{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 4647,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.32,
      "learning_rate": 1.78480740262535e-05,
      "loss": 0.4305,
      "step": 500
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.5696148052506995e-05,
      "loss": 0.3478,
      "step": 1000
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.3544222078760491e-05,
      "loss": 0.3352,
      "step": 1500
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.139229610501399e-05,
      "loss": 0.2272,
      "step": 2000
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.240370131267486e-06,
      "loss": 0.2143,
      "step": 2500
    },
    {
      "epoch": 1.94,
      "learning_rate": 7.088444157520982e-06,
      "loss": 0.202,
      "step": 3000
    },
    {
      "epoch": 2.26,
      "learning_rate": 4.936518183774478e-06,
      "loss": 0.147,
      "step": 3500
    },
    {
      "epoch": 2.58,
      "learning_rate": 2.7845922100279753e-06,
      "loss": 0.122,
      "step": 4000
    },
    {
      "epoch": 2.91,
      "learning_rate": 6.32666236281472e-07,
      "loss": 0.1166,
      "step": 4500
    },
    {
      "epoch": 3.0,
      "step": 4647,
      "total_flos": 9780232594314240.0,
      "train_loss": 0.23406425937416991,
      "train_runtime": 1839.4116,
      "train_samples_per_second": 80.833,
      "train_steps_per_second": 2.526
    }
  ],
  "max_steps": 4647,
  "num_train_epochs": 3,
  "total_flos": 9780232594314240.0,
  "trial_name": null,
  "trial_params": null
}