trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "global_step": 2374,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.42,
      "learning_rate": 2.3681550126368998e-05,
      "loss": 1.7828,
      "step": 500
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.7363100252737996e-05,
      "loss": 1.2157,
      "step": 1000
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.1044650379106993e-05,
      "loss": 0.9278,
      "step": 1500
    },
    {
      "epoch": 1.68,
      "learning_rate": 4.7262005054759895e-06,
      "loss": 0.7625,
      "step": 2000
    },
    {
      "epoch": 2.0,
      "step": 2374,
      "total_flos": 5580514833610752.0,
      "train_loss": 1.1029183468903025,
      "train_runtime": 1179.7724,
      "train_samples_per_second": 24.137,
      "train_steps_per_second": 2.012
    }
  ],
  "max_steps": 2374,
  "num_train_epochs": 2,
  "total_flos": 5580514833610752.0,
  "trial_name": null,
  "trial_params": null
}