trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.334722801167153,
  "eval_steps": 500,
  "global_step": 500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 3.33,
      "grad_norm": 2.0089468955993652,
      "learning_rate": 1.0396039603960395e-05,
      "loss": 0.5324,
      "step": 500
    },
    {
      "epoch": 3.33,
      "eval_loss": 0.39683377742767334,
      "eval_runtime": 31.7252,
      "eval_samples_per_second": 15.76,
      "eval_steps_per_second": 7.88,
      "step": 500
    }
  ],
  "logging_steps": 500,
  "max_steps": 745,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "total_flos": 1.948665249792e+16,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}