trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 10.0,
  "global_step": 2640,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.89,
      "learning_rate": 4.053030303030303e-05,
      "loss": 5.7009,
      "step": 500
    },
    {
      "epoch": 3.79,
      "learning_rate": 3.106060606060606e-05,
      "loss": 3.1027,
      "step": 1000
    },
    {
      "epoch": 5.68,
      "learning_rate": 2.1590909090909093e-05,
      "loss": 2.8741,
      "step": 1500
    },
    {
      "epoch": 7.58,
      "learning_rate": 1.2121212121212122e-05,
      "loss": 2.7506,
      "step": 2000
    },
    {
      "epoch": 9.47,
      "learning_rate": 2.651515151515152e-06,
      "loss": 2.6774,
      "step": 2500
    },
    {
      "epoch": 10.0,
      "step": 2640,
      "total_flos": 1.613998356083712e+16,
      "train_loss": 3.3821164564652877,
      "train_runtime": 1158.5519,
      "train_samples_per_second": 22.761,
      "train_steps_per_second": 2.279
    }
  ],
  "max_steps": 2640,
  "num_train_epochs": 10,
  "total_flos": 1.613998356083712e+16,
  "trial_name": null,
  "trial_params": null
}