{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 30.257186081694403,
  "eval_steps": 500,
  "global_step": 100000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.3,
      "learning_rate": 4.000000000000001e-06,
      "loss": 7.7084,
      "step": 1000
    },
    {
      "epoch": 0.61,
      "learning_rate": 8.000000000000001e-06,
      "loss": 7.1347,
      "step": 2000
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.2e-05,
      "loss": 6.9286,
      "step": 3000
    },
    {
      "epoch": 1.21,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 6.8404,
      "step": 4000
    },
    {
      "epoch": 1.51,
      "learning_rate": 2e-05,
      "loss": 6.811,
      "step": 5000
    },
    {
      "epoch": 1.82,
      "learning_rate": 2.4e-05,
      "loss": 6.7884,
      "step": 6000
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.8e-05,
      "loss": 6.7671,
      "step": 7000
    },
    {
      "epoch": 2.42,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 6.754,
      "step": 8000
    },
    {
      "epoch": 2.72,
      "learning_rate": 3.6e-05,
      "loss": 6.7459,
      "step": 9000
    },
    {
      "epoch": 3.03,
      "learning_rate": 4e-05,
      "loss": 6.7364,
      "step": 10000
    },
    {
      "epoch": 3.33,
      "learning_rate": 3.955555555555556e-05,
      "loss": 6.7283,
      "step": 11000
    },
    {
      "epoch": 3.63,
      "learning_rate": 3.9111111111111115e-05,
      "loss": 6.7226,
      "step": 12000
    },
    {
      "epoch": 3.93,
      "learning_rate": 3.866666666666667e-05,
      "loss": 6.7186,
      "step": 13000
    },
    {
      "epoch": 4.24,
      "learning_rate": 3.8222222222222226e-05,
      "loss": 6.715,
      "step": 14000
    },
    {
      "epoch": 4.54,
      "learning_rate": 3.777777777777778e-05,
      "loss": 6.7114,
      "step": 15000
    },
    {
      "epoch": 4.84,
      "learning_rate": 3.733333333333334e-05,
      "loss": 6.7088,
      "step": 16000
    },
    {
      "epoch": 5.14,
      "learning_rate": 3.6888888888888896e-05,
      "loss": 6.7072,
      "step": 17000
    },
    {
      "epoch": 5.45,
      "learning_rate": 3.644444444444445e-05,
      "loss": 6.7059,
      "step": 18000
    },
    {
      "epoch": 5.75,
      "learning_rate": 3.6e-05,
      "loss": 6.7039,
      "step": 19000
    },
    {
      "epoch": 6.05,
      "learning_rate": 3.555555555555555e-05,
      "loss": 6.7023,
      "step": 20000
    },
    {
      "epoch": 6.35,
      "learning_rate": 3.511111111111111e-05,
      "loss": 6.7002,
      "step": 21000
    },
    {
      "epoch": 6.66,
      "learning_rate": 3.466666666666667e-05,
      "loss": 6.698,
      "step": 22000
    },
    {
      "epoch": 6.96,
      "learning_rate": 3.4222222222222224e-05,
      "loss": 6.6975,
      "step": 23000
    },
    {
      "epoch": 7.26,
      "learning_rate": 3.377777777777778e-05,
      "loss": 6.6959,
      "step": 24000
    },
    {
      "epoch": 7.56,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 6.6929,
      "step": 25000
    },
    {
      "epoch": 7.87,
      "learning_rate": 3.288888888888889e-05,
      "loss": 6.6936,
      "step": 26000
    },
    {
      "epoch": 8.17,
      "learning_rate": 3.2444444444444446e-05,
      "loss": 6.6917,
      "step": 27000
    },
    {
      "epoch": 8.47,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 6.6915,
      "step": 28000
    },
    {
      "epoch": 8.77,
      "learning_rate": 3.155555555555556e-05,
      "loss": 6.69,
      "step": 29000
    },
    {
      "epoch": 9.08,
      "learning_rate": 3.111111111111112e-05,
      "loss": 6.6884,
      "step": 30000
    },
    {
      "epoch": 9.38,
      "learning_rate": 3.066666666666667e-05,
      "loss": 6.6864,
      "step": 31000
    },
    {
      "epoch": 19.37,
      "learning_rate": 3.0222222222222225e-05,
      "loss": 6.6229,
      "step": 32000
    },
    {
      "epoch": 19.97,
      "learning_rate": 2.977777777777778e-05,
      "loss": 6.5542,
      "step": 33000
    },
    {
      "epoch": 20.58,
      "learning_rate": 2.9333333333333333e-05,
      "loss": 6.5168,
      "step": 34000
    },
    {
      "epoch": 21.19,
      "learning_rate": 2.888888888888889e-05,
      "loss": 6.4593,
      "step": 35000
    },
    {
      "epoch": 21.79,
      "learning_rate": 2.8444444444444447e-05,
      "loss": 6.3693,
      "step": 36000
    },
    {
      "epoch": 22.4,
      "learning_rate": 2.8e-05,
      "loss": 6.2642,
      "step": 37000
    },
    {
      "epoch": 23.0,
      "learning_rate": 2.755555555555556e-05,
      "loss": 6.151,
      "step": 38000
    },
    {
      "epoch": 23.61,
      "learning_rate": 2.7111111111111114e-05,
      "loss": 6.0063,
      "step": 39000
    },
    {
      "epoch": 24.21,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 5.9027,
      "step": 40000
    },
    {
      "epoch": 24.82,
      "learning_rate": 2.6222222222222226e-05,
      "loss": 5.826,
      "step": 41000
    },
    {
      "epoch": 25.42,
      "learning_rate": 2.577777777777778e-05,
      "loss": 5.7664,
      "step": 42000
    },
    {
      "epoch": 26.03,
      "learning_rate": 2.5333333333333334e-05,
      "loss": 5.7222,
      "step": 43000
    },
    {
      "epoch": 26.63,
      "learning_rate": 2.4888888888888893e-05,
      "loss": 5.6801,
      "step": 44000
    },
    {
      "epoch": 27.24,
      "learning_rate": 2.444444444444445e-05,
      "loss": 5.6484,
      "step": 45000
    },
    {
      "epoch": 27.84,
      "learning_rate": 2.4e-05,
      "loss": 5.6185,
      "step": 46000
    },
    {
      "epoch": 28.45,
      "learning_rate": 2.355555555555556e-05,
      "loss": 5.5924,
      "step": 47000
    },
    {
      "epoch": 29.05,
      "learning_rate": 2.3111111111111112e-05,
      "loss": 5.5701,
      "step": 48000
    },
    {
      "epoch": 29.66,
      "learning_rate": 2.2666666666666668e-05,
      "loss": 5.5486,
      "step": 49000
    },
    {
      "epoch": 30.26,
      "learning_rate": 2.2222222222222227e-05,
      "loss": 5.5285,
      "step": 50000
    },
    {
      "epoch": 30.87,
      "learning_rate": 2.177777777777778e-05,
      "loss": 5.5128,
      "step": 51000
    },
    {
      "epoch": 31.47,
      "learning_rate": 2.1333333333333335e-05,
      "loss": 5.4946,
      "step": 52000
    },
    {
      "epoch": 32.08,
      "learning_rate": 2.088888888888889e-05,
      "loss": 5.4802,
      "step": 53000
    },
    {
      "epoch": 32.68,
      "learning_rate": 2.0444444444444446e-05,
      "loss": 5.4654,
      "step": 54000
    },
    {
      "epoch": 33.29,
      "learning_rate": 2e-05,
      "loss": 5.4553,
      "step": 55000
    },
    {
      "epoch": 16.94,
      "learning_rate": 1.9555555555555557e-05,
      "loss": 5.4459,
      "step": 56000
    },
    {
      "epoch": 17.25,
      "learning_rate": 1.9111111111111113e-05,
      "loss": 5.4354,
      "step": 57000
    },
    {
      "epoch": 17.55,
      "learning_rate": 1.866666666666667e-05,
      "loss": 5.4323,
      "step": 58000
    },
    {
      "epoch": 17.85,
      "learning_rate": 1.8222222222222224e-05,
      "loss": 5.4245,
      "step": 59000
    },
    {
      "epoch": 18.15,
      "learning_rate": 1.7777777777777777e-05,
      "loss": 5.4162,
      "step": 60000
    },
    {
      "epoch": 18.46,
      "learning_rate": 1.7333333333333336e-05,
      "loss": 5.4092,
      "step": 61000
    },
    {
      "epoch": 18.76,
      "learning_rate": 1.688888888888889e-05,
      "loss": 5.402,
      "step": 62000
    },
    {
      "epoch": 19.06,
      "learning_rate": 1.6444444444444444e-05,
      "loss": 5.3934,
      "step": 63000
    },
    {
      "epoch": 19.36,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 5.392,
      "step": 64000
    },
    {
      "epoch": 19.67,
      "learning_rate": 1.555555555555556e-05,
      "loss": 5.3835,
      "step": 65000
    },
    {
      "epoch": 19.97,
      "learning_rate": 1.5111111111111112e-05,
      "loss": 5.3828,
      "step": 66000
    },
    {
      "epoch": 20.27,
      "learning_rate": 1.4666666666666666e-05,
      "loss": 5.377,
      "step": 67000
    },
    {
      "epoch": 20.57,
      "learning_rate": 1.4222222222222224e-05,
      "loss": 5.3695,
      "step": 68000
    },
    {
      "epoch": 20.88,
      "learning_rate": 1.377777777777778e-05,
      "loss": 5.3634,
      "step": 69000
    },
    {
      "epoch": 21.18,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 5.358,
      "step": 70000
    },
    {
      "epoch": 21.48,
      "learning_rate": 1.288888888888889e-05,
      "loss": 5.3554,
      "step": 71000
    },
    {
      "epoch": 21.79,
      "learning_rate": 1.2444444444444446e-05,
      "loss": 5.3537,
      "step": 72000
    },
    {
      "epoch": 22.09,
      "learning_rate": 1.2e-05,
      "loss": 5.3486,
      "step": 73000
    },
    {
      "epoch": 22.39,
      "learning_rate": 1.1555555555555556e-05,
      "loss": 5.3431,
      "step": 74000
    },
    {
      "epoch": 22.69,
      "learning_rate": 1.1111111111111113e-05,
      "loss": 5.3404,
      "step": 75000
    },
    {
      "epoch": 23.0,
      "learning_rate": 1.0666666666666667e-05,
      "loss": 5.3377,
      "step": 76000
    },
    {
      "epoch": 23.3,
      "learning_rate": 1.0222222222222223e-05,
      "loss": 5.3317,
      "step": 77000
    },
    {
      "epoch": 23.6,
      "learning_rate": 9.777777777777779e-06,
      "loss": 5.3319,
      "step": 78000
    },
    {
      "epoch": 23.9,
      "learning_rate": 9.333333333333334e-06,
      "loss": 5.3299,
      "step": 79000
    },
    {
      "epoch": 24.21,
      "learning_rate": 8.888888888888888e-06,
      "loss": 5.3232,
      "step": 80000
    },
    {
      "epoch": 24.51,
      "learning_rate": 8.444444444444446e-06,
      "loss": 5.3249,
      "step": 81000
    },
    {
      "epoch": 24.81,
      "learning_rate": 8.000000000000001e-06,
      "loss": 5.3187,
      "step": 82000
    },
    {
      "epoch": 25.11,
      "learning_rate": 7.555555555555556e-06,
      "loss": 5.3174,
      "step": 83000
    },
    {
      "epoch": 25.42,
      "learning_rate": 7.111111111111112e-06,
      "loss": 5.3139,
      "step": 84000
    },
    {
      "epoch": 25.72,
      "learning_rate": 6.666666666666667e-06,
      "loss": 5.3119,
      "step": 85000
    },
    {
      "epoch": 26.02,
      "learning_rate": 6.222222222222223e-06,
      "loss": 5.3143,
      "step": 86000
    },
    {
      "epoch": 26.32,
      "learning_rate": 5.777777777777778e-06,
      "loss": 5.3063,
      "step": 87000
    },
    {
      "epoch": 26.63,
      "learning_rate": 5.333333333333334e-06,
      "loss": 5.3063,
      "step": 88000
    },
    {
      "epoch": 26.93,
      "learning_rate": 4.888888888888889e-06,
      "loss": 5.3096,
      "step": 89000
    },
    {
      "epoch": 27.23,
      "learning_rate": 4.444444444444444e-06,
      "loss": 5.2996,
      "step": 90000
    },
    {
      "epoch": 27.53,
      "learning_rate": 4.000000000000001e-06,
      "loss": 5.3011,
      "step": 91000
    },
    {
      "epoch": 27.84,
      "learning_rate": 3.555555555555556e-06,
      "loss": 5.3027,
      "step": 92000
    },
    {
      "epoch": 28.14,
      "learning_rate": 3.1111111111111116e-06,
      "loss": 5.3007,
      "step": 93000
    },
    {
      "epoch": 28.44,
      "learning_rate": 2.666666666666667e-06,
      "loss": 5.297,
      "step": 94000
    },
    {
      "epoch": 28.74,
      "learning_rate": 2.222222222222222e-06,
      "loss": 5.2979,
      "step": 95000
    },
    {
      "epoch": 29.05,
      "learning_rate": 1.777777777777778e-06,
      "loss": 5.2982,
      "step": 96000
    },
    {
      "epoch": 29.35,
      "learning_rate": 1.3333333333333334e-06,
      "loss": 5.2977,
      "step": 97000
    },
    {
      "epoch": 29.65,
      "learning_rate": 8.88888888888889e-07,
      "loss": 5.2955,
      "step": 98000
    },
    {
      "epoch": 29.95,
      "learning_rate": 4.444444444444445e-07,
      "loss": 5.2938,
      "step": 99000
    },
    {
      "epoch": 30.26,
      "learning_rate": 0.0,
      "loss": 5.2937,
      "step": 100000
    }
  ],
  "logging_steps": 1000,
  "max_steps": 100000,
  "num_train_epochs": 31,
  "save_steps": 1000,
  "total_flos": 9.396281398647587e+18,
  "trial_name": null,
  "trial_params": null
}