trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.2105263157894735,
  "eval_steps": 500,
  "global_step": 10,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.42,
      "learning_rate": 5e-05,
      "loss": 2.3831,
      "step": 1
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.5e-05,
      "loss": 3.273,
      "step": 2
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.5e-05,
      "loss": 2.3154,
      "step": 4
    },
    {
      "epoch": 2.95,
      "learning_rate": 2.5e-05,
      "loss": 1.1125,
      "step": 7
    },
    {
      "epoch": 3.79,
      "learning_rate": 1.5e-05,
      "loss": 0.9851,
      "step": 9
    },
    {
      "epoch": 4.21,
      "learning_rate": 1e-05,
      "loss": 0.3108,
      "step": 10
    }
  ],
  "logging_steps": 500,
  "max_steps": 10,
  "num_train_epochs": 5,
  "save_steps": 32,
  "total_flos": 2.50470614433792e+16,
  "trial_name": null,
  "trial_params": null
}