trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 10.0,
  "eval_steps": 1000,
  "global_step": 150,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 6.67,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 7.3982,
      "step": 100
    },
    {
      "epoch": 10.0,
      "step": 150,
      "total_flos": 0.0,
      "train_loss": 7.274218139648437,
      "train_runtime": 171.1896,
      "train_samples_per_second": 1.752,
      "train_steps_per_second": 0.876
    }
  ],
  "logging_steps": 100,
  "max_steps": 150,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "total_flos": 0.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}