trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.8860103626943006,
  "global_step": 1500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.3,
      "learning_rate": 1.4818652849740933e-05,
      "loss": 3.672,
      "step": 500
    },
    {
      "epoch": 2.59,
      "learning_rate": 9.637305699481867e-06,
      "loss": 3.1833,
      "step": 1000
    },
    {
      "epoch": 3.89,
      "learning_rate": 4.455958549222798e-06,
      "loss": 2.9613,
      "step": 1500
    }
  ],
  "max_steps": 1930,
  "num_train_epochs": 5,
  "total_flos": 840603004416000.0,
  "trial_name": null,
  "trial_params": null
}