The fix is to use some reverse psychology when training a model https://t.co/SsICpN7JlK ...