trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 149.90140845070422,
  "global_step": 600,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 124.9,
      "learning_rate": 1.4814814814814815e-05,
      "loss": 0.0513,
      "step": 500
    },
    {
      "epoch": 149.9,
      "step": 600,
      "total_flos": 3.308981186551543e+16,
      "train_loss": 0.04397463043530782,
      "train_runtime": 4092.2584,
      "train_samples_per_second": 5.168,
      "train_steps_per_second": 0.147
    }
  ],
  "max_steps": 600,
  "num_train_epochs": 150,
  "total_flos": 3.308981186551543e+16,
  "trial_name": null,
  "trial_params": null
}