News for hackers. Stuff that matters.

A comment on Keras vs PyTorch. Parent comment

Re:Keras vs PyTorch by alfalfasprout on Thursday June 28, 2018 @10:54AM (#17417328)
Yes, they're pretty ugly TBH. All they've done is provide some decent "canned" estimators but for anything custom you're still using the base tensorflow API. Not to mention feeding in something like numpy arrays > 2GB is a huge pain (their Dataset API doesn't fully work).
[ Parent | on HN ]
Re:Keras vs PyTorch by sandGorgon on Thursday June 28, 2018 @11:19AM (#17417588)
Interesting. So do you recommend Keras+Tf as well, or drop Tensorflow altogether ?
[ Parent | on HN ]
Re:Keras vs PyTorch by alfalfasprout on Friday June 29, 2018 @03:30PM (#17427096)
Keras + MxNet is far better. It's faster, you get multi-gpu out of the box, and it's reaaaally easy to export to ONNX for fast inference elsewhere.
[ Parent | on HN ]