Fast and Easy Infinitely Wide Networks with Neural Tangents

ai.googleblog.com

15 pointsrainboiboi3 comments

Can someone explain the implications of this for performance, or what we can do now that we couldn’t do before?

Well, they seem to be claiming it allows insight or understanding into how the networks work. However, they don't seem to demonstrate that as such.

It really looks more and more like JAX is the internal winner after the tf 2.0 fiasco