CuriouslyC 4 hours ago

If you try to optimize for everything you get a model that's good at nothing (or hyper expensive to train and run). Simple economics. There is no free lunch.