trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "global_step": 1750,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.43,
      "learning_rate": 3.571428571428572e-05,
      "loss": 0.8911,
      "step": 500
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.1428571428571428e-05,
      "loss": 0.0424,
      "step": 1000
    },
    {
      "epoch": 4.29,
      "learning_rate": 7.142857142857143e-06,
      "loss": 0.0178,
      "step": 1500
    },
    {
      "epoch": 5.0,
      "step": 1750,
      "total_flos": 29865071738880.0,
      "train_loss": 0.2740174835750035,
      "train_runtime": 117.3789,
      "train_samples_per_second": 59.636,
      "train_steps_per_second": 14.909
    }
  ],
  "max_steps": 1750,
  "num_train_epochs": 5,
  "total_flos": 29865071738880.0,
  "trial_name": null,
  "trial_params": null
}