{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 20205,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 0.0,
      "loss": 0.0817,
      "step": 30
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0,
      "loss": 0.0261,
      "step": 60
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0,
      "loss": 0.0766,
      "step": 90
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0,
      "loss": 0.086,
      "step": 120
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0,
      "loss": 0.0375,
      "step": 150
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0,
      "loss": 0.0229,
      "step": 180
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0,
      "loss": 0.0396,
      "step": 210
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0,
      "loss": 0.056,
      "step": 240
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0,
      "loss": 0.0579,
      "step": 270
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0,
      "loss": 0.0549,
      "step": 300
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0,
      "loss": 0.0648,
      "step": 330
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0,
      "loss": 0.1489,
      "step": 360
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0,
      "loss": 0.0751,
      "step": 390
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0,
      "loss": 0.0712,
      "step": 420
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0,
      "loss": 0.0607,
      "step": 450
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0,
      "loss": 0.034,
      "step": 480
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0,
      "loss": 0.1474,
      "step": 510
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0,
      "loss": 0.0299,
      "step": 540
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0,
      "loss": 0.1422,
      "step": 570
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0,
      "loss": 0.0407,
      "step": 600
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0,
      "loss": 0.0623,
      "step": 630
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0,
      "loss": 0.0343,
      "step": 660
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0,
      "loss": 0.0622,
      "step": 690
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0,
      "loss": 0.1015,
      "step": 720
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0,
      "loss": 0.0592,
      "step": 750
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0,
      "loss": 0.0768,
      "step": 780
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0,
      "loss": 0.0447,
      "step": 810
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0,
      "loss": 0.1074,
      "step": 840
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0,
      "loss": 0.0646,
      "step": 870
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0,
      "loss": 0.1365,
      "step": 900
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0,
      "loss": 0.0675,
      "step": 930
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0,
      "loss": 0.0656,
      "step": 960
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0,
      "loss": 0.0482,
      "step": 990
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0,
      "loss": 0.1055,
      "step": 1020
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0,
      "loss": 0.0523,
      "step": 1050
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0,
      "loss": 0.029,
      "step": 1080
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0,
      "loss": 0.0353,
      "step": 1110
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0,
      "loss": 0.0729,
      "step": 1140
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0,
      "loss": 0.0777,
      "step": 1170
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0,
      "loss": 0.0219,
      "step": 1200
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0,
      "loss": 0.0666,
      "step": 1230
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0,
      "loss": 0.0667,
      "step": 1260
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0,
      "loss": 0.1055,
      "step": 1290
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0,
      "loss": 0.0116,
      "step": 1320
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0,
      "loss": 0.0653,
      "step": 1350
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0,
      "loss": 0.0591,
      "step": 1380
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0,
      "loss": 0.0873,
      "step": 1410
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0,
      "loss": 0.0955,
      "step": 1440
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0,
      "loss": 0.0401,
      "step": 1470
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0,
      "loss": 0.0654,
      "step": 1500
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0,
      "loss": 0.0282,
      "step": 1530
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0,
      "loss": 0.0714,
      "step": 1560
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0,
      "loss": 0.0601,
      "step": 1590
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0,
      "loss": 0.045,
      "step": 1620
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0,
      "loss": 0.0646,
      "step": 1650
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0,
      "loss": 0.0687,
      "step": 1680
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0,
      "loss": 0.0392,
      "step": 1710
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0,
      "loss": 0.0484,
      "step": 1740
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0,
      "loss": 0.1054,
      "step": 1770
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0,
      "loss": 0.0774,
      "step": 1800
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0,
      "loss": 0.0981,
      "step": 1830
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0,
      "loss": 0.1039,
      "step": 1860
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0,
      "loss": 0.1142,
      "step": 1890
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0,
      "loss": 0.0411,
      "step": 1920
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0,
      "loss": 0.1089,
      "step": 1950
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0,
      "loss": 0.1218,
      "step": 1980
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0,
      "loss": 0.0586,
      "step": 2010
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0,
      "loss": 0.1293,
      "step": 2040
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0,
      "loss": 0.099,
      "step": 2070
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0,
      "loss": 0.122,
      "step": 2100
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0,
      "loss": 0.0378,
      "step": 2130
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0,
      "loss": 0.0459,
      "step": 2160
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0,
      "loss": 0.0891,
      "step": 2190
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0,
      "loss": 0.1211,
      "step": 2220
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0,
      "loss": 0.0728,
      "step": 2250
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0,
      "loss": 0.137,
      "step": 2280
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0,
      "loss": 0.063,
      "step": 2310
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0,
      "loss": 0.0641,
      "step": 2340
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0,
      "loss": 0.063,
      "step": 2370
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0,
      "loss": 0.1139,
      "step": 2400
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0,
      "loss": 0.0635,
      "step": 2430
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0,
      "loss": 0.1686,
      "step": 2460
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0,
      "loss": 0.0634,
      "step": 2490
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0,
      "loss": 0.0768,
      "step": 2520
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0,
      "loss": 0.0431,
      "step": 2550
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0,
      "loss": 0.036,
      "step": 2580
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0,
      "loss": 0.0538,
      "step": 2610
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0,
      "loss": 0.0432,
      "step": 2640
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0,
      "loss": 0.0332,
      "step": 2670
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0,
      "loss": 0.0292,
      "step": 2700
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0,
      "loss": 0.0604,
      "step": 2730
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0,
      "loss": 0.1014,
      "step": 2760
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0,
      "loss": 0.0594,
      "step": 2790
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0,
      "loss": 0.0691,
      "step": 2820
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0,
      "loss": 0.0353,
      "step": 2850
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0,
      "loss": 0.0507,
      "step": 2880
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0,
      "loss": 0.0334,
      "step": 2910
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0,
      "loss": 0.0884,
      "step": 2940
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0,
      "loss": 0.1243,
      "step": 2970
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0,
      "loss": 0.0736,
      "step": 3000
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0,
      "loss": 0.1229,
      "step": 3030
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0,
      "loss": 0.0787,
      "step": 3060
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0,
      "loss": 0.0526,
      "step": 3090
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0,
      "loss": 0.0702,
      "step": 3120
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0,
      "loss": 0.1156,
      "step": 3150
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0,
      "loss": 0.0353,
      "step": 3180
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0,
      "loss": 0.0415,
      "step": 3210
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0,
      "loss": 0.0834,
      "step": 3240
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0,
      "loss": 0.0678,
      "step": 3270
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0,
      "loss": 0.0438,
      "step": 3300
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0,
      "loss": 0.0927,
      "step": 3330
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0,
      "loss": 0.0776,
      "step": 3360
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0,
      "loss": 0.0336,
      "step": 3390
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0,
      "loss": 0.0639,
      "step": 3420
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0,
      "loss": 0.0738,
      "step": 3450
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0,
      "loss": 0.079,
      "step": 3480
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0,
      "loss": 0.1838,
      "step": 3510
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0,
      "loss": 0.1083,
      "step": 3540
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0,
      "loss": 0.0642,
      "step": 3570
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0,
      "loss": 0.0728,
      "step": 3600
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0,
      "loss": 0.0342,
      "step": 3630
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0,
      "loss": 0.0867,
      "step": 3660
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0,
      "loss": 0.1483,
      "step": 3690
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0,
      "loss": 0.0346,
      "step": 3720
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0,
      "loss": 0.107,
      "step": 3750
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0,
      "loss": 0.0428,
      "step": 3780
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0,
      "loss": 0.0922,
      "step": 3810
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0,
      "loss": 0.0726,
      "step": 3840
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0,
      "loss": 0.0371,
      "step": 3870
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0,
      "loss": 0.1188,
      "step": 3900
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0,
      "loss": 0.1053,
      "step": 3930
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0,
      "loss": 0.0906,
      "step": 3960
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0,
      "loss": 0.1179,
      "step": 3990
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0,
      "loss": 0.0883,
      "step": 4020
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0,
      "loss": 0.0257,
      "step": 4050
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0,
      "loss": 0.0392,
      "step": 4080
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0,
      "loss": 0.0904,
      "step": 4110
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0,
      "loss": 0.0156,
      "step": 4140
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0,
      "loss": 0.1197,
      "step": 4170
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0,
      "loss": 0.0236,
      "step": 4200
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0,
      "loss": 0.0465,
      "step": 4230
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0,
      "loss": 0.0165,
      "step": 4260
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0,
      "loss": 0.0862,
      "step": 4290
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0,
      "loss": 0.0598,
      "step": 4320
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0,
      "loss": 0.0118,
      "step": 4350
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0,
      "loss": 0.0712,
      "step": 4380
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0,
      "loss": 0.0731,
      "step": 4410
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0,
      "loss": 0.0653,
      "step": 4440
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0,
      "loss": 0.1226,
      "step": 4470
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0,
      "loss": 0.0852,
      "step": 4500
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0,
      "loss": 0.1108,
      "step": 4530
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0,
      "loss": 0.0994,
      "step": 4560
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0,
      "loss": 0.0463,
      "step": 4590
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0,
      "loss": 0.0494,
      "step": 4620
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0,
      "loss": 0.0918,
      "step": 4650
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0,
      "loss": 0.0257,
      "step": 4680
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0,
      "loss": 0.0314,
      "step": 4710
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0,
      "loss": 0.0436,
      "step": 4740
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0,
      "loss": 0.0174,
      "step": 4770
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0,
      "loss": 0.0521,
      "step": 4800
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0,
      "loss": 0.1076,
      "step": 4830
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0,
      "loss": 0.0506,
      "step": 4860
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0,
      "loss": 0.0619,
      "step": 4890
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0,
      "loss": 0.0782,
      "step": 4920
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0,
      "loss": 0.051,
      "step": 4950
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0,
      "loss": 0.0376,
      "step": 4980
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0,
      "loss": 0.0611,
      "step": 5010
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0,
      "loss": 0.0864,
      "step": 5040
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0,
      "loss": 0.1291,
      "step": 5070
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.0,
      "loss": 0.0943,
      "step": 5100
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.0,
      "loss": 0.1185,
      "step": 5130
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0,
      "loss": 0.0729,
      "step": 5160
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0,
      "loss": 0.0837,
      "step": 5190
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0,
      "loss": 0.1156,
      "step": 5220
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0,
      "loss": 0.0881,
      "step": 5250
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0,
      "loss": 0.0483,
      "step": 5280
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0,
      "loss": 0.0205,
      "step": 5310
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0,
      "loss": 0.0741,
      "step": 5340
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0,
      "loss": 0.0604,
      "step": 5370
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0,
      "loss": 0.0603,
      "step": 5400
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0,
      "loss": 0.0519,
      "step": 5430
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0,
      "loss": 0.0919,
      "step": 5460
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0,
      "loss": 0.0546,
      "step": 5490
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0,
      "loss": 0.0426,
      "step": 5520
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0,
      "loss": 0.0767,
      "step": 5550
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0,
      "loss": 0.0756,
      "step": 5580
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0,
      "loss": 0.0686,
      "step": 5610
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.0,
      "loss": 0.1233,
      "step": 5640
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.0,
      "loss": 0.0656,
      "step": 5670
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0,
      "loss": 0.0585,
      "step": 5700
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0,
      "loss": 0.0535,
      "step": 5730
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0,
      "loss": 0.1607,
      "step": 5760
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0,
      "loss": 0.0991,
      "step": 5790
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0,
      "loss": 0.0671,
      "step": 5820
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0,
      "loss": 0.1062,
      "step": 5850
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0,
      "loss": 0.0766,
      "step": 5880
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0,
      "loss": 0.0583,
      "step": 5910
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0,
      "loss": 0.1114,
      "step": 5940
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0,
      "loss": 0.0918,
      "step": 5970
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0,
      "loss": 0.0539,
      "step": 6000
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.0,
      "loss": 0.135,
      "step": 6030
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.0,
      "loss": 0.0425,
      "step": 6060
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.0,
      "loss": 0.1668,
      "step": 6090
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0,
      "loss": 0.0206,
      "step": 6120
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0,
      "loss": 0.0452,
      "step": 6150
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0,
      "loss": 0.1687,
      "step": 6180
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0,
      "loss": 0.1512,
      "step": 6210
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0,
      "loss": 0.0162,
      "step": 6240
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0,
      "loss": 0.046,
      "step": 6270
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0,
      "loss": 0.1362,
      "step": 6300
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0,
      "loss": 0.0913,
      "step": 6330
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0,
      "loss": 0.1346,
      "step": 6360
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0,
      "loss": 0.04,
      "step": 6390
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0,
      "loss": 0.1049,
      "step": 6420
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0,
      "loss": 0.0443,
      "step": 6450
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0,
      "loss": 0.0923,
      "step": 6480
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0,
      "loss": 0.0483,
      "step": 6510
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0,
      "loss": 0.0668,
      "step": 6540
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0,
      "loss": 0.0686,
      "step": 6570
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0,
      "loss": 0.1093,
      "step": 6600
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0,
      "loss": 0.0132,
      "step": 6630
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0,
      "loss": 0.0547,
      "step": 6660
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0,
      "loss": 0.056,
      "step": 6690
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0,
      "loss": 0.0847,
      "step": 6720
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.9415137614678899,
      "eval_loss": 0.3246009051799774,
      "eval_runtime": 2.5067,
      "eval_samples_per_second": 347.87,
      "eval_steps_per_second": 35.106,
      "step": 6735
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0,
      "loss": 0.0416,
      "step": 6750
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0,
      "loss": 0.0397,
      "step": 6780
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0,
      "loss": 0.1238,
      "step": 6810
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.0,
      "loss": 0.0772,
      "step": 6840
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.0,
      "loss": 0.0706,
      "step": 6870
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.0,
      "loss": 0.1078,
      "step": 6900
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0,
      "loss": 0.1038,
      "step": 6930
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0,
      "loss": 0.0703,
      "step": 6960
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0,
      "loss": 0.0859,
      "step": 6990
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0,
      "loss": 0.105,
      "step": 7020
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0,
      "loss": 0.074,
      "step": 7050
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0,
      "loss": 0.0429,
      "step": 7080
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0,
      "loss": 0.032,
      "step": 7110
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0,
      "loss": 0.1072,
      "step": 7140
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0,
      "loss": 0.0322,
      "step": 7170
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0,
      "loss": 0.0857,
      "step": 7200
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0,
      "loss": 0.0906,
      "step": 7230
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.0,
      "loss": 0.0334,
      "step": 7260
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.0,
      "loss": 0.0366,
      "step": 7290
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0,
      "loss": 0.0602,
      "step": 7320
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0,
      "loss": 0.0545,
      "step": 7350
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0,
      "loss": 0.1206,
      "step": 7380
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0,
      "loss": 0.0517,
      "step": 7410
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0,
      "loss": 0.0758,
      "step": 7440
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.0,
      "loss": 0.0472,
      "step": 7470
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.0,
      "loss": 0.0856,
      "step": 7500
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0,
      "loss": 0.1196,
      "step": 7530
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0,
      "loss": 0.076,
      "step": 7560
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0,
      "loss": 0.0305,
      "step": 7590
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0,
      "loss": 0.0697,
      "step": 7620
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0,
      "loss": 0.0413,
      "step": 7650
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0,
      "loss": 0.0913,
      "step": 7680
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0,
      "loss": 0.0663,
      "step": 7710
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0,
      "loss": 0.0995,
      "step": 7740
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0,
      "loss": 0.0995,
      "step": 7770
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.0,
      "loss": 0.0292,
      "step": 7800
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.0,
      "loss": 0.0609,
      "step": 7830
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0,
      "loss": 0.0766,
      "step": 7860
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0,
      "loss": 0.021,
      "step": 7890
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0,
      "loss": 0.0567,
      "step": 7920
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0,
      "loss": 0.0636,
      "step": 7950
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0,
      "loss": 0.0213,
      "step": 7980
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0,
      "loss": 0.0308,
      "step": 8010
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0,
      "loss": 0.0877,
      "step": 8040
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0,
      "loss": 0.0501,
      "step": 8070
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0,
      "loss": 0.0387,
      "step": 8100
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0,
      "loss": 0.0369,
      "step": 8130
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0,
      "loss": 0.0791,
      "step": 8160
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0,
      "loss": 0.1032,
      "step": 8190
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0,
      "loss": 0.082,
      "step": 8220
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0,
      "loss": 0.0352,
      "step": 8250
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0,
      "loss": 0.0494,
      "step": 8280
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0,
      "loss": 0.0788,
      "step": 8310
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.0,
      "loss": 0.0247,
      "step": 8340
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.0,
      "loss": 0.1044,
      "step": 8370
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0,
      "loss": 0.1092,
      "step": 8400
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0,
      "loss": 0.0788,
      "step": 8430
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0,
      "loss": 0.099,
      "step": 8460
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0,
      "loss": 0.0419,
      "step": 8490
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0,
      "loss": 0.0443,
      "step": 8520
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0,
      "loss": 0.0651,
      "step": 8550
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0,
      "loss": 0.0834,
      "step": 8580
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.0,
      "loss": 0.0812,
      "step": 8610
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.0,
      "loss": 0.0982,
      "step": 8640
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.0,
      "loss": 0.0102,
      "step": 8670
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.0,
      "loss": 0.0878,
      "step": 8700
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.0,
      "loss": 0.1202,
      "step": 8730
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.0,
      "loss": 0.1057,
      "step": 8760
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0,
      "loss": 0.0425,
      "step": 8790
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0,
      "loss": 0.0443,
      "step": 8820
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0,
      "loss": 0.0672,
      "step": 8850
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.0,
      "loss": 0.0554,
      "step": 8880
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.0,
      "loss": 0.0715,
      "step": 8910
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.0,
      "loss": 0.0992,
      "step": 8940
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.0,
      "loss": 0.0257,
      "step": 8970
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.0,
      "loss": 0.0716,
      "step": 9000
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.0,
      "loss": 0.0471,
      "step": 9030
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0,
      "loss": 0.0293,
      "step": 9060
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0,
      "loss": 0.1055,
      "step": 9090
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0,
      "loss": 0.0806,
      "step": 9120
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.0,
      "loss": 0.0881,
      "step": 9150
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.0,
      "loss": 0.0776,
      "step": 9180
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.0,
      "loss": 0.0329,
      "step": 9210
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.0,
      "loss": 0.0593,
      "step": 9240
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0,
      "loss": 0.1015,
      "step": 9270
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0,
      "loss": 0.1131,
      "step": 9300
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0,
      "loss": 0.0407,
      "step": 9330
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0,
      "loss": 0.0163,
      "step": 9360
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0,
      "loss": 0.0478,
      "step": 9390
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0,
      "loss": 0.0946,
      "step": 9420
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0,
      "loss": 0.0746,
      "step": 9450
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.0,
      "loss": 0.125,
      "step": 9480
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.0,
      "loss": 0.0326,
      "step": 9510
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.0,
      "loss": 0.0344,
      "step": 9540
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.0,
      "loss": 0.0997,
      "step": 9570
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.0,
      "loss": 0.0817,
      "step": 9600
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.0,
      "loss": 0.0579,
      "step": 9630
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.0,
      "loss": 0.0774,
      "step": 9660
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0,
      "loss": 0.1389,
      "step": 9690
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0,
      "loss": 0.1399,
      "step": 9720
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0,
      "loss": 0.0589,
      "step": 9750
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0,
      "loss": 0.1118,
      "step": 9780
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.0,
      "loss": 0.1206,
      "step": 9810
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.0,
      "loss": 0.116,
      "step": 9840
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0,
      "loss": 0.0421,
      "step": 9870
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0,
      "loss": 0.0894,
      "step": 9900
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0,
      "loss": 0.0312,
      "step": 9930
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0,
      "loss": 0.131,
      "step": 9960
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0,
      "loss": 0.1178,
      "step": 9990
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.0,
      "loss": 0.0722,
      "step": 10020
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.0,
      "loss": 0.0534,
      "step": 10050
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.0,
      "loss": 0.1002,
      "step": 10080
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.0,
      "loss": 0.0845,
      "step": 10110
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.0,
      "loss": 0.036,
      "step": 10140
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.0,
      "loss": 0.0868,
      "step": 10170
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.0,
      "loss": 0.0301,
      "step": 10200
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.0,
      "loss": 0.0892,
      "step": 10230
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.0,
      "loss": 0.0238,
      "step": 10260
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.0,
      "loss": 0.0668,
      "step": 10290
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.0,
      "loss": 0.0685,
      "step": 10320
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0,
      "loss": 0.105,
      "step": 10350
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0,
      "loss": 0.0894,
      "step": 10380
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.0,
      "loss": 0.0677,
      "step": 10410
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.0,
      "loss": 0.1072,
      "step": 10440
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.0,
      "loss": 0.1049,
      "step": 10470
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.0,
      "loss": 0.1096,
      "step": 10500
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.0,
      "loss": 0.0354,
      "step": 10530
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.0,
      "loss": 0.0925,
      "step": 10560
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.0,
      "loss": 0.1121,
      "step": 10590
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.0,
      "loss": 0.0955,
      "step": 10620
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.0,
      "loss": 0.0523,
      "step": 10650
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.0,
      "loss": 0.0239,
      "step": 10680
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.0,
      "loss": 0.063,
      "step": 10710
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.0,
      "loss": 0.0928,
      "step": 10740
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.0,
      "loss": 0.0736,
      "step": 10770
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.0,
      "loss": 0.2001,
      "step": 10800
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.0,
      "loss": 0.0887,
      "step": 10830
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.0,
      "loss": 0.0584,
      "step": 10860
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.0,
      "loss": 0.0825,
      "step": 10890
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.0,
      "loss": 0.1035,
      "step": 10920
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.0,
      "loss": 0.0991,
      "step": 10950
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.0,
      "loss": 0.0133,
      "step": 10980
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.0,
      "loss": 0.0265,
      "step": 11010
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.0,
      "loss": 0.1334,
      "step": 11040
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.0,
      "loss": 0.0862,
      "step": 11070
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.0,
      "loss": 0.0174,
      "step": 11100
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.0,
      "loss": 0.1254,
      "step": 11130
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.0,
      "loss": 0.1378,
      "step": 11160
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.0,
      "loss": 0.139,
      "step": 11190
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.0,
      "loss": 0.0925,
      "step": 11220
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.0,
      "loss": 0.0658,
      "step": 11250
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.0,
      "loss": 0.0758,
      "step": 11280
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.0,
      "loss": 0.0505,
      "step": 11310
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.0,
      "loss": 0.0726,
      "step": 11340
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.0,
      "loss": 0.0742,
      "step": 11370
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.0,
      "loss": 0.0372,
      "step": 11400
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.0,
      "loss": 0.0433,
      "step": 11430
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.0,
      "loss": 0.1069,
      "step": 11460
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.0,
      "loss": 0.0277,
      "step": 11490
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.0,
      "loss": 0.1009,
      "step": 11520
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.0,
      "loss": 0.0821,
      "step": 11550
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.0,
      "loss": 0.1083,
      "step": 11580
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.0,
      "loss": 0.1303,
      "step": 11610
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.0,
      "loss": 0.0565,
      "step": 11640
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.0,
      "loss": 0.0538,
      "step": 11670
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.0,
      "loss": 0.0528,
      "step": 11700
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.0,
      "loss": 0.0487,
      "step": 11730
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.0,
      "loss": 0.0321,
      "step": 11760
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.0,
      "loss": 0.0619,
      "step": 11790
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.0,
      "loss": 0.0711,
      "step": 11820
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.0,
      "loss": 0.0628,
      "step": 11850
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.0,
      "loss": 0.1428,
      "step": 11880
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.0,
      "loss": 0.0667,
      "step": 11910
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.0,
      "loss": 0.0933,
      "step": 11940
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.0,
      "loss": 0.056,
      "step": 11970
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.0,
      "loss": 0.0856,
      "step": 12000
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.0,
      "loss": 0.0835,
      "step": 12030
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.0,
      "loss": 0.0477,
      "step": 12060
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.0,
      "loss": 0.0868,
      "step": 12090
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.0,
      "loss": 0.0758,
      "step": 12120
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.0,
      "loss": 0.0988,
      "step": 12150
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.0,
      "loss": 0.1119,
      "step": 12180
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.0,
      "loss": 0.0566,
      "step": 12210
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.0,
      "loss": 0.0748,
      "step": 12240
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.0,
      "loss": 0.0542,
      "step": 12270
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.0,
      "loss": 0.0415,
      "step": 12300
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.0,
      "loss": 0.0656,
      "step": 12330
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.0,
      "loss": 0.1149,
      "step": 12360
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.0,
      "loss": 0.0645,
      "step": 12390
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.0,
      "loss": 0.0256,
      "step": 12420
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.0,
      "loss": 0.1142,
      "step": 12450
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.0,
      "loss": 0.0809,
      "step": 12480
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.0,
      "loss": 0.0714,
      "step": 12510
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.0,
      "loss": 0.0981,
      "step": 12540
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.0,
      "loss": 0.0218,
      "step": 12570
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.0,
      "loss": 0.0723,
      "step": 12600
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.0,
      "loss": 0.0674,
      "step": 12630
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.0,
      "loss": 0.1054,
      "step": 12660
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.0,
      "loss": 0.0889,
      "step": 12690
    },
    {
      "epoch": 1.89,
      "learning_rate": 0.0,
      "loss": 0.1059,
      "step": 12720
    },
    {
      "epoch": 1.89,
      "learning_rate": 0.0,
      "loss": 0.0404,
      "step": 12750
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.0,
      "loss": 0.1321,
      "step": 12780
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.0,
      "loss": 0.1299,
      "step": 12810
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.0,
      "loss": 0.0885,
      "step": 12840
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.0,
      "loss": 0.1057,
      "step": 12870
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.0,
      "loss": 0.0654,
      "step": 12900
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.0,
      "loss": 0.0655,
      "step": 12930
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.0,
      "loss": 0.0922,
      "step": 12960
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.0,
      "loss": 0.1027,
      "step": 12990
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.0,
      "loss": 0.0931,
      "step": 13020
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.0,
      "loss": 0.0727,
      "step": 13050
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.0,
      "loss": 0.1069,
      "step": 13080
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.0,
      "loss": 0.056,
      "step": 13110
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.0,
      "loss": 0.0419,
      "step": 13140
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.0,
      "loss": 0.0814,
      "step": 13170
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.0,
      "loss": 0.1313,
      "step": 13200
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.0,
      "loss": 0.0811,
      "step": 13230
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.0,
      "loss": 0.061,
      "step": 13260
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.0,
      "loss": 0.0655,
      "step": 13290
    },
    {
      "epoch": 1.98,
      "learning_rate": 0.0,
      "loss": 0.0502,
      "step": 13320
    },
    {
      "epoch": 1.98,
      "learning_rate": 0.0,
      "loss": 0.0462,
      "step": 13350
    },
    {
      "epoch": 1.99,
      "learning_rate": 0.0,
      "loss": 0.0606,
      "step": 13380
    },
    {
      "epoch": 1.99,
      "learning_rate": 0.0,
      "loss": 0.1204,
      "step": 13410
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.0,
      "loss": 0.0297,
      "step": 13440
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.0,
      "loss": 0.028,
      "step": 13470
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9415137614678899,
      "eval_loss": 0.3246009051799774,
      "eval_runtime": 4.8202,
      "eval_samples_per_second": 180.906,
      "eval_steps_per_second": 18.257,
      "step": 13470
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.0,
      "loss": 0.069,
      "step": 13500
    },
    {
      "epoch": 2.01,
      "learning_rate": 0.0,
      "loss": 0.0574,
      "step": 13530
    },
    {
      "epoch": 2.01,
      "learning_rate": 0.0,
      "loss": 0.0796,
      "step": 13560
    },
    {
      "epoch": 2.02,
      "learning_rate": 0.0,
      "loss": 0.0571,
      "step": 13590
    },
    {
      "epoch": 2.02,
      "learning_rate": 0.0,
      "loss": 0.0195,
      "step": 13620
    },
    {
      "epoch": 2.03,
      "learning_rate": 0.0,
      "loss": 0.1019,
      "step": 13650
    },
    {
      "epoch": 2.03,
      "learning_rate": 0.0,
      "loss": 0.0316,
      "step": 13680
    },
    {
      "epoch": 2.04,
      "learning_rate": 0.0,
      "loss": 0.0702,
      "step": 13710
    },
    {
      "epoch": 2.04,
      "learning_rate": 0.0,
      "loss": 0.1436,
      "step": 13740
    },
    {
      "epoch": 2.04,
      "learning_rate": 0.0,
      "loss": 0.0393,
      "step": 13770
    },
    {
      "epoch": 2.05,
      "learning_rate": 0.0,
      "loss": 0.09,
      "step": 13800
    },
    {
      "epoch": 2.05,
      "learning_rate": 0.0,
      "loss": 0.09,
      "step": 13830
    },
    {
      "epoch": 2.06,
      "learning_rate": 0.0,
      "loss": 0.1039,
      "step": 13860
    },
    {
      "epoch": 2.06,
      "learning_rate": 0.0,
      "loss": 0.1176,
      "step": 13890
    },
    {
      "epoch": 2.07,
      "learning_rate": 0.0,
      "loss": 0.0881,
      "step": 13920
    },
    {
      "epoch": 2.07,
      "learning_rate": 0.0,
      "loss": 0.1156,
      "step": 13950
    },
    {
      "epoch": 2.08,
      "learning_rate": 0.0,
      "loss": 0.0962,
      "step": 13980
    },
    {
      "epoch": 2.08,
      "learning_rate": 0.0,
      "loss": 0.093,
      "step": 14010
    },
    {
      "epoch": 2.08,
      "learning_rate": 0.0,
      "loss": 0.0674,
      "step": 14040
    },
    {
      "epoch": 2.09,
      "learning_rate": 0.0,
      "loss": 0.0798,
      "step": 14070
    },
    {
      "epoch": 2.09,
      "learning_rate": 0.0,
      "loss": 0.1274,
      "step": 14100
    },
    {
      "epoch": 2.1,
      "learning_rate": 0.0,
      "loss": 0.1226,
      "step": 14130
    },
    {
      "epoch": 2.1,
      "learning_rate": 0.0,
      "loss": 0.0788,
      "step": 14160
    },
    {
      "epoch": 2.11,
      "learning_rate": 0.0,
      "loss": 0.0314,
      "step": 14190
    },
    {
      "epoch": 2.11,
      "learning_rate": 0.0,
      "loss": 0.0951,
      "step": 14220
    },
    {
      "epoch": 2.12,
      "learning_rate": 0.0,
      "loss": 0.092,
      "step": 14250
    },
    {
      "epoch": 2.12,
      "learning_rate": 0.0,
      "loss": 0.1238,
      "step": 14280
    },
    {
      "epoch": 2.12,
      "learning_rate": 0.0,
      "loss": 0.06,
      "step": 14310
    },
    {
      "epoch": 2.13,
      "learning_rate": 0.0,
      "loss": 0.0595,
      "step": 14340
    },
    {
      "epoch": 2.13,
      "learning_rate": 0.0,
      "loss": 0.051,
      "step": 14370
    },
    {
      "epoch": 2.14,
      "learning_rate": 0.0,
      "loss": 0.0562,
      "step": 14400
    },
    {
      "epoch": 2.14,
      "learning_rate": 0.0,
      "loss": 0.0948,
      "step": 14430
    },
    {
      "epoch": 2.15,
      "learning_rate": 0.0,
      "loss": 0.0457,
      "step": 14460
    },
    {
      "epoch": 2.15,
      "learning_rate": 0.0,
      "loss": 0.0594,
      "step": 14490
    },
    {
      "epoch": 2.16,
      "learning_rate": 0.0,
      "loss": 0.0879,
      "step": 14520
    },
    {
      "epoch": 2.16,
      "learning_rate": 0.0,
      "loss": 0.0445,
      "step": 14550
    },
    {
      "epoch": 2.16,
      "learning_rate": 0.0,
      "loss": 0.0621,
      "step": 14580
    },
    {
      "epoch": 2.17,
      "learning_rate": 0.0,
      "loss": 0.0631,
      "step": 14610
    },
    {
      "epoch": 2.17,
      "learning_rate": 0.0,
      "loss": 0.0739,
      "step": 14640
    },
    {
      "epoch": 2.18,
      "learning_rate": 0.0,
      "loss": 0.0539,
      "step": 14670
    },
    {
      "epoch": 2.18,
      "learning_rate": 0.0,
      "loss": 0.0485,
      "step": 14700
    },
    {
      "epoch": 2.19,
      "learning_rate": 0.0,
      "loss": 0.0397,
      "step": 14730
    },
    {
      "epoch": 2.19,
      "learning_rate": 0.0,
      "loss": 0.0976,
      "step": 14760
    },
    {
      "epoch": 2.2,
      "learning_rate": 0.0,
      "loss": 0.0599,
      "step": 14790
    },
    {
      "epoch": 2.2,
      "learning_rate": 0.0,
      "loss": 0.0784,
      "step": 14820
    },
    {
      "epoch": 2.2,
      "learning_rate": 0.0,
      "loss": 0.045,
      "step": 14850
    },
    {
      "epoch": 2.21,
      "learning_rate": 0.0,
      "loss": 0.0451,
      "step": 14880
    },
    {
      "epoch": 2.21,
      "learning_rate": 0.0,
      "loss": 0.0915,
      "step": 14910
    },
    {
      "epoch": 2.22,
      "learning_rate": 0.0,
      "loss": 0.1122,
      "step": 14940
    },
    {
      "epoch": 2.22,
      "learning_rate": 0.0,
      "loss": 0.1391,
      "step": 14970
    },
    {
      "epoch": 2.23,
      "learning_rate": 0.0,
      "loss": 0.0815,
      "step": 15000
    },
    {
      "epoch": 2.23,
      "learning_rate": 0.0,
      "loss": 0.0131,
      "step": 15030
    },
    {
      "epoch": 2.24,
      "learning_rate": 0.0,
      "loss": 0.1057,
      "step": 15060
    },
    {
      "epoch": 2.24,
      "learning_rate": 0.0,
      "loss": 0.0904,
      "step": 15090
    },
    {
      "epoch": 2.24,
      "learning_rate": 0.0,
      "loss": 0.0438,
      "step": 15120
    },
    {
      "epoch": 2.25,
      "learning_rate": 0.0,
      "loss": 0.068,
      "step": 15150
    },
    {
      "epoch": 2.25,
      "learning_rate": 0.0,
      "loss": 0.0743,
      "step": 15180
    },
    {
      "epoch": 2.26,
      "learning_rate": 0.0,
      "loss": 0.1089,
      "step": 15210
    },
    {
      "epoch": 2.26,
      "learning_rate": 0.0,
      "loss": 0.0952,
      "step": 15240
    },
    {
      "epoch": 2.27,
      "learning_rate": 0.0,
      "loss": 0.128,
      "step": 15270
    },
    {
      "epoch": 2.27,
      "learning_rate": 0.0,
      "loss": 0.0373,
      "step": 15300
    },
    {
      "epoch": 2.28,
      "learning_rate": 0.0,
      "loss": 0.0527,
      "step": 15330
    },
    {
      "epoch": 2.28,
      "learning_rate": 0.0,
      "loss": 0.1093,
      "step": 15360
    },
    {
      "epoch": 2.29,
      "learning_rate": 0.0,
      "loss": 0.1115,
      "step": 15390
    },
    {
      "epoch": 2.29,
      "learning_rate": 0.0,
      "loss": 0.0385,
      "step": 15420
    },
    {
      "epoch": 2.29,
      "learning_rate": 0.0,
      "loss": 0.0309,
      "step": 15450
    },
    {
      "epoch": 2.3,
      "learning_rate": 0.0,
      "loss": 0.0509,
      "step": 15480
    },
    {
      "epoch": 2.3,
      "learning_rate": 0.0,
      "loss": 0.0783,
      "step": 15510
    },
    {
      "epoch": 2.31,
      "learning_rate": 0.0,
      "loss": 0.0658,
      "step": 15540
    },
    {
      "epoch": 2.31,
      "learning_rate": 0.0,
      "loss": 0.1936,
      "step": 15570
    },
    {
      "epoch": 2.32,
      "learning_rate": 0.0,
      "loss": 0.1358,
      "step": 15600
    },
    {
      "epoch": 2.32,
      "learning_rate": 0.0,
      "loss": 0.0406,
      "step": 15630
    },
    {
      "epoch": 2.33,
      "learning_rate": 0.0,
      "loss": 0.0971,
      "step": 15660
    },
    {
      "epoch": 2.33,
      "learning_rate": 0.0,
      "loss": 0.0731,
      "step": 15690
    },
    {
      "epoch": 2.33,
      "learning_rate": 0.0,
      "loss": 0.1039,
      "step": 15720
    },
    {
      "epoch": 2.34,
      "learning_rate": 0.0,
      "loss": 0.0713,
      "step": 15750
    },
    {
      "epoch": 2.34,
      "learning_rate": 0.0,
      "loss": 0.0398,
      "step": 15780
    },
    {
      "epoch": 2.35,
      "learning_rate": 0.0,
      "loss": 0.1228,
      "step": 15810
    },
    {
      "epoch": 2.35,
      "learning_rate": 0.0,
      "loss": 0.0909,
      "step": 15840
    },
    {
      "epoch": 2.36,
      "learning_rate": 0.0,
      "loss": 0.1098,
      "step": 15870
    },
    {
      "epoch": 2.36,
      "learning_rate": 0.0,
      "loss": 0.0583,
      "step": 15900
    },
    {
      "epoch": 2.37,
      "learning_rate": 0.0,
      "loss": 0.0608,
      "step": 15930
    },
    {
      "epoch": 2.37,
      "learning_rate": 0.0,
      "loss": 0.0909,
      "step": 15960
    },
    {
      "epoch": 2.37,
      "learning_rate": 0.0,
      "loss": 0.0613,
      "step": 15990
    },
    {
      "epoch": 2.38,
      "learning_rate": 0.0,
      "loss": 0.085,
      "step": 16020
    },
    {
      "epoch": 2.38,
      "learning_rate": 0.0,
      "loss": 0.0932,
      "step": 16050
    },
    {
      "epoch": 2.39,
      "learning_rate": 0.0,
      "loss": 0.1021,
      "step": 16080
    },
    {
      "epoch": 2.39,
      "learning_rate": 0.0,
      "loss": 0.0812,
      "step": 16110
    },
    {
      "epoch": 2.4,
      "learning_rate": 0.0,
      "loss": 0.1108,
      "step": 16140
    },
    {
      "epoch": 2.4,
      "learning_rate": 0.0,
      "loss": 0.0282,
      "step": 16170
    },
    {
      "epoch": 2.41,
      "learning_rate": 0.0,
      "loss": 0.0333,
      "step": 16200
    },
    {
      "epoch": 2.41,
      "learning_rate": 0.0,
      "loss": 0.0745,
      "step": 16230
    },
    {
      "epoch": 2.41,
      "learning_rate": 0.0,
      "loss": 0.0499,
      "step": 16260
    },
    {
      "epoch": 2.42,
      "learning_rate": 0.0,
      "loss": 0.1022,
      "step": 16290
    },
    {
      "epoch": 2.42,
      "learning_rate": 0.0,
      "loss": 0.1137,
      "step": 16320
    },
    {
      "epoch": 2.43,
      "learning_rate": 0.0,
      "loss": 0.0414,
      "step": 16350
    },
    {
      "epoch": 2.43,
      "learning_rate": 0.0,
      "loss": 0.0463,
      "step": 16380
    },
    {
      "epoch": 2.44,
      "learning_rate": 0.0,
      "loss": 0.0981,
      "step": 16410
    },
    {
      "epoch": 2.44,
      "learning_rate": 0.0,
      "loss": 0.0465,
      "step": 16440
    },
    {
      "epoch": 2.45,
      "learning_rate": 0.0,
      "loss": 0.0795,
      "step": 16470
    },
    {
      "epoch": 2.45,
      "learning_rate": 0.0,
      "loss": 0.0494,
      "step": 16500
    },
    {
      "epoch": 2.45,
      "learning_rate": 0.0,
      "loss": 0.0686,
      "step": 16530
    },
    {
      "epoch": 2.46,
      "learning_rate": 0.0,
      "loss": 0.0589,
      "step": 16560
    },
    {
      "epoch": 2.46,
      "learning_rate": 0.0,
      "loss": 0.0487,
      "step": 16590
    },
    {
      "epoch": 2.47,
      "learning_rate": 0.0,
      "loss": 0.0868,
      "step": 16620
    },
    {
      "epoch": 2.47,
      "learning_rate": 0.0,
      "loss": 0.0762,
      "step": 16650
    },
    {
      "epoch": 2.48,
      "learning_rate": 0.0,
      "loss": 0.0804,
      "step": 16680
    },
    {
      "epoch": 2.48,
      "learning_rate": 0.0,
      "loss": 0.0654,
      "step": 16710
    },
    {
      "epoch": 2.49,
      "learning_rate": 0.0,
      "loss": 0.1255,
      "step": 16740
    },
    {
      "epoch": 2.49,
      "learning_rate": 0.0,
      "loss": 0.0961,
      "step": 16770
    },
    {
      "epoch": 2.49,
      "learning_rate": 0.0,
      "loss": 0.1008,
      "step": 16800
    },
    {
      "epoch": 2.5,
      "learning_rate": 0.0,
      "loss": 0.0761,
      "step": 16830
    },
    {
      "epoch": 2.5,
      "learning_rate": 0.0,
      "loss": 0.0204,
      "step": 16860
    },
    {
      "epoch": 2.51,
      "learning_rate": 0.0,
      "loss": 0.0559,
      "step": 16890
    },
    {
      "epoch": 2.51,
      "learning_rate": 0.0,
      "loss": 0.0974,
      "step": 16920
    },
    {
      "epoch": 2.52,
      "learning_rate": 0.0,
      "loss": 0.0122,
      "step": 16950
    },
    {
      "epoch": 2.52,
      "learning_rate": 0.0,
      "loss": 0.097,
      "step": 16980
    },
    {
      "epoch": 2.53,
      "learning_rate": 0.0,
      "loss": 0.1055,
      "step": 17010
    },
    {
      "epoch": 2.53,
      "learning_rate": 0.0,
      "loss": 0.0582,
      "step": 17040
    },
    {
      "epoch": 2.53,
      "learning_rate": 0.0,
      "loss": 0.0884,
      "step": 17070
    },
    {
      "epoch": 2.54,
      "learning_rate": 0.0,
      "loss": 0.0634,
      "step": 17100
    },
    {
      "epoch": 2.54,
      "learning_rate": 0.0,
      "loss": 0.0527,
      "step": 17130
    },
    {
      "epoch": 2.55,
      "learning_rate": 0.0,
      "loss": 0.0653,
      "step": 17160
    },
    {
      "epoch": 2.55,
      "learning_rate": 0.0,
      "loss": 0.0989,
      "step": 17190
    },
    {
      "epoch": 2.56,
      "learning_rate": 0.0,
      "loss": 0.1646,
      "step": 17220
    },
    {
      "epoch": 2.56,
      "learning_rate": 0.0,
      "loss": 0.1076,
      "step": 17250
    },
    {
      "epoch": 2.57,
      "learning_rate": 0.0,
      "loss": 0.0211,
      "step": 17280
    },
    {
      "epoch": 2.57,
      "learning_rate": 0.0,
      "loss": 0.0626,
      "step": 17310
    },
    {
      "epoch": 2.57,
      "learning_rate": 0.0,
      "loss": 0.1055,
      "step": 17340
    },
    {
      "epoch": 2.58,
      "learning_rate": 0.0,
      "loss": 0.0293,
      "step": 17370
    },
    {
      "epoch": 2.58,
      "learning_rate": 0.0,
      "loss": 0.1164,
      "step": 17400
    },
    {
      "epoch": 2.59,
      "learning_rate": 0.0,
      "loss": 0.0381,
      "step": 17430
    },
    {
      "epoch": 2.59,
      "learning_rate": 0.0,
      "loss": 0.049,
      "step": 17460
    },
    {
      "epoch": 2.6,
      "learning_rate": 0.0,
      "loss": 0.0639,
      "step": 17490
    },
    {
      "epoch": 2.6,
      "learning_rate": 0.0,
      "loss": 0.039,
      "step": 17520
    },
    {
      "epoch": 2.61,
      "learning_rate": 0.0,
      "loss": 0.0959,
      "step": 17550
    },
    {
      "epoch": 2.61,
      "learning_rate": 0.0,
      "loss": 0.0909,
      "step": 17580
    },
    {
      "epoch": 2.61,
      "learning_rate": 0.0,
      "loss": 0.1018,
      "step": 17610
    },
    {
      "epoch": 2.62,
      "learning_rate": 0.0,
      "loss": 0.0695,
      "step": 17640
    },
    {
      "epoch": 2.62,
      "learning_rate": 0.0,
      "loss": 0.1116,
      "step": 17670
    },
    {
      "epoch": 2.63,
      "learning_rate": 0.0,
      "loss": 0.0418,
      "step": 17700
    },
    {
      "epoch": 2.63,
      "learning_rate": 0.0,
      "loss": 0.1098,
      "step": 17730
    },
    {
      "epoch": 2.64,
      "learning_rate": 0.0,
      "loss": 0.1354,
      "step": 17760
    },
    {
      "epoch": 2.64,
      "learning_rate": 0.0,
      "loss": 0.0651,
      "step": 17790
    },
    {
      "epoch": 2.65,
      "learning_rate": 0.0,
      "loss": 0.1013,
      "step": 17820
    },
    {
      "epoch": 2.65,
      "learning_rate": 0.0,
      "loss": 0.0742,
      "step": 17850
    },
    {
      "epoch": 2.65,
      "learning_rate": 0.0,
      "loss": 0.0631,
      "step": 17880
    },
    {
      "epoch": 2.66,
      "learning_rate": 0.0,
      "loss": 0.1206,
      "step": 17910
    },
    {
      "epoch": 2.66,
      "learning_rate": 0.0,
      "loss": 0.081,
      "step": 17940
    },
    {
      "epoch": 2.67,
      "learning_rate": 0.0,
      "loss": 0.0913,
      "step": 17970
    },
    {
      "epoch": 2.67,
      "learning_rate": 0.0,
      "loss": 0.0775,
      "step": 18000
    },
    {
      "epoch": 2.68,
      "learning_rate": 0.0,
      "loss": 0.0544,
      "step": 18030
    },
    {
      "epoch": 2.68,
      "learning_rate": 0.0,
      "loss": 0.0367,
      "step": 18060
    },
    {
      "epoch": 2.69,
      "learning_rate": 0.0,
      "loss": 0.0315,
      "step": 18090
    },
    {
      "epoch": 2.69,
      "learning_rate": 0.0,
      "loss": 0.0742,
      "step": 18120
    },
    {
      "epoch": 2.69,
      "learning_rate": 0.0,
      "loss": 0.0642,
      "step": 18150
    },
    {
      "epoch": 2.7,
      "learning_rate": 0.0,
      "loss": 0.0434,
      "step": 18180
    },
    {
      "epoch": 2.7,
      "learning_rate": 0.0,
      "loss": 0.0311,
      "step": 18210
    },
    {
      "epoch": 2.71,
      "learning_rate": 0.0,
      "loss": 0.112,
      "step": 18240
    },
    {
      "epoch": 2.71,
      "learning_rate": 0.0,
      "loss": 0.061,
      "step": 18270
    },
    {
      "epoch": 2.72,
      "learning_rate": 0.0,
      "loss": 0.0375,
      "step": 18300
    },
    {
      "epoch": 2.72,
      "learning_rate": 0.0,
      "loss": 0.0581,
      "step": 18330
    },
    {
      "epoch": 2.73,
      "learning_rate": 0.0,
      "loss": 0.0953,
      "step": 18360
    },
    {
      "epoch": 2.73,
      "learning_rate": 0.0,
      "loss": 0.0537,
      "step": 18390
    },
    {
      "epoch": 2.73,
      "learning_rate": 0.0,
      "loss": 0.0804,
      "step": 18420
    },
    {
      "epoch": 2.74,
      "learning_rate": 0.0,
      "loss": 0.0292,
      "step": 18450
    },
    {
      "epoch": 2.74,
      "learning_rate": 0.0,
      "loss": 0.048,
      "step": 18480
    },
    {
      "epoch": 2.75,
      "learning_rate": 0.0,
      "loss": 0.0932,
      "step": 18510
    },
    {
      "epoch": 2.75,
      "learning_rate": 0.0,
      "loss": 0.0801,
      "step": 18540
    },
    {
      "epoch": 2.76,
      "learning_rate": 0.0,
      "loss": 0.0621,
      "step": 18570
    },
    {
      "epoch": 2.76,
      "learning_rate": 0.0,
      "loss": 0.1034,
      "step": 18600
    },
    {
      "epoch": 2.77,
      "learning_rate": 0.0,
      "loss": 0.1303,
      "step": 18630
    },
    {
      "epoch": 2.77,
      "learning_rate": 0.0,
      "loss": 0.049,
      "step": 18660
    },
    {
      "epoch": 2.78,
      "learning_rate": 0.0,
      "loss": 0.0765,
      "step": 18690
    },
    {
      "epoch": 2.78,
      "learning_rate": 0.0,
      "loss": 0.0733,
      "step": 18720
    },
    {
      "epoch": 2.78,
      "learning_rate": 0.0,
      "loss": 0.035,
      "step": 18750
    },
    {
      "epoch": 2.79,
      "learning_rate": 0.0,
      "loss": 0.0691,
      "step": 18780
    },
    {
      "epoch": 2.79,
      "learning_rate": 0.0,
      "loss": 0.0583,
      "step": 18810
    },
    {
      "epoch": 2.8,
      "learning_rate": 0.0,
      "loss": 0.1347,
      "step": 18840
    },
    {
      "epoch": 2.8,
      "learning_rate": 0.0,
      "loss": 0.0502,
      "step": 18870
    },
    {
      "epoch": 2.81,
      "learning_rate": 0.0,
      "loss": 0.0376,
      "step": 18900
    },
    {
      "epoch": 2.81,
      "learning_rate": 0.0,
      "loss": 0.0509,
      "step": 18930
    },
    {
      "epoch": 2.82,
      "learning_rate": 0.0,
      "loss": 0.0804,
      "step": 18960
    },
    {
      "epoch": 2.82,
      "learning_rate": 0.0,
      "loss": 0.0763,
      "step": 18990
    },
    {
      "epoch": 2.82,
      "learning_rate": 0.0,
      "loss": 0.0699,
      "step": 19020
    },
    {
      "epoch": 2.83,
      "learning_rate": 0.0,
      "loss": 0.1181,
      "step": 19050
    },
    {
      "epoch": 2.83,
      "learning_rate": 0.0,
      "loss": 0.0505,
      "step": 19080
    },
    {
      "epoch": 2.84,
      "learning_rate": 0.0,
      "loss": 0.0195,
      "step": 19110
    },
    {
      "epoch": 2.84,
      "learning_rate": 0.0,
      "loss": 0.0431,
      "step": 19140
    },
    {
      "epoch": 2.85,
      "learning_rate": 0.0,
      "loss": 0.0617,
      "step": 19170
    },
    {
      "epoch": 2.85,
      "learning_rate": 0.0,
      "loss": 0.0801,
      "step": 19200
    },
    {
      "epoch": 2.86,
      "learning_rate": 0.0,
      "loss": 0.0386,
      "step": 19230
    },
    {
      "epoch": 2.86,
      "learning_rate": 0.0,
      "loss": 0.066,
      "step": 19260
    },
    {
      "epoch": 2.86,
      "learning_rate": 0.0,
      "loss": 0.0259,
      "step": 19290
    },
    {
      "epoch": 2.87,
      "learning_rate": 0.0,
      "loss": 0.0507,
      "step": 19320
    },
    {
      "epoch": 2.87,
      "learning_rate": 0.0,
      "loss": 0.0988,
      "step": 19350
    },
    {
      "epoch": 2.88,
      "learning_rate": 0.0,
      "loss": 0.053,
      "step": 19380
    },
    {
      "epoch": 2.88,
      "learning_rate": 0.0,
      "loss": 0.0929,
      "step": 19410
    },
    {
      "epoch": 2.89,
      "learning_rate": 0.0,
      "loss": 0.1143,
      "step": 19440
    },
    {
      "epoch": 2.89,
      "learning_rate": 0.0,
      "loss": 0.0529,
      "step": 19470
    },
    {
      "epoch": 2.9,
      "learning_rate": 0.0,
      "loss": 0.049,
      "step": 19500
    },
    {
      "epoch": 2.9,
      "learning_rate": 0.0,
      "loss": 0.0842,
      "step": 19530
    },
    {
      "epoch": 2.9,
      "learning_rate": 0.0,
      "loss": 0.058,
      "step": 19560
    },
    {
      "epoch": 2.91,
      "learning_rate": 0.0,
      "loss": 0.0934,
      "step": 19590
    },
    {
      "epoch": 2.91,
      "learning_rate": 0.0,
      "loss": 0.0733,
      "step": 19620
    },
    {
      "epoch": 2.92,
      "learning_rate": 0.0,
      "loss": 0.1,
      "step": 19650
    },
    {
      "epoch": 2.92,
      "learning_rate": 0.0,
      "loss": 0.0434,
      "step": 19680
    },
    {
      "epoch": 2.93,
      "learning_rate": 0.0,
      "loss": 0.0871,
      "step": 19710
    },
    {
      "epoch": 2.93,
      "learning_rate": 0.0,
      "loss": 0.0417,
      "step": 19740
    },
    {
      "epoch": 2.94,
      "learning_rate": 0.0,
      "loss": 0.055,
      "step": 19770
    },
    {
      "epoch": 2.94,
      "learning_rate": 0.0,
      "loss": 0.1,
      "step": 19800
    },
    {
      "epoch": 2.94,
      "learning_rate": 0.0,
      "loss": 0.0745,
      "step": 19830
    },
    {
      "epoch": 2.95,
      "learning_rate": 0.0,
      "loss": 0.0903,
      "step": 19860
    },
    {
      "epoch": 2.95,
      "learning_rate": 0.0,
      "loss": 0.0968,
      "step": 19890
    },
    {
      "epoch": 2.96,
      "learning_rate": 0.0,
      "loss": 0.0321,
      "step": 19920
    },
    {
      "epoch": 2.96,
      "learning_rate": 0.0,
      "loss": 0.0696,
      "step": 19950
    },
    {
      "epoch": 2.97,
      "learning_rate": 0.0,
      "loss": 0.1039,
      "step": 19980
    },
    {
      "epoch": 2.97,
      "learning_rate": 0.0,
      "loss": 0.1003,
      "step": 20010
    },
    {
      "epoch": 2.98,
      "learning_rate": 0.0,
      "loss": 0.1039,
      "step": 20040
    },
    {
      "epoch": 2.98,
      "learning_rate": 0.0,
      "loss": 0.0965,
      "step": 20070
    },
    {
      "epoch": 2.98,
      "learning_rate": 0.0,
      "loss": 0.1773,
      "step": 20100
    },
    {
      "epoch": 2.99,
      "learning_rate": 0.0,
      "loss": 0.0727,
      "step": 20130
    },
    {
      "epoch": 2.99,
      "learning_rate": 0.0,
      "loss": 0.0601,
      "step": 20160
    },
    {
      "epoch": 3.0,
      "learning_rate": 0.0,
      "loss": 0.0545,
      "step": 20190
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9415137614678899,
      "eval_loss": 0.3246009051799774,
      "eval_runtime": 4.3961,
      "eval_samples_per_second": 198.357,
      "eval_steps_per_second": 20.018,
      "step": 20205
    }
  ],
  "max_steps": 20205,
  "num_train_epochs": 3,
  "total_flos": 3913626734210772.0,
  "trial_name": null,
  "trial_params": null
}