trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.760546642899584,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.19,
      "learning_rate": 1.9920000000000002e-05,
      "loss": 1.8071,
      "step": 500
    },
    {
      "epoch": 2.38,
      "learning_rate": 3.9920000000000004e-05,
      "loss": 1.2711,
      "step": 1000
    },
    {
      "epoch": 3.57,
      "learning_rate": 3.951414634146342e-05,
      "loss": 0.9311,
      "step": 1500
    },
    {
      "epoch": 4.76,
      "learning_rate": 3.9026341463414634e-05,
      "loss": 0.6448,
      "step": 2000
    }
  ],
  "max_steps": 42000,
  "num_train_epochs": 100,
  "total_flos": 1.4237038007161651e+17,
  "trial_name": null,
  "trial_params": null
}