run-1/checkpoint-15500/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.4646130586790136,
  "eval_steps": 500,
  "global_step": 15500,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.42,
      "learning_rate": 2.4108812116072453e-06,
      "loss": 1.0113,
      "step": 15000
    }
  ],
  "logging_steps": 15000,
  "max_steps": 31749,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 108439435898244.0,
  "trial_name": null,
  "trial_params": {
    "learning_rate": 4.570008214658692e-06,
    "per_device_train_batch_size": 1
  }
}