trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 50.0,
  "global_step": 50,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 10.0,
      "learning_rate": 0.00016,
      "loss": 0.1359,
      "step": 10
    },
    {
      "epoch": 20.0,
      "learning_rate": 0.00012,
      "loss": 0.0002,
      "step": 20
    },
    {
      "epoch": 30.0,
      "learning_rate": 8e-05,
      "loss": 0.0001,
      "step": 30
    },
    {
      "epoch": 40.0,
      "learning_rate": 4e-05,
      "loss": 0.0001,
      "step": 40
    },
    {
      "epoch": 50.0,
      "learning_rate": 0.0,
      "loss": 0.0001,
      "step": 50
    },
    {
      "epoch": 50.0,
      "step": 50,
      "total_flos": 6.19947029495808e+16,
      "train_loss": 0.027269574254751204,
      "train_runtime": 19.4285,
      "train_samples_per_second": 41.177,
      "train_steps_per_second": 2.574
    }
  ],
  "max_steps": 50,
  "num_train_epochs": 50,
  "total_flos": 6.19947029495808e+16,
  "trial_name": null,
  "trial_params": null
}