trainer_state.json

{
  "best_metric": 1.1936179399490356,
  "best_model_checkpoint": "sean_test_out/checkpoint-84",
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 84,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "eval_accuracy": 0.4791,
      "eval_loss": 1.1936179399490356,
      "eval_runtime": 59.2942,
      "eval_samples_per_second": 168.651,
      "eval_steps_per_second": 7.033,
      "step": 84
    }
  ],
  "logging_steps": 500,
  "max_steps": 84,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "total_flos": 611842914385920.0,
  "train_batch_size": 24,
  "trial_name": null,
  "trial_params": null
}