trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.7857142857142856,
  "global_step": 100,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.79,
      "learning_rate": 0.0,
      "loss": 26.9592,
      "step": 100
    },
    {
      "epoch": 1.79,
      "step": 100,
      "total_flos": 2.251300101436512e+16,
      "train_loss": 26.95923095703125,
      "train_runtime": 53.1379,
      "train_samples_per_second": 3.764,
      "train_steps_per_second": 1.882
    }
  ],
  "max_steps": 100,
  "num_train_epochs": 2,
  "total_flos": 2.251300101436512e+16,
  "trial_name": null,
  "trial_params": null
}