trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.471169686985173,
  "global_step": 1500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.82,
      "learning_rate": 4.5881383855024714e-05,
      "loss": 0.3825,
      "step": 500
    },
    {
      "epoch": 1.65,
      "learning_rate": 4.1762767710049426e-05,
      "loss": 0.2088,
      "step": 1000
    },
    {
      "epoch": 2.47,
      "learning_rate": 3.764415156507414e-05,
      "loss": 0.1715,
      "step": 1500
    }
  ],
  "max_steps": 6070,
  "num_train_epochs": 10,
  "total_flos": 2317640762820000.0,
  "trial_name": null,
  "trial_params": null
}