checkpoint-1500/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.11532362692806689,
  "eval_steps": 500,
  "global_step": 1500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "ep_loss": 4.4238,
      "epoch": 0.04,
      "learning_rate": 2.5e-05,
      "loss": 11.1467,
      "mlm_loss": 6.7229,
      "step": 500
    },
    {
      "ep_loss": 0.4305,
      "epoch": 0.08,
      "learning_rate": 5e-05,
      "loss": 2.9841,
      "mlm_loss": 2.5537,
      "step": 1000
    },
    {
      "ep_loss": 0.3724,
      "epoch": 0.12,
      "learning_rate": 7.5e-05,
      "loss": 2.2371,
      "mlm_loss": 1.8647,
      "step": 1500
    }
  ],
  "logging_steps": 500,
  "max_steps": 520240,
  "num_train_epochs": 40,
  "save_steps": 500,
  "total_flos": 1.4322259079253197e+18,
  "trial_name": null,
  "trial_params": null
}