trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.7337031900138697,
  "global_step": 2500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.35,
      "learning_rate": 5e-05,
      "loss": 2.0383,
      "step": 500
    },
    {
      "epoch": 0.69,
      "learning_rate": 5e-05,
      "loss": 1.8608,
      "step": 1000
    },
    {
      "epoch": 1.04,
      "learning_rate": 5e-05,
      "loss": 1.7928,
      "step": 1500
    },
    {
      "epoch": 1.39,
      "learning_rate": 5e-05,
      "loss": 1.5415,
      "step": 2000
    },
    {
      "epoch": 1.73,
      "learning_rate": 5e-05,
      "loss": 1.5206,
      "step": 2500
    }
  ],
  "max_steps": 4326,
  "num_train_epochs": 3,
  "total_flos": 3.867114196697088e+16,
  "trial_name": null,
  "trial_params": null
}