trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "global_step": 2605,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.96,
      "learning_rate": 4.040307101727447e-05,
      "loss": 0.1122,
      "step": 500
    },
    {
      "epoch": 1.92,
      "learning_rate": 3.080614203454895e-05,
      "loss": 0.0432,
      "step": 1000
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.1209213051823416e-05,
      "loss": 0.0261,
      "step": 1500
    },
    {
      "epoch": 3.84,
      "learning_rate": 1.161228406909789e-05,
      "loss": 0.0177,
      "step": 2000
    },
    {
      "epoch": 4.8,
      "learning_rate": 2.015355086372361e-06,
      "loss": 0.0121,
      "step": 2500
    },
    {
      "epoch": 5.0,
      "step": 2605,
      "total_flos": 1.087585926764544e+16,
      "train_loss": 0.04096609338002562,
      "train_runtime": 1383.1538,
      "train_samples_per_second": 30.091,
      "train_steps_per_second": 1.883
    }
  ],
  "max_steps": 2605,
  "num_train_epochs": 5,
  "total_flos": 1.087585926764544e+16,
  "trial_name": null,
  "trial_params": null
}