trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 49.857142857142854,
  "global_step": 150,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 33.29,
      "learning_rate": 3.675e-06,
      "loss": 17.0308,
      "step": 100
    },
    {
      "epoch": 49.86,
      "step": 150,
      "total_flos": 7.155226924023398e+17,
      "train_loss": 15.253912760416666,
      "train_runtime": 565.3944,
      "train_samples_per_second": 9.905,
      "train_steps_per_second": 0.265
    }
  ],
  "max_steps": 150,
  "num_train_epochs": 50,
  "total_flos": 7.155226924023398e+17,
  "trial_name": null,
  "trial_params": null
}