trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "global_step": 2950,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.85,
      "learning_rate": 1.6610169491525424e-05,
      "loss": 0.6633,
      "step": 500
    },
    {
      "epoch": 1.69,
      "learning_rate": 1.3227118644067798e-05,
      "loss": 0.5801,
      "step": 1000
    },
    {
      "epoch": 2.54,
      "learning_rate": 9.844067796610171e-06,
      "loss": 0.4258,
      "step": 1500
    },
    {
      "epoch": 3.39,
      "learning_rate": 6.4542372881355944e-06,
      "loss": 0.2802,
      "step": 2000
    },
    {
      "epoch": 4.24,
      "learning_rate": 3.0644067796610173e-06,
      "loss": 0.1836,
      "step": 2500
    },
    {
      "epoch": 5.0,
      "step": 2950,
      "total_flos": 3100434898598400.0,
      "train_loss": 0.38077404086872685,
      "train_runtime": 143.3727,
      "train_samples_per_second": 328.759,
      "train_steps_per_second": 20.576
    }
  ],
  "max_steps": 2950,
  "num_train_epochs": 5,
  "total_flos": 3100434898598400.0,
  "trial_name": null,
  "trial_params": null
}