4 ms·
Fast and Easy Infinitely Wide Networks with Neural Tangents
- dgreensp 7y agoCan someone explain the implications of this for performance, or what we can do now that we couldn’t do before?
- perl4ever 7y agoWell, 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.
- rllin 7y agoIt really looks more and more like JAX is the internal winner after the tf 2.0 fiasco
- miscPerson 7y agoFor the contextless, what was the TF 2.0 fiasco?