trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 2631,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.57,
      "learning_rate": 4.0497909540098824e-05,
      "loss": 0.2038,
      "step": 500
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.0995819080197646e-05,
      "loss": 0.0882,
      "step": 1000
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.1493728620296465e-05,
      "loss": 0.0401,
      "step": 1500
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.1991638160395288e-05,
      "loss": 0.0309,
      "step": 2000
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.489547700494109e-06,
      "loss": 0.0173,
      "step": 2500
    },
    {
      "epoch": 3.0,
      "step": 2631,
      "total_flos": 1.7600391032419123e+17,
      "train_loss": 0.07335638256863196,
      "train_runtime": 838.7854,
      "train_samples_per_second": 50.187,
      "train_steps_per_second": 3.137
    }
  ],
  "max_steps": 2631,
  "num_train_epochs": 3,
  "start_time": 1660825259.040938,
  "total_flos": 1.7600391032419123e+17,
  "trial_name": null,
  "trial_params": null
}