{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.19220604488011148,
  "eval_steps": 500,
  "global_step": 2500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "ep_loss": 4.7343,
      "epoch": 0.04,
      "learning_rate": 2.5e-05,
      "loss": 11.6218,
      "mlm_loss": 6.8875,
      "step": 500
    },
    {
      "ep_loss": 0.4315,
      "epoch": 0.08,
      "learning_rate": 5e-05,
      "loss": 3.0285,
      "mlm_loss": 2.5969,
      "step": 1000
    },
    {
      "ep_loss": 0.3811,
      "epoch": 0.12,
      "learning_rate": 7.5e-05,
      "loss": 2.2571,
      "mlm_loss": 1.8761,
      "step": 1500
    },
    {
      "ep_loss": 0.37,
      "epoch": 0.15,
      "learning_rate": 0.0001,
      "loss": 1.9695,
      "mlm_loss": 1.5994,
      "step": 2000
    },
    {
      "ep_loss": 0.3813,
      "epoch": 0.19,
      "learning_rate": 0.00012495,
      "loss": 1.5597,
      "mlm_loss": 1.1785,
      "step": 2500
    }
  ],
  "logging_steps": 500,
  "max_steps": 520240,
  "num_train_epochs": 40,
  "save_steps": 500,
  "total_flos": 2.387043179875533e+18,
  "trial_name": null,
  "trial_params": null
}