{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 9,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.67,
      "learning_rate": 0.000175,
      "loss": 17.8562,
      "step": 2
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.000125,
      "loss": 17.4723,
      "step": 4
    },
    {
      "epoch": 2.0,
      "learning_rate": 7.500000000000001e-05,
      "loss": 17.0703,
      "step": 6
    },
    {
      "epoch": 2.67,
      "learning_rate": 2.5e-05,
      "loss": 16.8618,
      "step": 8
    }
  ],
  "logging_steps": 2,
  "max_steps": 9,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 4771393247232.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}