422797 (1) [Avatar] Offline
#1
Hi,

First thanks for this interesting book. Well written, easy to use and very useful !

I might be wrong but, when moving from learning building blocks of Tensorflow and Machine Learning / Deep Learning technics, I think one important point might be missing in the book content ( or is yet to come in next chapters!) :

In my own opinion, when moving from "learning/experiments" to "real world projects" (aka large models with big data), we will certainly need to move from one step single server to a bigger cluster with several servers, most probably loaded with one of even many GPU. I believe this could have an impact on how to write code in tensorflow or even how to proceed depending on the algorithm used (and the limits, constraints, best practices, ...).

Having an appendix to get some insights on current GPU offers (cloud or Nvidia hardware) would also be a plus, especially for beginners.

So to summarize, it would also be useful to learn how to scale with Tensorflow.

Any thoughts on this ?