      "loss": 0.2085,
      "step": 970
    },
    {
      "epoch": 4.43,
      "learning_rate": 8.64756158873806e-06,
      "loss": 0.4583,
      "step": 980
    },
    {
      "epoch": 4.47,
      "learning_rate": 8.622423328305683e-06,
      "loss": 0.1944,
      "step": 990
    },
    {
      "epoch": 4.52,
      "learning_rate": 8.597285067873304e-06,
      "loss": 0.1342,
      "step": 1000
    },
    {
      "epoch": 4.56,
      "learning_rate": 8.572146807440926e-06,
      "loss": 0.3174,
      "step": 1010
    },
    {
      "epoch": 4.61,
      "learning_rate": 8.547008547008548e-06,
      "loss": 0.0971,
      "step": 1020
    },
    {
      "epoch": 4.66,
      "learning_rate": 8.52187028657617e-06,
      "loss": 0.2681,
      "step": 1030
    },
    {
      "epoch": 4.7,
      "learning_rate": 8.496732026143791e-06,
      "loss": 0.3496,
      "step": 1040
    },
    {
      "epoch": 4.75,
      "learning_rate": 8.471593765711414e-06,
      "loss": 0.1596,
      "step": 1050
    },
    {
      "epoch": 4.79,
      "learning_rate": 8.446455505279036e-06,
      "loss": 0.2241,
      "step": 1060
    },
    {
      "epoch": 4.84,
      "learning_rate": 8.421317244846657e-06,
      "loss": 0.1701,
      "step": 1070
    },
    {
      "epoch": 4.88,
      "learning_rate": 8.396178984414279e-06,
      "loss": 0.1263,
      "step": 1080
    },
    {
      "epoch": 4.93,
      "learning_rate": 8.371040723981902e-06,
      "loss": 0.318,
      "step": 1090
    },
    {
      "epoch": 4.97,
      "learning_rate": 8.345902463549524e-06,
      "loss": 0.265,
      "step": 1100
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.8578680203045685,
      "eval_loss": 0.5272068977355957,
      "eval_runtime": 11.4684,
      "eval_samples_per_second": 17.178,
      "eval_steps_per_second": 8.632,
      "step": 1106
    },
    {
      "epoch": 5.02,
      "learning_rate": 8.320764203117145e-06,
      "loss": 0.259,
      "step": 1110
    },
    {
      "epoch": 5.06,
      "learning_rate": 8.295625942684767e-06,
      "loss": 0.1618,
      "step": 1120
    },
    {
      "epoch": 5.11,
      "learning_rate": 8.270487682252389e-06,
      "loss": 0.1645,
      "step": 1130
    },
    {
      "epoch": 5.15,
      "learning_rate": 8.24534942182001e-06,
      "loss": 0.1389,
      "step": 1140
    },
    {
      "epoch": 5.2,
      "learning_rate": 8.220211161387632e-06,
      "loss": 0.2253,
      "step": 1150
    },
    {
      "epoch": 5.24,
      "learning_rate": 8.195072900955255e-06,
      "loss": 0.2173,
      "step": 1160
    },
    {
      "epoch": 5.29,
      "learning_rate": 8.169934640522877e-06,
      "loss": 0.0624,
      "step": 1170
    },
    {
      "epoch": 5.33,
      "learning_rate": 8.144796380090498e-06,
      "loss": 0.1958,
      "step": 1180
    },
    {
      "epoch": 5.38,
      "learning_rate": 8.11965811965812e-06,
      "loss": 0.1418,
      "step": 1190
    },
    {
      "epoch": 5.42,
      "learning_rate": 8.094519859225743e-06,
      "loss": 0.2998,
      "step": 1200
    },
    {
      "epoch": 5.47,
      "learning_rate": 8.069381598793365e-06,
      "loss": 0.0822,
      "step": 1210
    },
    {
      "epoch": 5.51,
      "learning_rate": 8.044243338360986e-06,
      "loss": 0.4893,
      "step": 1220
    },
    {
      "epoch": 5.56,
      "learning_rate": 8.019105077928608e-06,
      "loss": 0.2722,
      "step": 1230
    },
    {
      "epoch": 5.6,
      "learning_rate": 7.993966817496231e-06,
      "loss": 0.0453,
      "step": 1240
    },
    {
      "epoch": 5.65,
      "learning_rate": 7.968828557063853e-06,
      "loss": 0.124,
      "step": 1250
    },
    {
      "epoch": 5.69,
      "learning_rate": 7.943690296631473e-06,
      "loss": 0.267,
      "step": 1260
    },
    {
      "epoch": 5.74,
      "learning_rate": 7.918552036199096e-06,
      "loss": 0.3905,
      "step": 1270
    },
    {
      "epoch": 5.79,
      "learning_rate": 7.893413775766717e-06,
      "loss": 0.3556,
      "step": 1280
    },
    {
      "epoch": 5.83,
      "learning_rate": 7.868275515334339e-06,
      "loss": 0.2221,
      "step": 1290
    },
    {
      "epoch": 5.88,
      "learning_rate": 7.84313725490196e-06,
      "loss": 0.0581,
      "step": 1300
    },
    {
      "epoch": 5.92,
      "learning_rate": 7.817998994469584e-06,
      "loss": 0.1313,
      "step": 1310
    },
    {
      "epoch": 5.97,
      "learning_rate": 7.792860734037206e-06,
      "loss": 0.1638,
      "step": 1320
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.9035532994923858,
      "eval_loss": 0.4494355022907257,
      "eval_runtime": 11.4563,
      "eval_samples_per_second": 17.196,
      "eval_steps_per_second": 8.642,
      "step": 1327
    },
    {
      "epoch": 6.01,
      "learning_rate": 7.767722473604827e-06,
      "loss": 0.0616,
      "step": 1330
    },
    {
      "epoch": 6.06,
      "learning_rate": 7.742584213172449e-06,
      "loss": 0.181,
      "step": 1340
    },
    {
      "epoch": 6.1,
      "learning_rate": 7.717445952740072e-06,
      "loss": 0.3007,
      "step": 1350
    },
    {
      "epoch": 6.15,
      "learning_rate": 7.692307692307694e-06,
      "loss": 0.083,
      "step": 1360
    },
    {
      "epoch": 6.19,
      "learning_rate": 7.667169431875315e-06,
      "loss": 0.1376,
      "step": 1370
    },
    {
      "epoch": 6.24,
      "learning_rate": 7.642031171442937e-06,
      "loss": 0.1786,
      "step": 1380
    },
    {
      "epoch": 6.28,
      "learning_rate": 7.616892911010558e-06,
      "loss": 0.2023,
      "step": 1390
    },
    {
      "epoch": 6.33,
      "learning_rate": 7.59175465057818e-06,
      "loss": 0.2793,
      "step": 1400
    },
    {
      "epoch": 6.37,
      "learning_rate": 7.566616390145802e-06,
      "loss": 0.0139,
      "step": 1410
    },
    {
      "epoch": 6.42,
      "learning_rate": 7.541478129713424e-06,
      "loss": 0.1125,
      "step": 1420
    },
    {
      "epoch": 6.46,
      "learning_rate": 7.516339869281046e-06,
      "loss": 0.1029,
      "step": 1430
    },
    {
      "epoch": 6.51,
      "learning_rate": 7.491201608848668e-06,
      "loss": 0.0955,
      "step": 1440
    },
    {
      "epoch": 6.55,
      "learning_rate": 7.4660633484162904e-06,
      "loss": 0.242,
      "step": 1450
    },
    {
      "epoch": 6.6,
      "learning_rate": 7.440925087983912e-06,
      "loss": 0.3091,
      "step": 1460
    },
    {
      "epoch": 6.64,
      "learning_rate": 7.4157868275515345e-06,
      "loss": 0.2234,
      "step": 1470
    },
    {
      "epoch": 6.69,
      "learning_rate": 7.390648567119156e-06,
      "loss": 0.135,
      "step": 1480
    },
    {
      "epoch": 6.73,
      "learning_rate": 7.3655103066867785e-06,
      "loss": 0.0903,
      "step": 1490
    },
    {
      "epoch": 6.78,
      "learning_rate": 7.340372046254399e-06,
      "loss": 0.1068,
      "step": 1500
    },
    {
      "epoch": 6.82,
      "learning_rate": 7.315233785822022e-06,
      "loss": 0.1832,
      "step": 1510
    },
    {
      "epoch": 6.87,
      "learning_rate": 7.290095525389643e-06,
      "loss": 0.338,
      "step": 1520
    },
    {
      "epoch": 6.92,
      "learning_rate": 7.264957264957266e-06,
      "loss": 0.1379,
      "step": 1530
    },
    {
      "epoch": 6.96,
      "learning_rate": 7.239819004524887e-06,
      "loss": 0.1767,
      "step": 1540
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.8883248730964467,
      "eval_loss": 0.580611526966095,
      "eval_runtime": 11.2566,
      "eval_samples_per_second": 17.501,
      "eval_steps_per_second": 8.795,
      "step": 1548
    },
    {
      "epoch": 7.01,
      "learning_rate": 7.214680744092509e-06,
      "loss": 0.1469,
      "step": 1550
    },
    {
      "epoch": 7.05,
      "learning_rate": 7.189542483660131e-06,
      "loss": 0.3954,
      "step": 1560
    },
    {
      "epoch": 7.1,
      "learning_rate": 7.164404223227753e-06,
      "loss": 0.0218,
      "step": 1570
    },
    {
      "epoch": 7.14,
      "learning_rate": 7.139265962795375e-06,
      "loss": 0.2107,
      "step": 1580
    },
    {
      "epoch": 7.19,
      "learning_rate": 7.114127702362997e-06,
      "loss": 0.2268,
      "step": 1590
    },
    {
      "epoch": 7.23,
      "learning_rate": 7.088989441930619e-06,
      "loss": 0.1166,
      "step": 1600
    },
    {
      "epoch": 7.28,
      "learning_rate": 7.063851181498241e-06,
      "loss": 0.0958,
      "step": 1610
    },
    {
      "epoch": 7.32,
      "learning_rate": 7.038712921065863e-06,
      "loss": 0.0712,
      "step": 1620
    },
    {
      "epoch": 7.37,
      "learning_rate": 7.013574660633484e-06,
      "loss": 0.3218,
      "step": 1630
    },
    {
      "epoch": 7.41,
      "learning_rate": 6.9884364002011066e-06,
      "loss": 0.1145,
      "step": 1640
    },
    {
      "epoch": 7.46,
      "learning_rate": 6.963298139768728e-06,
      "loss": 0.3381,
      "step": 1650
    },
    {
      "epoch": 7.5,
      "learning_rate": 6.938159879336351e-06,
      "loss": 0.2609,
      "step": 1660
    },
    {
      "epoch": 7.55,
      "learning_rate": 6.913021618903972e-06,
      "loss": 0.1109,
      "step": 1670
    },
    {
      "epoch": 7.59,
      "learning_rate": 6.887883358471595e-06,
      "loss": 0.1425,
      "step": 1680
    },
    {
      "epoch": 7.64,
      "learning_rate": 6.862745098039216e-06,
      "loss": 0.0786,
      "step": 1690
    },
    {
      "epoch": 7.68,
      "learning_rate": 6.837606837606839e-06,
      "loss": 0.0241,
      "step": 1700
    },
    {
      "epoch": 7.73,
      "learning_rate": 6.81246857717446e-06,
      "loss": 0.0539,
      "step": 1710
    },
    {
      "epoch": 7.77,
      "learning_rate": 6.787330316742083e-06,
      "loss": 0.1844,
      "step": 1720
    },
    {
      "epoch": 7.82,
      "learning_rate": 6.762192056309704e-06,
      "loss": 0.0446,
      "step": 1730
    },
    {
      "epoch": 7.86,
      "learning_rate": 6.737053795877327e-06,
      "loss": 0.2129,
      "step": 1740
    },
    {
      "epoch": 7.91,
      "learning_rate": 6.7119155354449474e-06,
      "loss": 0.065,
      "step": 1750
    },
    {
      "epoch": 7.95,
      "learning_rate": 6.686777275012569e-06,
      "loss": 0.1402,
      "step": 1760
    },
    {
      "epoch": 8.0,
      "learning_rate": 6.6616390145801915e-06,
      "loss": 0.1587,
      "step": 1770
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.8984771573604061,
      "eval_loss": 0.5724393725395203,
      "eval_runtime": 11.4108,
      "eval_samples_per_second": 17.264,
      "eval_steps_per_second": 8.676,
      "step": 1770
    },
    {
      "epoch": 8.05,
      "learning_rate": 6.636500754147813e-06,
      "loss": 0.1339,
      "step": 1780
    },
    {
      "epoch": 8.09,
      "learning_rate": 6.6113624937154355e-06,
      "loss": 0.122,
      "step": 1790
    },
    {
      "epoch": 8.14,
      "learning_rate": 6.586224233283057e-06,
      "loss": 0.2488,
      "step": 1800
    },
    {
      "epoch": 8.18,
      "learning_rate": 6.5610859728506795e-06,
      "loss": 0.0059,
      "step": 1810
    },
    {
      "epoch": 8.23,
      "learning_rate": 6.535947712418301e-06,
      "loss": 0.0735,
      "step": 1820
    },
    {
      "epoch": 8.27,
      "learning_rate": 6.5108094519859236e-06,
      "loss": 0.2475,
      "step": 1830
    },
    {
      "epoch": 8.32,
      "learning_rate": 6.485671191553545e-06,
      "loss": 0.1837,
      "step": 1840
    },
    {
      "epoch": 8.36,
      "learning_rate": 6.460532931121168e-06,
      "loss": 0.0566,
      "step": 1850
    },
    {
      "epoch": 8.41,
      "learning_rate": 6.435394670688789e-06,
      "loss": 0.1014,
      "step": 1860
    },
    {
      "epoch": 8.45,
      "learning_rate": 6.410256410256412e-06,
      "loss": 0.04,
      "step": 1870
    },
    {
      "epoch": 8.5,
      "learning_rate": 6.385118149824032e-06,
      "loss": 0.0315,
      "step": 1880
    },
    {
      "epoch": 8.54,
      "learning_rate": 6.359979889391654e-06,
      "loss": 0.0776,
      "step": 1890
    },
    {
      "epoch": 8.59,
      "learning_rate": 6.334841628959276e-06,
      "loss": 0.1041,
      "step": 1900
    },
    {
      "epoch": 8.63,
      "learning_rate": 6.309703368526898e-06,
      "loss": 0.1221,
      "step": 1910
    },
    {
      "epoch": 8.68,
      "learning_rate": 6.28456510809452e-06,
      "loss": 0.0816,
      "step": 1920
    },
    {
      "epoch": 8.72,
      "learning_rate": 6.259426847662142e-06,
      "loss": 0.2014,
      "step": 1930
    },
    {
      "epoch": 8.77,
      "learning_rate": 6.2342885872297644e-06,
      "loss": 0.0842,
      "step": 1940
    },
    {
      "epoch": 8.81,
      "learning_rate": 6.209150326797386e-06,
      "loss": 0.2245,
      "step": 1950
    },
    {
      "epoch": 8.86,
      "learning_rate": 6.1840120663650085e-06,
      "loss": 0.1937,
      "step": 1960
    },
    {
      "epoch": 8.9,
      "learning_rate": 6.15887380593263e-06,
      "loss": 0.128,
      "step": 1970
    },
    {
      "epoch": 8.95,
      "learning_rate": 6.1337355455002525e-06,
      "loss": 0.1745,
      "step": 1980
    },
    {
      "epoch": 8.99,
      "learning_rate": 6.108597285067874e-06,
      "loss": 0.5961,
      "step": 1990
    },
    {
      "epoch": 9.0,
      "eval_accuracy": 0.883248730964467,
      "eval_loss": 0.6212002635002136,
      "eval_runtime": 11.4043,
      "eval_samples_per_second": 17.274,
      "eval_steps_per_second": 8.681,
      "step": 1991
    },
    {
      "epoch": 9.04,
      "learning_rate": 6.083459024635495e-06,
      "loss": 0.0176,
      "step": 2000
    },
    {
      "epoch": 9.08,
      "learning_rate": 6.058320764203117e-06,
      "loss": 0.132,
      "step": 2010
    },
    {
      "epoch": 9.13,
      "learning_rate": 6.033182503770739e-06,
      "loss": 0.0346,
      "step": 2020
    },
    {
      "epoch": 9.18,
      "learning_rate": 6.008044243338361e-06,
      "loss": 0.0088,
      "step": 2030
    },
    {
      "epoch": 9.22,
      "learning_rate": 5.982905982905983e-06,
      "loss": 0.0187,
      "step": 2040
    },
    {
      "epoch": 9.27,
      "learning_rate": 5.957767722473605e-06,
      "loss": 0.056,
      "step": 2050
    },
    {
      "epoch": 9.31,
      "learning_rate": 5.932629462041227e-06,
      "loss": 0.1644,
      "step": 2060
    },
    {
      "epoch": 9.36,
      "learning_rate": 5.907491201608849e-06,
      "loss": 0.1428,
      "step": 2070
    },
    {
      "epoch": 9.4,
      "learning_rate": 5.882352941176471e-06,
      "loss": 0.2165,
      "step": 2080
    },
    {
      "epoch": 9.45,
      "learning_rate": 5.857214680744093e-06,
      "loss": 0.1284,
      "step": 2090
    },
    {
      "epoch": 9.49,
      "learning_rate": 5.832076420311715e-06,
      "loss": 0.1231,
      "step": 2100
    },
    {
      "epoch": 9.54,
      "learning_rate": 5.806938159879337e-06,
      "loss": 0.1071,
      "step": 2110
    },
    {
      "epoch": 9.58,
      "learning_rate": 5.781799899446959e-06,
      "loss": 0.09,
      "step": 2120
    },
    {
      "epoch": 9.63,
      "learning_rate": 5.7566616390145806e-06,
      "loss": 0.0583,
      "step": 2130
    },
    {
      "epoch": 9.67,
      "learning_rate": 5.731523378582202e-06,
      "loss": 0.121,
      "step": 2140
    },
    {
      "epoch": 9.72,
      "learning_rate": 5.706385118149825e-06,
      "loss": 0.096,
      "step": 2150
    },
    {
      "epoch": 9.76,
      "learning_rate": 5.681246857717446e-06,
      "loss": 0.0394,
      "step": 2160
    },
    {
      "epoch": 9.81,
      "learning_rate": 5.656108597285069e-06,
      "loss": 0.0298,
      "step": 2170
    },
    {
      "epoch": 9.85,
      "learning_rate": 5.63097033685269e-06,
      "loss": 0.2333,
      "step": 2180
    },
    {
      "epoch": 9.9,
      "learning_rate": 5.605832076420313e-06,
      "loss": 0.2992,
      "step": 2190
    },
    {
      "epoch": 9.94,
      "learning_rate": 5.580693815987934e-06,
      "loss": 0.2866,
      "step": 2200
    },
    {
      "epoch": 9.99,
      "learning_rate": 5.555555555555557e-06,
      "loss": 0.1293,
      "step": 2210
    },
    {
      "epoch": 10.0,
      "eval_accuracy": 0.9238578680203046,
      "eval_loss": 0.46713295578956604,
      "eval_runtime": 11.499,
      "eval_samples_per_second": 17.132,
      "eval_steps_per_second": 8.609,
      "step": 2212
    },
    {
      "epoch": 10.03,
      "learning_rate": 5.530417295123178e-06,
      "loss": 0.1584,
      "step": 2220
    },
    {
      "epoch": 10.08,
      "learning_rate": 5.505279034690801e-06,
      "loss": 0.0545,
      "step": 2230
    },
    {
      "epoch": 10.12,
      "learning_rate": 5.480140774258422e-06,
      "loss": 0.0611,
      "step": 2240
    },
    {
      "epoch": 10.17,
      "learning_rate": 5.455002513826043e-06,
      "loss": 0.0912,
      "step": 2250
    },
    {
      "epoch": 10.21,
      "learning_rate": 5.4298642533936655e-06,
      "loss": 0.2301,
      "step": 2260
    },
    {
      "epoch": 10.26,
      "learning_rate": 5.404725992961287e-06,
      "loss": 0.0344,
      "step": 2270
    },
    {
      "epoch": 10.31,
      "learning_rate": 5.3795877325289095e-06,
      "loss": 0.1263,
      "step": 2280
    },
    {
      "epoch": 10.35,
      "learning_rate": 5.354449472096531e-06,
      "loss": 0.1295,
      "step": 2290
    },
    {
      "epoch": 10.4,
      "learning_rate": 5.3293112116641535e-06,
      "loss": 0.1283,
      "step": 2300
    },
    {
      "epoch": 10.44,
      "learning_rate": 5.304172951231775e-06,
      "loss": 0.0085,
      "step": 2310
    },
    {
      "epoch": 10.49,
      "learning_rate": 5.2790346907993975e-06,
      "loss": 0.0396,
      "step": 2320
    },
    {
      "epoch": 10.53,
      "learning_rate": 5.253896430367019e-06,
      "loss": 0.0942,
      "step": 2330
    },
    {
      "epoch": 10.58,
      "learning_rate": 5.2287581699346416e-06,
      "loss": 0.099,
      "step": 2340
    },
    {
      "epoch": 10.62,
      "learning_rate": 5.203619909502263e-06,
      "loss": 0.0341,
      "step": 2350
    },
    {
      "epoch": 10.67,
      "learning_rate": 5.178481649069886e-06,
      "loss": 0.2573,
      "step": 2360
    },
    {
      "epoch": 10.71,
      "learning_rate": 5.153343388637506e-06,
      "loss": 0.0754,
      "step": 2370
    },
    {
      "epoch": 10.76,
      "learning_rate": 5.128205128205128e-06,
      "loss": 0.002,
      "step": 2380
    },
    {
      "epoch": 10.8,
      "learning_rate": 5.10306686777275e-06,
      "loss": 0.0832,
      "step": 2390
    },
    {
      "epoch": 10.85,
      "learning_rate": 5.077928607340372e-06,
      "loss": 0.14,
      "step": 2400
    },
    {
      "epoch": 10.89,
      "learning_rate": 5.052790346907994e-06,
      "loss": 0.0127,
      "step": 2410
    },
    {
      "epoch": 10.94,
      "learning_rate": 5.027652086475616e-06,
      "loss": 0.1633,
      "step": 2420
    },
    {
      "epoch": 10.98,
      "learning_rate": 5.002513826043238e-06,
      "loss": 0.2886,
      "step": 2430
    },
    {
      "epoch": 11.0,
      "eval_accuracy": 0.9137055837563451,
      "eval_loss": 0.4880651831626892,
      "eval_runtime": 11.3471,
      "eval_samples_per_second": 17.361,
      "eval_steps_per_second": 8.725,
      "step": 2433
    },
    {
      "epoch": 11.03,
      "learning_rate": 4.97737556561086e-06,
      "loss": 0.0932,
      "step": 2440
    },
    {
      "epoch": 11.07,
      "learning_rate": 4.9522373051784824e-06,
      "loss": 0.233,
      "step": 2450
    },
    {
      "epoch": 11.12,
      "learning_rate": 4.927099044746104e-06,
      "loss": 0.0427,
      "step": 2460
    },
    {
      "epoch": 11.16,
      "learning_rate": 4.901960784313726e-06,
      "loss": 0.1207,
      "step": 2470
    },
    {
      "epoch": 11.21,
      "learning_rate": 4.876822523881348e-06,
      "loss": 0.0242,
      "step": 2480
    },
    {
      "epoch": 11.25,
      "learning_rate": 4.85168426344897e-06,
      "loss": 0.022,
      "step": 2490
    },
    {
      "epoch": 11.3,
      "learning_rate": 4.826546003016592e-06,
      "loss": 0.048,
      "step": 2500
    },
    {
      "epoch": 11.34,
      "learning_rate": 4.801407742584214e-06,
      "loss": 0.0177,
      "step": 2510
    },
    {
      "epoch": 11.39,
      "learning_rate": 4.776269482151836e-06,
      "loss": 0.0086,
      "step": 2520
    },
    {
      "epoch": 11.44,
      "learning_rate": 4.751131221719457e-06,
      "loss": 0.0145,
      "step": 2530
    },
    {
      "epoch": 11.48,
      "learning_rate": 4.725992961287079e-06,
      "loss": 0.0371,
      "step": 2540
    },
    {
      "epoch": 11.53,
      "learning_rate": 4.700854700854701e-06,
      "loss": 0.3427,
      "step": 2550
    },
    {
      "epoch": 11.57,
      "learning_rate": 4.675716440422323e-06,
      "loss": 0.1189,
      "step": 2560
    },
    {
      "epoch": 11.62,
      "learning_rate": 4.650578179989945e-06,
      "loss": 0.0668,
      "step": 2570
    },
    {
      "epoch": 11.66,
      "learning_rate": 4.625439919557567e-06,
      "loss": 0.0659,
      "step": 2580
    },
    {
      "epoch": 11.71,
      "learning_rate": 4.600301659125189e-06,
      "loss": 0.1244,
      "step": 2590
    },
    {
      "epoch": 11.75,
      "learning_rate": 4.5751633986928105e-06,
      "loss": 0.0327,
      "step": 2600
    },
    {
      "epoch": 11.8,
      "learning_rate": 4.550025138260433e-06,
      "loss": 0.1169,
      "step": 2610
    },
    {
      "epoch": 11.84,
      "learning_rate": 4.5248868778280546e-06,
      "loss": 0.2857,
      "step": 2620
    },
    {
      "epoch": 11.89,
      "learning_rate": 4.499748617395677e-06,
      "loss": 0.0479,
      "step": 2630
    },
    {
      "epoch": 11.93,
      "learning_rate": 4.474610356963299e-06,
      "loss": 0.0672,
      "step": 2640
    },
    {
      "epoch": 11.98,
      "learning_rate": 4.44947209653092e-06,
      "loss": 0.1336,
      "step": 2650
    },
    {
      "epoch": 12.0,
      "eval_accuracy": 0.9035532994923858,
      "eval_loss": 0.6028070449829102,
      "eval_runtime": 11.4741,
      "eval_samples_per_second": 17.169,
      "eval_steps_per_second": 8.628,
      "step": 2655
    },
    {
      "epoch": 12.02,
      "learning_rate": 4.424333836098542e-06,
      "loss": 0.0059,
      "step": 2660
    },
    {
      "epoch": 12.07,
      "learning_rate": 4.399195575666164e-06,
      "loss": 0.0869,
      "step": 2670
    },
    {
      "epoch": 12.11,
      "learning_rate": 4.374057315233786e-06,
      "loss": 0.0766,
      "step": 2680
    },
    {
      "epoch": 12.16,
      "learning_rate": 4.348919054801408e-06,
      "loss": 0.0793,
      "step": 2690
    },
    {
      "epoch": 12.2,
      "learning_rate": 4.32378079436903e-06,
      "loss": 0.0057,
      "step": 2700
    },
    {
      "epoch": 12.25,
      "learning_rate": 4.298642533936652e-06,
      "loss": 0.079,
      "step": 2710
    },
    {
      "epoch": 12.29,
      "learning_rate": 4.273504273504274e-06,
      "loss": 0.0177,
      "step": 2720
    },
    {
      "epoch": 12.34,
      "learning_rate": 4.2483660130718954e-06,
      "loss": 0.0025,
      "step": 2730
    },
    {
      "epoch": 12.38,
      "learning_rate": 4.223227752639518e-06,
      "loss": 0.0953,
      "step": 2740
    },
    {
      "epoch": 12.43,
      "learning_rate": 4.1980894922071395e-06,
      "loss": 0.0128,
      "step": 2750
    },
    {
      "epoch": 12.47,
      "learning_rate": 4.172951231774762e-06,
      "loss": 0.2451,
      "step": 2760
    },
    {
      "epoch": 12.52,
      "learning_rate": 4.1478129713423835e-06,
      "loss": 0.2394,
      "step": 2770
    },
    {
      "epoch": 12.56,
      "learning_rate": 4.122674710910005e-06,
      "loss": 0.0102,
      "step": 2780
    },
    {
      "epoch": 12.61,
      "learning_rate": 4.0975364504776275e-06,
      "loss": 0.0627,
      "step": 2790
    },
    {
      "epoch": 12.66,
      "learning_rate": 4.072398190045249e-06,
      "loss": 0.0262,
      "step": 2800
    },
    {
      "epoch": 12.7,
      "learning_rate": 4.0472599296128715e-06,
      "loss": 0.0216,
      "step": 2810
    },
    {
      "epoch": 12.75,
      "learning_rate": 4.022121669180493e-06,
      "loss": 0.0017,
      "step": 2820
    },
    {
      "epoch": 12.79,
      "learning_rate": 3.9969834087481156e-06,
      "loss": 0.0313,
      "step": 2830
    },
    {
      "epoch": 12.84,
      "learning_rate": 3.971845148315736e-06,
      "loss": 0.0889,
      "step": 2840
    },
    {
      "epoch": 12.88,
      "learning_rate": 3.946706887883359e-06,
      "loss": 0.0683,
      "step": 2850
    },
    {
      "epoch": 12.93,
      "learning_rate": 3.92156862745098e-06,
      "loss": 0.0795,
      "step": 2860
    },
    {
      "epoch": 12.97,
      "learning_rate": 3.896430367018603e-06,
      "loss": 0.0061,
      "step": 2870
    },
    {
      "epoch": 13.0,
      "eval_accuracy": 0.9035532994923858,
      "eval_loss": 0.6166318655014038,
      "eval_runtime": 11.4255,
      "eval_samples_per_second": 17.242,
      "eval_steps_per_second": 8.665,
      "step": 2876
    },
    {
      "epoch": 13.02,
      "learning_rate": 3.871292106586224e-06,
      "loss": 0.032,
      "step": 2880
    },
    {
      "epoch": 13.06,
      "learning_rate": 3.846153846153847e-06,
      "loss": 0.0039,
      "step": 2890
    },
    {
      "epoch": 13.11,
      "learning_rate": 3.821015585721468e-06,
      "loss": 0.2127,
      "step": 2900
    },
    {
      "epoch": 13.15,
      "learning_rate": 3.79587732528909e-06,
      "loss": 0.1421,
      "step": 2910
    },
    {
      "epoch": 13.2,
      "learning_rate": 3.770739064856712e-06,
      "loss": 0.0917,
      "step": 2920
    },
    {
      "epoch": 13.24,
      "learning_rate": 3.745600804424334e-06,
      "loss": 0.0375,
      "step": 2930
    },
    {
      "epoch": 13.29,
      "learning_rate": 3.720462543991956e-06,
      "loss": 0.1163,
      "step": 2940
    },
    {
      "epoch": 13.33,
      "learning_rate": 3.695324283559578e-06,
      "loss": 0.0423,
      "step": 2950
    },
    {
      "epoch": 13.38,
      "learning_rate": 3.6701860231271996e-06,
      "loss": 0.1147,
      "step": 2960
    },
    {
      "epoch": 13.42,
      "learning_rate": 3.6450477626948216e-06,
      "loss": 0.0798,
      "step": 2970
    },
    {
      "epoch": 13.47,
      "learning_rate": 3.6199095022624436e-06,
      "loss": 0.0008,
      "step": 2980
    },
    {
      "epoch": 13.51,
      "learning_rate": 3.5947712418300657e-06,
      "loss": 0.0423,
      "step": 2990
    },
    {
      "epoch": 13.56,
      "learning_rate": 3.5696329813976877e-06,
      "loss": 0.0006,
      "step": 3000
    },
    {
      "epoch": 13.6,
      "learning_rate": 3.5444947209653097e-06,
      "loss": 0.0101,
      "step": 3010
    },
    {
      "epoch": 13.65,
      "learning_rate": 3.5193564605329317e-06,
      "loss": 0.2554,
      "step": 3020
    },
    {
      "epoch": 13.69,
      "learning_rate": 3.4942182001005533e-06,
      "loss": 0.0564,
      "step": 3030
    },
    {
      "epoch": 13.74,
      "learning_rate": 3.4690799396681753e-06,
      "loss": 0.0761,
      "step": 3040
    },
    {
      "epoch": 13.79,
      "learning_rate": 3.4439416792357973e-06,
      "loss": 0.0437,
      "step": 3050
    },
    {
      "epoch": 13.83,
      "learning_rate": 3.4188034188034193e-06,
      "loss": 0.1901,
      "step": 3060
    },
    {
      "epoch": 13.88,
      "learning_rate": 3.3936651583710413e-06,
      "loss": 0.0284,
      "step": 3070
    },
    {
      "epoch": 13.92,
      "learning_rate": 3.3685268979386634e-06,
      "loss": 0.0715,
      "step": 3080
    },
    {
      "epoch": 13.97,
      "learning_rate": 3.3433886375062845e-06,
      "loss": 0.127,
      "step": 3090
    },
    {
      "epoch": 14.0,
      "eval_accuracy": 0.9086294416243654,
      "eval_loss": 0.6385547518730164,
      "eval_runtime": 11.4426,
      "eval_samples_per_second": 17.216,
      "eval_steps_per_second": 8.652,
      "step": 3097
    },
    {
      "epoch": 14.01,
      "learning_rate": 3.3182503770739065e-06,
      "loss": 0.0727,
      "step": 3100
    },
    {
      "epoch": 14.06,
      "learning_rate": 3.2931121166415285e-06,
      "loss": 0.0006,
      "step": 3110
    },
    {
      "epoch": 14.1,
      "learning_rate": 3.2679738562091506e-06,
      "loss": 0.0011,
      "step": 3120
    },
    {
      "epoch": 14.15,
      "learning_rate": 3.2428355957767726e-06,
      "loss": 0.0675,
      "step": 3130
    },
    {
      "epoch": 14.19,
      "learning_rate": 3.2176973353443946e-06,
      "loss": 0.0083,
      "step": 3140
    },
    {
      "epoch": 14.24,
      "learning_rate": 3.192559074912016e-06,
      "loss": 0.0101,
      "step": 3150
    },
    {
      "epoch": 14.28,
      "learning_rate": 3.167420814479638e-06,
      "loss": 0.0186,
      "step": 3160
    },
    {
      "epoch": 14.33,
      "learning_rate": 3.14228255404726e-06,
      "loss": 0.0423,
      "step": 3170
    },
    {
      "epoch": 14.37,
      "learning_rate": 3.1171442936148822e-06,
      "loss": 0.0093,
      "step": 3180
    },
    {
      "epoch": 14.42,
      "learning_rate": 3.0920060331825042e-06,
      "loss": 0.0577,
      "step": 3190
    },
    {
      "epoch": 14.46,
      "learning_rate": 3.0668677727501262e-06,
      "loss": 0.003,
      "step": 3200
    },
    {
      "epoch": 14.51,
      "learning_rate": 3.0417295123177474e-06,
      "loss": 0.1598,
      "step": 3210
    },
    {
      "epoch": 14.55,
      "learning_rate": 3.0165912518853694e-06,
      "loss": 0.0112,
      "step": 3220
    },
    {
      "epoch": 14.6,
      "learning_rate": 2.9914529914529914e-06,
      "loss": 0.1371,
      "step": 3230
    },
    {
      "epoch": 14.64,
      "learning_rate": 2.9663147310206134e-06,
      "loss": 0.0799,
      "step": 3240
    },
    {
      "epoch": 14.69,
      "learning_rate": 2.9411764705882355e-06,
      "loss": 0.2422,
      "step": 3250
    },
    {
      "epoch": 14.73,
      "learning_rate": 2.9160382101558575e-06,
      "loss": 0.0882,
      "step": 3260
    },
    {
      "epoch": 14.78,
      "learning_rate": 2.8908999497234795e-06,
      "loss": 0.0022,
      "step": 3270
    },
    {
      "epoch": 14.82,
      "learning_rate": 2.865761689291101e-06,
      "loss": 0.038,
      "step": 3280
    },
    {
      "epoch": 14.87,
      "learning_rate": 2.840623428858723e-06,
      "loss": 0.0602,
      "step": 3290
    },
    {
      "epoch": 14.92,
      "learning_rate": 2.815485168426345e-06,
      "loss": 0.0054,
      "step": 3300
    },
    {
      "epoch": 14.96,
      "learning_rate": 2.790346907993967e-06,
      "loss": 0.1755,
      "step": 3310
    },
    {
      "epoch": 15.0,
      "eval_accuracy": 0.9035532994923858,
      "eval_loss": 0.6393283605575562,
      "eval_runtime": 11.4409,
      "eval_samples_per_second": 17.219,
      "eval_steps_per_second": 8.653,
      "step": 3318
    },
    {
      "epoch": 15.01,
      "learning_rate": 2.765208647561589e-06,
      "loss": 0.1946,
      "step": 3320
    },
    {
      "epoch": 15.05,
      "learning_rate": 2.740070387129211e-06,
      "loss": 0.0461,
      "step": 3330
    },
    {
      "epoch": 15.1,
      "learning_rate": 2.7149321266968327e-06,
      "loss": 0.0123,
      "step": 3340
    },
    {
      "epoch": 15.14,
      "learning_rate": 2.6897938662644547e-06,
      "loss": 0.0279,
      "step": 3350
    },
    {
      "epoch": 15.19,
      "learning_rate": 2.6646556058320768e-06,
      "loss": 0.0621,
      "step": 3360
    },
    {
      "epoch": 15.23,
      "learning_rate": 2.6395173453996988e-06,
      "loss": 0.0001,
      "step": 3370
    },
    {
      "epoch": 15.28,
      "learning_rate": 2.6143790849673208e-06,
      "loss": 0.1222,
      "step": 3380
    },
    {
      "epoch": 15.32,
      "learning_rate": 2.589240824534943e-06,
      "loss": 0.0328,
      "step": 3390
    },
    {
      "epoch": 15.37,
      "learning_rate": 2.564102564102564e-06,
      "loss": 0.1822,
      "step": 3400
    },
    {
      "epoch": 15.41,
      "learning_rate": 2.538964303670186e-06,
      "loss": 0.0244,
      "step": 3410
    },
    {
      "epoch": 15.46,
      "learning_rate": 2.513826043237808e-06,
      "loss": 0.2829,
      "step": 3420
    },
    {
      "epoch": 15.5,
      "learning_rate": 2.48868778280543e-06,
      "loss": 0.0047,
      "step": 3430
    },
    {
      "epoch": 15.55,
      "learning_rate": 2.463549522373052e-06,
      "loss": 0.1227,
      "step": 3440
    },
    {
      "epoch": 15.59,
      "learning_rate": 2.438411261940674e-06,
      "loss": 0.1557,
      "step": 3450
    },
    {
      "epoch": 15.64,
      "learning_rate": 2.413273001508296e-06,
      "loss": 0.0398,
      "step": 3460
    },
    {
      "epoch": 15.68,
      "learning_rate": 2.388134741075918e-06,
      "loss": 0.0064,
      "step": 3470
    },
    {
      "epoch": 15.73,
      "learning_rate": 2.3629964806435396e-06,
      "loss": 0.1254,
      "step": 3480
    },
    {
      "epoch": 15.77,
      "learning_rate": 2.3378582202111617e-06,
      "loss": 0.0909,
      "step": 3490
    },
    {
      "epoch": 15.82,
      "learning_rate": 2.3127199597787837e-06,
      "loss": 0.0617,
      "step": 3500
    },
    {
      "epoch": 15.86,
      "learning_rate": 2.2875816993464053e-06,
      "loss": 0.0454,
      "step": 3510
    },
    {
      "epoch": 15.91,
      "learning_rate": 2.2624434389140273e-06,
      "loss": 0.0825,
      "step": 3520
    },
    {
      "epoch": 15.95,
      "learning_rate": 2.2373051784816493e-06,
      "loss": 0.0296,
      "step": 3530
    },
    {
      "epoch": 16.0,
      "learning_rate": 2.212166918049271e-06,
      "loss": 0.0308,
      "step": 3540
    },
    {
      "epoch": 16.0,
      "eval_accuracy": 0.9086294416243654,
      "eval_loss": 0.6942557692527771,
      "eval_runtime": 11.4217,
      "eval_samples_per_second": 17.248,
      "eval_steps_per_second": 8.668,
      "step": 3540
    },
    {
      "epoch": 16.05,
      "learning_rate": 2.187028657616893e-06,
      "loss": 0.0447,
      "step": 3550
    },
    {
      "epoch": 16.09,
      "learning_rate": 2.161890397184515e-06,
      "loss": 0.0946,
      "step": 3560
    },
    {
      "epoch": 16.14,
      "learning_rate": 2.136752136752137e-06,
      "loss": 0.1738,
      "step": 3570
    },
    {
      "epoch": 16.18,
      "learning_rate": 2.111613876319759e-06,
      "loss": 0.071,
      "step": 3580
    },
    {
      "epoch": 16.23,
      "learning_rate": 2.086475615887381e-06,
      "loss": 0.0176,
      "step": 3590
    },
    {
      "epoch": 16.27,
      "learning_rate": 2.0613373554550025e-06,
      "loss": 0.1395,
      "step": 3600
    },
    {
      "epoch": 16.32,
      "learning_rate": 2.0361990950226245e-06,
      "loss": 0.0002,
      "step": 3610
    },
    {
      "epoch": 16.36,
      "learning_rate": 2.0110608345902466e-06,
      "loss": 0.047,
      "step": 3620
    },
    {
      "epoch": 16.41,
      "learning_rate": 1.985922574157868e-06,
      "loss": 0.0049,
      "step": 3630
    },
    {
      "epoch": 16.45,
      "learning_rate": 1.96078431372549e-06,
      "loss": 0.05,
      "step": 3640
    },
    {
      "epoch": 16.5,
      "learning_rate": 1.935646053293112e-06,
      "loss": 0.2121,
      "step": 3650
    },
    {
      "epoch": 16.54,
      "learning_rate": 1.910507792860734e-06,
      "loss": 0.0364,
      "step": 3660
    },
    {
      "epoch": 16.59,
      "learning_rate": 1.885369532428356e-06,
      "loss": 0.0433,
      "step": 3670
    },
    {
      "epoch": 16.63,
      "learning_rate": 1.860231271995978e-06,
      "loss": 0.0016,
      "step": 3680
    },
    {
      "epoch": 16.68,
      "learning_rate": 1.8350930115635998e-06,
      "loss": 0.0008,
      "step": 3690
    },
    {
      "epoch": 16.72,
      "learning_rate": 1.8099547511312218e-06,
      "loss": 0.0001,
      "step": 3700
    },
    {
      "epoch": 16.77,
      "learning_rate": 1.7848164906988438e-06,
      "loss": 0.001,
      "step": 3710
    },
    {
      "epoch": 16.81,
      "learning_rate": 1.7596782302664658e-06,
      "loss": 0.0269,
      "step": 3720
    },
    {
      "epoch": 16.86,
      "learning_rate": 1.7345399698340876e-06,
      "loss": 0.001,
      "step": 3730
    },
    {
      "epoch": 16.9,
      "learning_rate": 1.7094017094017097e-06,
      "loss": 0.2083,
      "step": 3740
    },
    {
      "epoch": 16.95,
      "learning_rate": 1.6842634489693317e-06,
      "loss": 0.0007,
      "step": 3750
    },
    {
      "epoch": 16.99,
      "learning_rate": 1.6591251885369533e-06,
      "loss": 0.2561,
      "step": 3760
    },
    {
      "epoch": 17.0,
      "eval_accuracy": 0.9035532994923858,
      "eval_loss": 0.6907753944396973,
      "eval_runtime": 11.5041,
      "eval_samples_per_second": 17.124,
      "eval_steps_per_second": 8.606,
      "step": 3761
    },
    {
      "epoch": 17.04,
      "learning_rate": 1.6339869281045753e-06,
      "loss": 0.0011,
      "step": 3770
    },
    {
      "epoch": 17.08,
      "learning_rate": 1.6088486676721973e-06,
      "loss": 0.0129,
      "step": 3780
    },
    {
      "epoch": 17.13,
      "learning_rate": 1.583710407239819e-06,
      "loss": 0.0053,
      "step": 3790
    },
    {
      "epoch": 17.18,
      "learning_rate": 1.5585721468074411e-06,
      "loss": 0.0083,
      "step": 3800
    },
    {
      "epoch": 17.22,
      "learning_rate": 1.5334338863750631e-06,
      "loss": 0.0651,
      "step": 3810
    },
    {
      "epoch": 17.27,
      "learning_rate": 1.5082956259426847e-06,
      "loss": 0.0002,
      "step": 3820
    },
    {
      "epoch": 17.31,
      "learning_rate": 1.4831573655103067e-06,
      "loss": 0.0009,
      "step": 3830
    },
    {
      "epoch": 17.36,
      "learning_rate": 1.4580191050779287e-06,
      "loss": 0.0047,
      "step": 3840
    },
    {
      "epoch": 17.4,
      "learning_rate": 1.4328808446455505e-06,
      "loss": 0.1318,
      "step": 3850
    },
    {
      "epoch": 17.45,
      "learning_rate": 1.4077425842131726e-06,
      "loss": 0.0419,
      "step": 3860
    },
    {
      "epoch": 17.49,
      "learning_rate": 1.3826043237807946e-06,
      "loss": 0.0751,
      "step": 3870
    },
    {
      "epoch": 17.54,
      "learning_rate": 1.3574660633484164e-06,
      "loss": 0.0012,
      "step": 3880
    },
    {
      "epoch": 17.58,
      "learning_rate": 1.3323278029160384e-06,
      "loss": 0.0019,
      "step": 3890
    },
    {
      "epoch": 17.63,
      "learning_rate": 1.3071895424836604e-06,
      "loss": 0.0001,
      "step": 3900
    },
    {
      "epoch": 17.67,
      "learning_rate": 1.282051282051282e-06,
      "loss": 0.0061,
      "step": 3910
    },
    {
      "epoch": 17.72,
      "learning_rate": 1.256913021618904e-06,
      "loss": 0.001,
      "step": 3920
    },
    {
      "epoch": 17.76,
      "learning_rate": 1.231774761186526e-06,
      "loss": 0.0584,
      "step": 3930
    },
    {
      "epoch": 17.81,
      "learning_rate": 1.206636500754148e-06,
      "loss": 0.1281,
      "step": 3940
    },
    {
      "epoch": 17.85,
      "learning_rate": 1.1814982403217698e-06,
      "loss": 0.0692,
      "step": 3950
    },
    {
      "epoch": 17.9,
      "learning_rate": 1.1563599798893918e-06,
      "loss": 0.0185,
      "step": 3960
    },
    {
      "epoch": 17.94,
      "learning_rate": 1.1312217194570136e-06,
      "loss": 0.0065,
      "step": 3970
    },
    {
      "epoch": 17.99,
      "learning_rate": 1.1060834590246354e-06,
      "loss": 0.1023,
      "step": 3980
    },
    {
      "epoch": 18.0,
      "eval_accuracy": 0.9086294416243654,
      "eval_loss": 0.6389538645744324,
      "eval_runtime": 11.4495,
      "eval_samples_per_second": 17.206,
      "eval_steps_per_second": 8.647,
      "step": 3982
    },
    {
      "epoch": 18.03,
      "learning_rate": 1.0809451985922575e-06,
      "loss": 0.0001,
      "step": 3990
    },
    {
      "epoch": 18.08,
      "learning_rate": 1.0558069381598795e-06,
      "loss": 0.1817,
      "step": 4000
    },
    {
      "epoch": 18.12,
      "learning_rate": 1.0306686777275013e-06,
      "loss": 0.0287,
      "step": 4010
    },
    {
      "epoch": 18.17,
      "learning_rate": 1.0055304172951233e-06,
      "loss": 0.0985,
      "step": 4020
    },
    {
      "epoch": 18.21,
      "learning_rate": 9.80392156862745e-07,
      "loss": 0.0051,
      "step": 4030
    },
    {
      "epoch": 18.26,
      "learning_rate": 9.55253896430367e-07,
      "loss": 0.133,
      "step": 4040
    },
    {
      "epoch": 18.31,
      "learning_rate": 9.30115635997989e-07,
      "loss": 0.0007,
      "step": 4050
    },
    {
      "epoch": 18.35,
      "learning_rate": 9.049773755656109e-07,
      "loss": 0.1533,
      "step": 4060
    },
    {
      "epoch": 18.4,
      "learning_rate": 8.798391151332329e-07,
      "loss": 0.1159,
      "step": 4070
    },
    {
      "epoch": 18.44,
      "learning_rate": 8.547008547008548e-07,
      "loss": 0.0169,
      "step": 4080
    },
    {
      "epoch": 18.49,
      "learning_rate": 8.295625942684766e-07,
      "loss": 0.0002,
      "step": 4090
    },
    {
      "epoch": 18.53,
      "learning_rate": 8.044243338360986e-07,
      "loss": 0.0093,
      "step": 4100
    },
    {
      "epoch": 18.58,
      "learning_rate": 7.792860734037206e-07,
      "loss": 0.0536,
      "step": 4110
    },
    {
      "epoch": 18.62,
      "learning_rate": 7.541478129713424e-07,
      "loss": 0.3254,
      "step": 4120
    },
    {
      "epoch": 18.67,
      "learning_rate": 7.290095525389644e-07,
      "loss": 0.0229,
      "step": 4130
    },
    {
      "epoch": 18.71,
      "learning_rate": 7.038712921065863e-07,
      "loss": 0.1067,
      "step": 4140
    },
    {
      "epoch": 18.76,
      "learning_rate": 6.787330316742082e-07,
      "loss": 0.0969,
      "step": 4150
    },
    {
      "epoch": 18.8,
      "learning_rate": 6.535947712418302e-07,
      "loss": 0.0014,
      "step": 4160
    },
    {
      "epoch": 18.85,
      "learning_rate": 6.28456510809452e-07,
      "loss": 0.0007,
      "step": 4170
    },
    {
      "epoch": 18.89,
      "learning_rate": 6.03318250377074e-07,
      "loss": 0.059,
      "step": 4180
    },
    {
      "epoch": 18.94,
      "learning_rate": 5.781799899446959e-07,
      "loss": 0.0337,
      "step": 4190
    },
    {
      "epoch": 18.98,
      "learning_rate": 5.530417295123177e-07,
      "loss": 0.0688,
      "step": 4200
    },
    {
      "epoch": 19.0,
      "eval_accuracy": 0.9137055837563451,
      "eval_loss": 0.6214128136634827,
      "eval_runtime": 11.4137,
      "eval_samples_per_second": 17.26,
      "eval_steps_per_second": 8.674,
      "step": 4203
    },
    {
      "epoch": 19.03,
      "learning_rate": 5.279034690799397e-07,
      "loss": 0.0002,
      "step": 4210
    },
    {
      "epoch": 19.07,
      "learning_rate": 5.027652086475616e-07,
      "loss": 0.1321,
      "step": 4220
    },
    {
      "epoch": 19.12,
      "learning_rate": 4.776269482151835e-07,
      "loss": 0.1035,
      "step": 4230
    },
    {
      "epoch": 19.16,
      "learning_rate": 4.5248868778280546e-07,
      "loss": 0.0437,
      "step": 4240
    },
    {
      "epoch": 19.21,
      "learning_rate": 4.273504273504274e-07,
      "loss": 0.0034,
      "step": 4250
    },
    {
      "epoch": 19.25,
      "learning_rate": 4.022121669180493e-07,
      "loss": 0.0119,
      "step": 4260
    },
    {
      "epoch": 19.3,
      "learning_rate": 3.770739064856712e-07,
      "loss": 0.0119,
      "step": 4270
    },
    {
      "epoch": 19.34,
      "learning_rate": 3.5193564605329314e-07,
      "loss": 0.0521,
      "step": 4280
    },
    {
      "epoch": 19.39,
      "learning_rate": 3.267973856209151e-07,
      "loss": 0.0065,
      "step": 4290
    },
    {
      "epoch": 19.44,
      "learning_rate": 3.01659125188537e-07,
      "loss": 0.0738,
      "step": 4300
    },
    {
      "epoch": 19.48,
      "learning_rate": 2.7652086475615886e-07,
      "loss": 0.0108,
      "step": 4310
    },
    {
      "epoch": 19.53,
      "learning_rate": 2.513826043237808e-07,
      "loss": 0.1008,
      "step": 4320
    },
    {
      "epoch": 19.57,
      "learning_rate": 2.2624434389140273e-07,
      "loss": 0.004,
      "step": 4330
    },
    {
      "epoch": 19.62,
      "learning_rate": 2.0110608345902466e-07,
      "loss": 0.0277,
      "step": 4340
    },
    {
      "epoch": 19.66,
      "learning_rate": 1.7596782302664657e-07,
      "loss": 0.0003,
      "step": 4350
    },
    {
      "epoch": 19.71,
      "learning_rate": 1.508295625942685e-07,
      "loss": 0.0205,
      "step": 4360
    },
    {
      "epoch": 19.75,
      "learning_rate": 1.256913021618904e-07,
      "loss": 0.0225,
      "step": 4370
    },
    {
      "epoch": 19.8,
      "learning_rate": 1.0055304172951233e-07,
      "loss": 0.0101,
      "step": 4380
    },
    {
      "epoch": 19.84,
      "learning_rate": 7.541478129713425e-08,
      "loss": 0.1101,
      "step": 4390
    },
    {
      "epoch": 19.89,
      "learning_rate": 5.0276520864756165e-08,
      "loss": 0.1248,
      "step": 4400
    },
    {
      "epoch": 19.93,
      "learning_rate": 2.5138260432378083e-08,
      "loss": 0.005,
      "step": 4410
    },
    {
      "epoch": 19.98,
      "learning_rate": 0.0,
      "loss": 0.0542,
      "step": 4420
    },
    {
      "epoch": 19.98,
      "eval_accuracy": 0.9086294416243654,
      "eval_loss": 0.6275140643119812,
      "eval_runtime": 11.4471,
      "eval_samples_per_second": 17.21,
      "eval_steps_per_second": 8.648,
      "step": 4420
    },
    {
      "epoch": 19.98,
      "step": 4420,
      "total_flos": 1.8435127960865341e+19,
      "train_loss": 0.19852465775689138,
      "train_runtime": 6461.3145,
      "train_samples_per_second": 5.479,
      "train_steps_per_second": 0.684
    }
  ],
  "logging_steps": 10,
  "max_steps": 4420,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 20,
  "save_steps": 500,
  "total_flos": 1.8435127960865341e+19,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}