trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.29698375870069604,
  "eval_steps": 500,
  "global_step": 3,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.1,
      "learning_rate": 2e-05,
      "loss": 1.3053,
      "step": 1
    },
    {
      "epoch": 0.3,
      "eval_loss": 1.2855874300003052,
      "eval_runtime": 54.5811,
      "eval_samples_per_second": 10.535,
      "eval_steps_per_second": 1.319,
      "step": 3
    },
    {
      "epoch": 0.3,
      "step": 3,
      "total_flos": 1.7380242293260288e+17,
      "train_loss": 1.6732101043065388,
      "train_runtime": 1592.7916,
      "train_samples_per_second": 3.245,
      "train_steps_per_second": 0.006
    }
  ],
  "logging_steps": 5,
  "max_steps": 10,
  "num_train_epochs": 1,
  "save_steps": 500,
  "total_flos": 1.7380242293260288e+17,
  "trial_name": null,
  "trial_params": null
}