trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "global_step": 1242,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.4,
      "learning_rate": 1.7922705314009663e-05,
      "loss": 1.5077,
      "step": 500
    },
    {
      "epoch": 0.81,
      "learning_rate": 5.845410628019324e-06,
      "loss": 1.2752,
      "step": 1000
    },
    {
      "epoch": 1.0,
      "step": 1242,
      "total_flos": 2920187229092352.0,
      "train_loss": 1.3494737904620824,
      "train_runtime": 616.7037,
      "train_samples_per_second": 24.162,
      "train_steps_per_second": 2.014
    }
  ],
  "max_steps": 1242,
  "num_train_epochs": 1,
  "total_flos": 2920187229092352.0,
  "trial_name": null,
  "trial_params": null
}