trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 153.84615384615384,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 38.46,
      "learning_rate": 0.00075,
      "loss": 0.1293,
      "step": 500
    },
    {
      "epoch": 76.92,
      "learning_rate": 0.0005,
      "loss": 0.0576,
      "step": 1000
    },
    {
      "epoch": 115.38,
      "learning_rate": 0.00025,
      "loss": 0.0387,
      "step": 1500
    },
    {
      "epoch": 153.85,
      "learning_rate": 0.0,
      "loss": 0.0315,
      "step": 2000
    }
  ],
  "max_steps": 2000,
  "num_train_epochs": 154,
  "total_flos": 2083470892210176.0,
  "trial_name": null,
  "trial_params": null
}