{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.3066037735849054,
  "eval_steps": 500,
  "global_step": 195,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.2,
      "learning_rate": 0.00013125000000000002,
      "loss": 0.8436,
      "step": 42
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.000193006993006993,
      "loss": 0.481,
      "step": 84
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.00017832167832167832,
      "loss": 0.3865,
      "step": 126
    },
    {
      "epoch": 2.18,
      "learning_rate": 0.00016363636363636366,
      "loss": 0.2925,
      "step": 168
    }
  ],
  "logging_steps": 42,
  "max_steps": 636,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 4046868719861760.0,
  "trial_name": null,
  "trial_params": null
}