> "The Little Book of Deep Learning"

I wouldn't recommend it. My copy is dogeared - but for me as a concise mini-reference for a working statistician who has to work with data science people, but not actually implement anything. It's more of a high-level handwavy "these are the concepts involved".