trainer_state.json

{
  "best_metric": 1.4518036842346191,
  "best_model_checkpoint": "sean_test_out/checkpoint-42",
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 42,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "eval_accuracy": 0.353,
      "eval_loss": 1.4518036842346191,
      "eval_runtime": 58.1635,
      "eval_samples_per_second": 171.929,
      "eval_steps_per_second": 7.169,
      "step": 42
    }
  ],
  "logging_steps": 500,
  "max_steps": 42,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "total_flos": 311359843123200.0,
  "train_batch_size": 24,
  "trial_name": null,
  "trial_params": null
}