What are good methods of determining when the model is giving a definitely wrong answer, and when your not-a-model is grossly overconfident and wrong?
What are good methods of determining when the model is giving a definitely wrong answer, and when your not-a-model is grossly overconfident and wrong?