{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 28.320558806871816,
  "eval_steps": 500,
  "global_step": 100000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.28,
      "learning_rate": 4.000000000000001e-06,
      "loss": 7.4059,
      "step": 1000
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.000000000000001e-06,
      "loss": 6.7738,
      "step": 2000
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.2e-05,
      "loss": 6.5767,
      "step": 3000
    },
    {
      "epoch": 1.13,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 6.4861,
      "step": 4000
    },
    {
      "epoch": 1.42,
      "learning_rate": 2e-05,
      "loss": 6.4545,
      "step": 5000
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.4e-05,
      "loss": 6.4371,
      "step": 6000
    },
    {
      "epoch": 1.98,
      "learning_rate": 2.8e-05,
      "loss": 6.4229,
      "step": 7000
    },
    {
      "epoch": 2.27,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 6.4124,
      "step": 8000
    },
    {
      "epoch": 2.55,
      "learning_rate": 3.6e-05,
      "loss": 6.4036,
      "step": 9000
    },
    {
      "epoch": 2.83,
      "learning_rate": 4e-05,
      "loss": 6.3949,
      "step": 10000
    },
    {
      "epoch": 3.11,
      "learning_rate": 3.955555555555556e-05,
      "loss": 6.3881,
      "step": 11000
    },
    {
      "epoch": 3.4,
      "learning_rate": 3.9111111111111115e-05,
      "loss": 6.3816,
      "step": 12000
    },
    {
      "epoch": 3.68,
      "learning_rate": 3.866666666666667e-05,
      "loss": 6.3782,
      "step": 13000
    },
    {
      "epoch": 3.96,
      "learning_rate": 3.8222222222222226e-05,
      "loss": 6.3737,
      "step": 14000
    },
    {
      "epoch": 4.25,
      "learning_rate": 3.777777777777778e-05,
      "loss": 6.3709,
      "step": 15000
    },
    {
      "epoch": 4.53,
      "learning_rate": 3.733333333333334e-05,
      "loss": 6.3688,
      "step": 16000
    },
    {
      "epoch": 4.81,
      "learning_rate": 3.6888888888888896e-05,
      "loss": 6.366,
      "step": 17000
    },
    {
      "epoch": 5.1,
      "learning_rate": 3.644444444444445e-05,
      "loss": 6.3647,
      "step": 18000
    },
    {
      "epoch": 5.38,
      "learning_rate": 3.6e-05,
      "loss": 6.3626,
      "step": 19000
    },
    {
      "epoch": 5.66,
      "learning_rate": 3.555555555555555e-05,
      "loss": 6.3604,
      "step": 20000
    },
    {
      "epoch": 5.95,
      "learning_rate": 3.511111111111111e-05,
      "loss": 6.3601,
      "step": 21000
    },
    {
      "epoch": 6.23,
      "learning_rate": 3.466666666666667e-05,
      "loss": 6.3572,
      "step": 22000
    },
    {
      "epoch": 6.51,
      "learning_rate": 3.4222222222222224e-05,
      "loss": 6.3567,
      "step": 23000
    },
    {
      "epoch": 6.8,
      "learning_rate": 3.377777777777778e-05,
      "loss": 6.3569,
      "step": 24000
    },
    {
      "epoch": 7.08,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 6.3555,
      "step": 25000
    },
    {
      "epoch": 7.36,
      "learning_rate": 3.288888888888889e-05,
      "loss": 6.3541,
      "step": 26000
    },
    {
      "epoch": 7.65,
      "learning_rate": 3.2444444444444446e-05,
      "loss": 6.353,
      "step": 27000
    },
    {
      "epoch": 7.93,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 6.352,
      "step": 28000
    },
    {
      "epoch": 8.21,
      "learning_rate": 3.155555555555556e-05,
      "loss": 6.3374,
      "step": 29000
    },
    {
      "epoch": 8.5,
      "learning_rate": 3.111111111111112e-05,
      "loss": 6.274,
      "step": 30000
    },
    {
      "epoch": 8.78,
      "learning_rate": 3.066666666666667e-05,
      "loss": 6.2403,
      "step": 31000
    },
    {
      "epoch": 9.06,
      "learning_rate": 3.0222222222222225e-05,
      "loss": 6.219,
      "step": 32000
    },
    {
      "epoch": 9.35,
      "learning_rate": 2.977777777777778e-05,
      "loss": 6.194,
      "step": 33000
    },
    {
      "epoch": 9.63,
      "learning_rate": 2.9333333333333333e-05,
      "loss": 6.1634,
      "step": 34000
    },
    {
      "epoch": 9.91,
      "learning_rate": 2.888888888888889e-05,
      "loss": 6.136,
      "step": 35000
    },
    {
      "epoch": 10.2,
      "learning_rate": 2.8444444444444447e-05,
      "loss": 6.0914,
      "step": 36000
    },
    {
      "epoch": 10.48,
      "learning_rate": 2.8e-05,
      "loss": 6.0379,
      "step": 37000
    },
    {
      "epoch": 10.76,
      "learning_rate": 2.755555555555556e-05,
      "loss": 5.9794,
      "step": 38000
    },
    {
      "epoch": 11.04,
      "learning_rate": 2.7111111111111114e-05,
      "loss": 5.9123,
      "step": 39000
    },
    {
      "epoch": 11.33,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 5.8473,
      "step": 40000
    },
    {
      "epoch": 11.61,
      "learning_rate": 2.6222222222222226e-05,
      "loss": 5.761,
      "step": 41000
    },
    {
      "epoch": 11.89,
      "learning_rate": 2.577777777777778e-05,
      "loss": 5.6793,
      "step": 42000
    },
    {
      "epoch": 12.18,
      "learning_rate": 2.5333333333333334e-05,
      "loss": 5.6205,
      "step": 43000
    },
    {
      "epoch": 12.46,
      "learning_rate": 2.4888888888888893e-05,
      "loss": 5.5763,
      "step": 44000
    },
    {
      "epoch": 12.74,
      "learning_rate": 2.444444444444445e-05,
      "loss": 5.5355,
      "step": 45000
    },
    {
      "epoch": 13.03,
      "learning_rate": 2.4e-05,
      "loss": 5.5028,
      "step": 46000
    },
    {
      "epoch": 13.31,
      "learning_rate": 2.355555555555556e-05,
      "loss": 5.4741,
      "step": 47000
    },
    {
      "epoch": 13.59,
      "learning_rate": 2.3111111111111112e-05,
      "loss": 5.4451,
      "step": 48000
    },
    {
      "epoch": 13.88,
      "learning_rate": 2.2666666666666668e-05,
      "loss": 5.4244,
      "step": 49000
    },
    {
      "epoch": 14.16,
      "learning_rate": 2.2222222222222227e-05,
      "loss": 5.4003,
      "step": 50000
    },
    {
      "epoch": 14.44,
      "learning_rate": 2.177777777777778e-05,
      "loss": 5.3802,
      "step": 51000
    },
    {
      "epoch": 14.73,
      "learning_rate": 2.1333333333333335e-05,
      "loss": 5.3606,
      "step": 52000
    },
    {
      "epoch": 15.01,
      "learning_rate": 2.088888888888889e-05,
      "loss": 5.3435,
      "step": 53000
    },
    {
      "epoch": 15.29,
      "learning_rate": 2.0444444444444446e-05,
      "loss": 5.3308,
      "step": 54000
    },
    {
      "epoch": 15.58,
      "learning_rate": 2e-05,
      "loss": 5.3103,
      "step": 55000
    },
    {
      "epoch": 15.86,
      "learning_rate": 1.9555555555555557e-05,
      "loss": 5.2965,
      "step": 56000
    },
    {
      "epoch": 16.14,
      "learning_rate": 1.9111111111111113e-05,
      "loss": 5.2886,
      "step": 57000
    },
    {
      "epoch": 16.43,
      "learning_rate": 1.866666666666667e-05,
      "loss": 5.2752,
      "step": 58000
    },
    {
      "epoch": 16.71,
      "learning_rate": 1.8222222222222224e-05,
      "loss": 5.2649,
      "step": 59000
    },
    {
      "epoch": 16.99,
      "learning_rate": 1.7777777777777777e-05,
      "loss": 5.2527,
      "step": 60000
    },
    {
      "epoch": 17.27,
      "learning_rate": 1.7333333333333336e-05,
      "loss": 5.2436,
      "step": 61000
    },
    {
      "epoch": 17.56,
      "learning_rate": 1.688888888888889e-05,
      "loss": 5.2318,
      "step": 62000
    },
    {
      "epoch": 17.84,
      "learning_rate": 1.6444444444444444e-05,
      "loss": 5.2264,
      "step": 63000
    },
    {
      "epoch": 18.12,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 5.218,
      "step": 64000
    },
    {
      "epoch": 18.41,
      "learning_rate": 1.555555555555556e-05,
      "loss": 5.2067,
      "step": 65000
    },
    {
      "epoch": 18.69,
      "learning_rate": 1.5111111111111112e-05,
      "loss": 5.2002,
      "step": 66000
    },
    {
      "epoch": 18.97,
      "learning_rate": 1.4666666666666666e-05,
      "loss": 5.1996,
      "step": 67000
    },
    {
      "epoch": 19.26,
      "learning_rate": 1.4222222222222224e-05,
      "loss": 5.1862,
      "step": 68000
    },
    {
      "epoch": 19.54,
      "learning_rate": 1.377777777777778e-05,
      "loss": 5.1821,
      "step": 69000
    },
    {
      "epoch": 19.82,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 5.1776,
      "step": 70000
    },
    {
      "epoch": 20.11,
      "learning_rate": 1.288888888888889e-05,
      "loss": 5.1709,
      "step": 71000
    },
    {
      "epoch": 20.39,
      "learning_rate": 1.2444444444444446e-05,
      "loss": 5.1644,
      "step": 72000
    },
    {
      "epoch": 20.67,
      "learning_rate": 1.2e-05,
      "loss": 5.1592,
      "step": 73000
    },
    {
      "epoch": 20.96,
      "learning_rate": 1.1555555555555556e-05,
      "loss": 5.158,
      "step": 74000
    },
    {
      "epoch": 21.24,
      "learning_rate": 1.1111111111111113e-05,
      "loss": 5.1505,
      "step": 75000
    },
    {
      "epoch": 21.52,
      "learning_rate": 1.0666666666666667e-05,
      "loss": 5.1463,
      "step": 76000
    },
    {
      "epoch": 21.81,
      "learning_rate": 1.0222222222222223e-05,
      "loss": 5.1434,
      "step": 77000
    },
    {
      "epoch": 22.09,
      "learning_rate": 9.777777777777779e-06,
      "loss": 5.1396,
      "step": 78000
    },
    {
      "epoch": 22.37,
      "learning_rate": 9.333333333333334e-06,
      "loss": 5.1338,
      "step": 79000
    },
    {
      "epoch": 22.66,
      "learning_rate": 8.888888888888888e-06,
      "loss": 5.1323,
      "step": 80000
    },
    {
      "epoch": 22.94,
      "learning_rate": 8.444444444444446e-06,
      "loss": 5.1303,
      "step": 81000
    },
    {
      "epoch": 23.22,
      "learning_rate": 8.000000000000001e-06,
      "loss": 5.1224,
      "step": 82000
    },
    {
      "epoch": 23.51,
      "learning_rate": 7.555555555555556e-06,
      "loss": 5.1214,
      "step": 83000
    },
    {
      "epoch": 23.79,
      "learning_rate": 7.111111111111112e-06,
      "loss": 5.1226,
      "step": 84000
    },
    {
      "epoch": 24.07,
      "learning_rate": 6.666666666666667e-06,
      "loss": 5.1185,
      "step": 85000
    },
    {
      "epoch": 24.36,
      "learning_rate": 6.222222222222223e-06,
      "loss": 5.1089,
      "step": 86000
    },
    {
      "epoch": 24.64,
      "learning_rate": 5.777777777777778e-06,
      "loss": 5.111,
      "step": 87000
    },
    {
      "epoch": 24.92,
      "learning_rate": 5.333333333333334e-06,
      "loss": 5.1099,
      "step": 88000
    },
    {
      "epoch": 25.2,
      "learning_rate": 4.888888888888889e-06,
      "loss": 5.1088,
      "step": 89000
    },
    {
      "epoch": 25.49,
      "learning_rate": 4.444444444444444e-06,
      "loss": 5.1057,
      "step": 90000
    },
    {
      "epoch": 25.77,
      "learning_rate": 4.000000000000001e-06,
      "loss": 5.1054,
      "step": 91000
    },
    {
      "epoch": 26.05,
      "learning_rate": 3.555555555555556e-06,
      "loss": 5.0996,
      "step": 92000
    },
    {
      "epoch": 26.34,
      "learning_rate": 3.1111111111111116e-06,
      "loss": 5.1013,
      "step": 93000
    },
    {
      "epoch": 26.62,
      "learning_rate": 2.666666666666667e-06,
      "loss": 5.1,
      "step": 94000
    },
    {
      "epoch": 26.9,
      "learning_rate": 2.222222222222222e-06,
      "loss": 5.0977,
      "step": 95000
    },
    {
      "epoch": 27.19,
      "learning_rate": 1.777777777777778e-06,
      "loss": 5.0966,
      "step": 96000
    },
    {
      "epoch": 27.47,
      "learning_rate": 1.3333333333333334e-06,
      "loss": 5.098,
      "step": 97000
    },
    {
      "epoch": 27.75,
      "learning_rate": 8.88888888888889e-07,
      "loss": 5.0919,
      "step": 98000
    },
    {
      "epoch": 28.04,
      "learning_rate": 4.444444444444445e-07,
      "loss": 5.0967,
      "step": 99000
    },
    {
      "epoch": 28.32,
      "learning_rate": 0.0,
      "loss": 5.0951,
      "step": 100000
    }
  ],
  "logging_steps": 1000,
  "max_steps": 100000,
  "num_train_epochs": 29,
  "save_steps": 1000,
  "total_flos": 7.577756588702695e+18,
  "trial_name": null,
  "trial_params": null
}