trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.6,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.4,
      "learning_rate": 8.689839572192514e-05,
      "loss": 5.1589,
      "step": 500
    },
    {
      "epoch": 0.8,
      "learning_rate": 7.352941176470589e-05,
      "loss": 4.2823,
      "step": 1000
    },
    {
      "epoch": 1.2,
      "learning_rate": 6.016042780748663e-05,
      "loss": 3.8577,
      "step": 1500
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.679144385026738e-05,
      "loss": 3.6414,
      "step": 2000
    }
  ],
  "max_steps": 3750,
  "num_train_epochs": 3,
  "total_flos": 2003670839033856.0,
  "trial_name": null,
  "trial_params": null
}