{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.32439446366782004,
  "eval_steps": 500,
  "global_step": 1500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.11,
      "learning_rate": 4.819780853517878e-05,
      "loss": 1.1469,
      "step": 500
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.639561707035756e-05,
      "loss": 0.698,
      "step": 1000
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.459342560553633e-05,
      "loss": 0.6134,
      "step": 1500
    }
  ],
  "logging_steps": 500,
  "max_steps": 13872,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 2.563531776e+16,
  "train_batch_size": 128,
  "trial_name": null,
  "trial_params": null
}