{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.05,
  "global_step": 50000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 4.7524,
      "step": 1000
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.000000000000001e-06,
      "loss": 3.5548,
      "step": 2000
    },
    {
      "epoch": 0.0,
      "learning_rate": 6e-06,
      "loss": 2.9636,
      "step": 3000
    },
    {
      "epoch": 0.0,
      "learning_rate": 8.000000000000001e-06,
      "loss": 2.7618,
      "step": 4000
    },
    {
      "epoch": 0.01,
      "learning_rate": 1e-05,
      "loss": 2.6257,
      "step": 5000
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.2e-05,
      "loss": 2.5517,
      "step": 6000
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.4000000000000001e-05,
      "loss": 2.4628,
      "step": 7000
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 2.4338,
      "step": 8000
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.8e-05,
      "loss": 2.3362,
      "step": 9000
    },
    {
      "epoch": 0.01,
      "learning_rate": 2e-05,
      "loss": 2.2746,
      "step": 10000
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.2000000000000003e-05,
      "loss": 2.1795,
      "step": 11000
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4e-05,
      "loss": 2.2309,
      "step": 12000
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.6000000000000002e-05,
      "loss": 1.8605,
      "step": 13000
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.8000000000000003e-05,
      "loss": 1.7232,
      "step": 14000
    },
    {
      "epoch": 0.01,
      "learning_rate": 3e-05,
      "loss": 1.6872,
      "step": 15000
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 2.1581,
      "step": 16000
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.4000000000000007e-05,
      "loss": 2.0653,
      "step": 17000
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6e-05,
      "loss": 2.049,
      "step": 18000
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.8e-05,
      "loss": 2.0773,
      "step": 19000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4e-05,
      "loss": 1.9397,
      "step": 20000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.2e-05,
      "loss": 1.8899,
      "step": 21000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 1.8474,
      "step": 22000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.600000000000001e-05,
      "loss": 1.8034,
      "step": 23000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8e-05,
      "loss": 1.7666,
      "step": 24000
    },
    {
      "epoch": 0.03,
      "learning_rate": 5e-05,
      "loss": 1.7899,
      "step": 25000
    },
    {
      "epoch": 0.03,
      "learning_rate": 5.2000000000000004e-05,
      "loss": 1.4265,
      "step": 26000
    },
    {
      "epoch": 0.03,
      "learning_rate": 5.4000000000000005e-05,
      "loss": 1.3535,
      "step": 27000
    },
    {
      "epoch": 0.03,
      "learning_rate": 5.6000000000000006e-05,
      "loss": 1.1703,
      "step": 28000
    },
    {
      "epoch": 0.03,
      "learning_rate": 5.8e-05,
      "loss": 1.129,
      "step": 29000
    },
    {
      "epoch": 0.03,
      "learning_rate": 6e-05,
      "loss": 1.2811,
      "step": 30000
    },
    {
      "epoch": 0.03,
      "learning_rate": 6.2e-05,
      "loss": 1.2189,
      "step": 31000
    },
    {
      "epoch": 0.03,
      "learning_rate": 6.400000000000001e-05,
      "loss": 1.4371,
      "step": 32000
    },
    {
      "epoch": 0.03,
      "learning_rate": 6.6e-05,
      "loss": 1.3551,
      "step": 33000
    },
    {
      "epoch": 0.03,
      "learning_rate": 6.800000000000001e-05,
      "loss": 1.3616,
      "step": 34000
    },
    {
      "epoch": 0.04,
      "learning_rate": 7e-05,
      "loss": 1.2483,
      "step": 35000
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.2e-05,
      "loss": 1.2174,
      "step": 36000
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.4e-05,
      "loss": 1.1861,
      "step": 37000
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.6e-05,
      "loss": 1.1465,
      "step": 38000
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.800000000000001e-05,
      "loss": 1.1957,
      "step": 39000
    },
    {
      "epoch": 0.04,
      "learning_rate": 8e-05,
      "loss": 1.136,
      "step": 40000
    },
    {
      "epoch": 0.04,
      "learning_rate": 8.2e-05,
      "loss": 1.1122,
      "step": 41000
    },
    {
      "epoch": 0.04,
      "learning_rate": 8.4e-05,
      "loss": 1.0905,
      "step": 42000
    },
    {
      "epoch": 0.04,
      "learning_rate": 8.6e-05,
      "loss": 1.0996,
      "step": 43000
    },
    {
      "epoch": 0.04,
      "learning_rate": 8.800000000000001e-05,
      "loss": 1.0632,
      "step": 44000
    },
    {
      "epoch": 0.04,
      "learning_rate": 9e-05,
      "loss": 1.1387,
      "step": 45000
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.200000000000001e-05,
      "loss": 1.2049,
      "step": 46000
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.4e-05,
      "loss": 1.1783,
      "step": 47000
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.6e-05,
      "loss": 1.1043,
      "step": 48000
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.8e-05,
      "loss": 1.0633,
      "step": 49000
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0001,
      "loss": 1.1331,
      "step": 50000
    },
    {
      "epoch": 0.05,
      "eval_loss": 0.753776490688324,
      "eval_runtime": 33.8225,
      "eval_samples_per_second": 147.83,
      "eval_steps_per_second": 2.336,
      "step": 50000
    }
  ],
  "max_steps": 1000000,
  "num_train_epochs": 9223372036854775807,
  "total_flos": 2.9824949157888e+18,
  "trial_name": null,
  "trial_params": null
}