I meant specifically the book, which doe not have any of those things you mentioned.

Also nobody fits neural networks and use variation inference using any priors that aren’t some standard form that makes algorithm easy