trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 10.0,
  "global_step": 3340,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.5,
      "learning_rate": 4.251497005988024e-05,
      "loss": 3.2334,
      "step": 500
    },
    {
      "epoch": 2.99,
      "learning_rate": 3.502994011976048e-05,
      "loss": 1.5687,
      "step": 1000
    },
    {
      "epoch": 4.49,
      "learning_rate": 2.754491017964072e-05,
      "loss": 0.6862,
      "step": 1500
    },
    {
      "epoch": 5.99,
      "learning_rate": 2.0059880239520957e-05,
      "loss": 0.3808,
      "step": 2000
    },
    {
      "epoch": 7.49,
      "learning_rate": 1.2574850299401197e-05,
      "loss": 0.2231,
      "step": 2500
    },
    {
      "epoch": 8.98,
      "learning_rate": 5.0898203592814375e-06,
      "loss": 0.1795,
      "step": 3000
    },
    {
      "epoch": 10.0,
      "step": 3340,
      "total_flos": 1.2436072056963072e+16,
      "train_loss": 0.9540257048464107,
      "train_runtime": 1339.3762,
      "train_samples_per_second": 24.915,
      "train_steps_per_second": 2.494
    }
  ],
  "max_steps": 3340,
  "num_train_epochs": 10,
  "total_flos": 1.2436072056963072e+16,
  "trial_name": null,
  "trial_params": null
}