trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0568031704095113,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.26,
      "learning_rate": 4.399471534950757e-05,
      "loss": 1.9949,
      "step": 500
    },
    {
      "epoch": 0.53,
      "learning_rate": 3.7989430699015134e-05,
      "loss": 2.021,
      "step": 1000
    },
    {
      "epoch": 0.79,
      "learning_rate": 3.199615661782369e-05,
      "loss": 2.0538,
      "step": 1500
    },
    {
      "epoch": 1.06,
      "learning_rate": 2.5990871967331253e-05,
      "loss": 2.0468,
      "step": 2000
    }
  ],
  "max_steps": 4163,
  "num_train_epochs": 3,
  "total_flos": 1.51211470749696e+17,
  "trial_name": null,
  "trial_params": null
}