{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 70.17543859649123,
  "eval_steps": 500,
  "global_step": 100000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.7,
      "learning_rate": 4.000000000000001e-06,
      "loss": 6.8126,
      "step": 1000
    },
    {
      "epoch": 1.4,
      "learning_rate": 8.000000000000001e-06,
      "loss": 6.2581,
      "step": 2000
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.2e-05,
      "loss": 6.1279,
      "step": 3000
    },
    {
      "epoch": 2.81,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 6.0905,
      "step": 4000
    },
    {
      "epoch": 3.51,
      "learning_rate": 2e-05,
      "loss": 6.0732,
      "step": 5000
    },
    {
      "epoch": 4.21,
      "learning_rate": 2.4e-05,
      "loss": 6.0619,
      "step": 6000
    },
    {
      "epoch": 4.91,
      "learning_rate": 2.8e-05,
      "loss": 6.0544,
      "step": 7000
    },
    {
      "epoch": 5.61,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 6.0456,
      "step": 8000
    },
    {
      "epoch": 6.32,
      "learning_rate": 3.6e-05,
      "loss": 6.0417,
      "step": 9000
    },
    {
      "epoch": 7.02,
      "learning_rate": 4e-05,
      "loss": 6.0371,
      "step": 10000
    },
    {
      "epoch": 7.72,
      "learning_rate": 3.955555555555556e-05,
      "loss": 6.033,
      "step": 11000
    },
    {
      "epoch": 8.42,
      "learning_rate": 3.9111111111111115e-05,
      "loss": 6.03,
      "step": 12000
    },
    {
      "epoch": 9.12,
      "learning_rate": 3.866666666666667e-05,
      "loss": 6.0276,
      "step": 13000
    },
    {
      "epoch": 9.82,
      "learning_rate": 3.8222222222222226e-05,
      "loss": 6.0267,
      "step": 14000
    },
    {
      "epoch": 10.53,
      "learning_rate": 3.777777777777778e-05,
      "loss": 6.0213,
      "step": 15000
    },
    {
      "epoch": 11.23,
      "learning_rate": 3.733333333333334e-05,
      "loss": 5.9847,
      "step": 16000
    },
    {
      "epoch": 11.93,
      "learning_rate": 3.6888888888888896e-05,
      "loss": 5.8953,
      "step": 17000
    },
    {
      "epoch": 12.63,
      "learning_rate": 3.644444444444445e-05,
      "loss": 5.8317,
      "step": 18000
    },
    {
      "epoch": 13.33,
      "learning_rate": 3.6e-05,
      "loss": 5.7704,
      "step": 19000
    },
    {
      "epoch": 14.04,
      "learning_rate": 3.555555555555555e-05,
      "loss": 5.7067,
      "step": 20000
    },
    {
      "epoch": 14.74,
      "learning_rate": 3.511111111111111e-05,
      "loss": 5.6463,
      "step": 21000
    },
    {
      "epoch": 15.44,
      "learning_rate": 3.466666666666667e-05,
      "loss": 5.5902,
      "step": 22000
    },
    {
      "epoch": 16.14,
      "learning_rate": 3.4222222222222224e-05,
      "loss": 5.5284,
      "step": 23000
    },
    {
      "epoch": 16.84,
      "learning_rate": 3.377777777777778e-05,
      "loss": 5.4464,
      "step": 24000
    },
    {
      "epoch": 17.54,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 5.2621,
      "step": 25000
    },
    {
      "epoch": 18.25,
      "learning_rate": 3.288888888888889e-05,
      "loss": 5.1088,
      "step": 26000
    },
    {
      "epoch": 18.95,
      "learning_rate": 3.2444444444444446e-05,
      "loss": 5.0189,
      "step": 27000
    },
    {
      "epoch": 19.65,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 4.9616,
      "step": 28000
    },
    {
      "epoch": 20.35,
      "learning_rate": 3.155555555555556e-05,
      "loss": 4.9103,
      "step": 29000
    },
    {
      "epoch": 21.05,
      "learning_rate": 3.111111111111112e-05,
      "loss": 4.8747,
      "step": 30000
    },
    {
      "epoch": 21.75,
      "learning_rate": 3.066666666666667e-05,
      "loss": 4.8424,
      "step": 31000
    },
    {
      "epoch": 22.46,
      "learning_rate": 3.0222222222222225e-05,
      "loss": 4.8151,
      "step": 32000
    },
    {
      "epoch": 23.16,
      "learning_rate": 2.977777777777778e-05,
      "loss": 4.7914,
      "step": 33000
    },
    {
      "epoch": 23.86,
      "learning_rate": 2.9333333333333333e-05,
      "loss": 4.7693,
      "step": 34000
    },
    {
      "epoch": 24.56,
      "learning_rate": 2.888888888888889e-05,
      "loss": 4.7507,
      "step": 35000
    },
    {
      "epoch": 25.26,
      "learning_rate": 2.8444444444444447e-05,
      "loss": 4.7284,
      "step": 36000
    },
    {
      "epoch": 25.96,
      "learning_rate": 2.8e-05,
      "loss": 4.7164,
      "step": 37000
    },
    {
      "epoch": 26.67,
      "learning_rate": 2.755555555555556e-05,
      "loss": 4.7017,
      "step": 38000
    },
    {
      "epoch": 27.37,
      "learning_rate": 2.7111111111111114e-05,
      "loss": 4.6878,
      "step": 39000
    },
    {
      "epoch": 28.07,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 4.6739,
      "step": 40000
    },
    {
      "epoch": 28.77,
      "learning_rate": 2.6222222222222226e-05,
      "loss": 4.665,
      "step": 41000
    },
    {
      "epoch": 29.47,
      "learning_rate": 2.577777777777778e-05,
      "loss": 4.652,
      "step": 42000
    },
    {
      "epoch": 30.18,
      "learning_rate": 2.5333333333333334e-05,
      "loss": 4.6436,
      "step": 43000
    },
    {
      "epoch": 30.88,
      "learning_rate": 2.4888888888888893e-05,
      "loss": 4.6312,
      "step": 44000
    },
    {
      "epoch": 31.58,
      "learning_rate": 2.444444444444445e-05,
      "loss": 4.6219,
      "step": 45000
    },
    {
      "epoch": 32.28,
      "learning_rate": 2.4e-05,
      "loss": 4.6139,
      "step": 46000
    },
    {
      "epoch": 32.98,
      "learning_rate": 2.355555555555556e-05,
      "loss": 4.6054,
      "step": 47000
    },
    {
      "epoch": 33.68,
      "learning_rate": 2.3111111111111112e-05,
      "loss": 4.5959,
      "step": 48000
    },
    {
      "epoch": 34.39,
      "learning_rate": 2.2666666666666668e-05,
      "loss": 4.5901,
      "step": 49000
    },
    {
      "epoch": 35.09,
      "learning_rate": 2.2222222222222227e-05,
      "loss": 4.5829,
      "step": 50000
    },
    {
      "epoch": 35.79,
      "learning_rate": 2.177777777777778e-05,
      "loss": 4.5763,
      "step": 51000
    },
    {
      "epoch": 36.49,
      "learning_rate": 2.1333333333333335e-05,
      "loss": 4.5683,
      "step": 52000
    },
    {
      "epoch": 37.19,
      "learning_rate": 2.088888888888889e-05,
      "loss": 4.5641,
      "step": 53000
    },
    {
      "epoch": 37.89,
      "learning_rate": 2.0444444444444446e-05,
      "loss": 4.5564,
      "step": 54000
    },
    {
      "epoch": 38.6,
      "learning_rate": 2e-05,
      "loss": 4.5514,
      "step": 55000
    },
    {
      "epoch": 39.3,
      "learning_rate": 1.9555555555555557e-05,
      "loss": 4.5475,
      "step": 56000
    },
    {
      "epoch": 40.0,
      "learning_rate": 1.9111111111111113e-05,
      "loss": 4.5393,
      "step": 57000
    },
    {
      "epoch": 40.7,
      "learning_rate": 1.866666666666667e-05,
      "loss": 4.5354,
      "step": 58000
    },
    {
      "epoch": 41.4,
      "learning_rate": 1.8222222222222224e-05,
      "loss": 4.529,
      "step": 59000
    },
    {
      "epoch": 42.11,
      "learning_rate": 1.7777777777777777e-05,
      "loss": 4.5271,
      "step": 60000
    },
    {
      "epoch": 42.81,
      "learning_rate": 1.7333333333333336e-05,
      "loss": 4.521,
      "step": 61000
    },
    {
      "epoch": 43.51,
      "learning_rate": 1.688888888888889e-05,
      "loss": 4.5139,
      "step": 62000
    },
    {
      "epoch": 44.21,
      "learning_rate": 1.6444444444444444e-05,
      "loss": 4.513,
      "step": 63000
    },
    {
      "epoch": 44.91,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 4.5078,
      "step": 64000
    },
    {
      "epoch": 45.61,
      "learning_rate": 1.555555555555556e-05,
      "loss": 4.5057,
      "step": 65000
    },
    {
      "epoch": 46.32,
      "learning_rate": 1.5111111111111112e-05,
      "loss": 4.4992,
      "step": 66000
    },
    {
      "epoch": 47.02,
      "learning_rate": 1.4666666666666666e-05,
      "loss": 4.4969,
      "step": 67000
    },
    {
      "epoch": 47.72,
      "learning_rate": 1.4222222222222224e-05,
      "loss": 4.4919,
      "step": 68000
    },
    {
      "epoch": 48.42,
      "learning_rate": 1.377777777777778e-05,
      "loss": 4.4912,
      "step": 69000
    },
    {
      "epoch": 49.12,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 4.4854,
      "step": 70000
    },
    {
      "epoch": 49.82,
      "learning_rate": 1.288888888888889e-05,
      "loss": 4.4829,
      "step": 71000
    },
    {
      "epoch": 50.53,
      "learning_rate": 1.2444444444444446e-05,
      "loss": 4.4789,
      "step": 72000
    },
    {
      "epoch": 51.23,
      "learning_rate": 1.2e-05,
      "loss": 4.4767,
      "step": 73000
    },
    {
      "epoch": 51.93,
      "learning_rate": 1.1555555555555556e-05,
      "loss": 4.4742,
      "step": 74000
    },
    {
      "epoch": 52.63,
      "learning_rate": 1.1111111111111113e-05,
      "loss": 4.4719,
      "step": 75000
    },
    {
      "epoch": 53.33,
      "learning_rate": 1.0666666666666667e-05,
      "loss": 4.4681,
      "step": 76000
    },
    {
      "epoch": 54.04,
      "learning_rate": 1.0222222222222223e-05,
      "loss": 4.4674,
      "step": 77000
    },
    {
      "epoch": 54.74,
      "learning_rate": 9.777777777777779e-06,
      "loss": 4.4643,
      "step": 78000
    },
    {
      "epoch": 55.44,
      "learning_rate": 9.333333333333334e-06,
      "loss": 4.4599,
      "step": 79000
    },
    {
      "epoch": 56.14,
      "learning_rate": 8.888888888888888e-06,
      "loss": 4.4617,
      "step": 80000
    },
    {
      "epoch": 56.84,
      "learning_rate": 8.444444444444446e-06,
      "loss": 4.4572,
      "step": 81000
    },
    {
      "epoch": 57.54,
      "learning_rate": 8.000000000000001e-06,
      "loss": 4.4539,
      "step": 82000
    },
    {
      "epoch": 58.25,
      "learning_rate": 7.555555555555556e-06,
      "loss": 4.4574,
      "step": 83000
    },
    {
      "epoch": 58.95,
      "learning_rate": 7.111111111111112e-06,
      "loss": 4.4503,
      "step": 84000
    },
    {
      "epoch": 59.65,
      "learning_rate": 6.666666666666667e-06,
      "loss": 4.4499,
      "step": 85000
    },
    {
      "epoch": 60.35,
      "learning_rate": 6.222222222222223e-06,
      "loss": 4.4484,
      "step": 86000
    },
    {
      "epoch": 61.05,
      "learning_rate": 5.777777777777778e-06,
      "loss": 4.4481,
      "step": 87000
    },
    {
      "epoch": 61.75,
      "learning_rate": 5.333333333333334e-06,
      "loss": 4.4447,
      "step": 88000
    },
    {
      "epoch": 62.46,
      "learning_rate": 4.888888888888889e-06,
      "loss": 4.4435,
      "step": 89000
    },
    {
      "epoch": 63.16,
      "learning_rate": 4.444444444444444e-06,
      "loss": 4.4435,
      "step": 90000
    },
    {
      "epoch": 63.86,
      "learning_rate": 4.000000000000001e-06,
      "loss": 4.4434,
      "step": 91000
    },
    {
      "epoch": 64.56,
      "learning_rate": 3.555555555555556e-06,
      "loss": 4.4409,
      "step": 92000
    },
    {
      "epoch": 65.26,
      "learning_rate": 3.1111111111111116e-06,
      "loss": 4.4403,
      "step": 93000
    },
    {
      "epoch": 65.96,
      "learning_rate": 2.666666666666667e-06,
      "loss": 4.4409,
      "step": 94000
    },
    {
      "epoch": 66.67,
      "learning_rate": 2.222222222222222e-06,
      "loss": 4.4382,
      "step": 95000
    },
    {
      "epoch": 67.37,
      "learning_rate": 1.777777777777778e-06,
      "loss": 4.437,
      "step": 96000
    },
    {
      "epoch": 68.07,
      "learning_rate": 1.3333333333333334e-06,
      "loss": 4.4376,
      "step": 97000
    },
    {
      "epoch": 68.77,
      "learning_rate": 8.88888888888889e-07,
      "loss": 4.4357,
      "step": 98000
    },
    {
      "epoch": 69.47,
      "learning_rate": 4.444444444444445e-07,
      "loss": 4.4373,
      "step": 99000
    },
    {
      "epoch": 70.18,
      "learning_rate": 0.0,
      "loss": 4.4348,
      "step": 100000
    },
    {
      "epoch": 70.18,
      "step": 100000,
      "total_flos": 2.020741224679342e+19,
      "train_loss": 4.91067076171875,
      "train_runtime": 82541.1995,
      "train_samples_per_second": 930.444,
      "train_steps_per_second": 1.212
    }
  ],
  "logging_steps": 1000,
  "max_steps": 100000,
  "num_train_epochs": 71,
  "save_steps": 1000,
  "total_flos": 2.020741224679342e+19,
  "trial_name": null,
  "trial_params": null
}