trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 100.0,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 50.0,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 2.2247,
      "step": 100
    },
    {
      "epoch": 100.0,
      "learning_rate": 0.0,
      "loss": 0.0076,
      "step": 200
    },
    {
      "epoch": 100.0,
      "step": 200,
      "total_flos": 255121579477200.0,
      "train_loss": 1.1161543041467668,
      "train_runtime": 70.3956,
      "train_samples_per_second": 34.093,
      "train_steps_per_second": 2.841
    }
  ],
  "max_steps": 200,
  "num_train_epochs": 100,
  "total_flos": 255121579477200.0,
  "trial_name": null,
  "trial_params": null
}