IndoLEM/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 20.0,
  "global_step": 3160,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 3.16,
      "learning_rate": 4.208860759493671e-05,
      "loss": 1.608,
      "step": 500
    },
    {
      "epoch": 6.33,
      "learning_rate": 3.4177215189873416e-05,
      "loss": 0.6735,
      "step": 1000
    },
    {
      "epoch": 9.49,
      "learning_rate": 2.626582278481013e-05,
      "loss": 0.4976,
      "step": 1500
    },
    {
      "epoch": 12.66,
      "learning_rate": 1.8354430379746836e-05,
      "loss": 0.3982,
      "step": 2000
    },
    {
      "epoch": 15.82,
      "learning_rate": 1.0443037974683544e-05,
      "loss": 0.3188,
      "step": 2500
    },
    {
      "epoch": 18.99,
      "learning_rate": 2.531645569620253e-06,
      "loss": 0.2669,
      "step": 3000
    }
  ],
  "max_steps": 3160,
  "num_train_epochs": 20,
  "total_flos": 1.321116402057216e+16,
  "trial_name": null,
  "trial_params": null
}