Aug 24, 2017 · 1 min read
Agreed with the point that Pytorch is easy to start with but what about production? If you are going to ship that thing on a chip, you still need to code your model in Tensorflow and then optimize it for inference.
Agreed with the point that Pytorch is easy to start with but what about production? If you are going to ship that thing on a chip, you still need to code your model in Tensorflow and then optimize it for inference.