Is There an Open Research Problem in Mecha Optimisation?

How often are neural networks that we train mess optimizers? Yes, that's a really good question. I have no idea. My guess would be that most of the systems that we train currently don't have this sort of internal structure. There might be some systems which do. We talk a little in the rescoplization paper about some of the sorts of related work that might sort of habts have these sorts of properties. This having good enough at producing models ich are capable of doing really coherent opthonization on and off distribution is an open research problem.