trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.658536585365854,
  "global_step": 19500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.94,
      "learning_rate": 9.063614186526554e-05,
      "loss": 0.9314,
      "step": 5000
    },
    {
      "epoch": 1.88,
      "learning_rate": 8.125351848376806e-05,
      "loss": 0.6435,
      "step": 10000
    },
    {
      "epoch": 2.81,
      "learning_rate": 7.18708951022706e-05,
      "loss": 0.5023,
      "step": 15000
    }
  ],
  "max_steps": 53300,
  "num_train_epochs": 10,
  "total_flos": 1.697286170345472e+17,
  "trial_name": null,
  "trial_params": null
}