keepsake icon indicating copy to clipboard operation
keepsake copied to clipboard

Published 20 hours ago • verified publisher icon replicate

PyTorch Lightning tests shouldn't hit network

Open bfirsh opened this issue 3 years ago • 4 comments

Currently the PyTorch Lightning tests download mnist and train a real model. We shouldn't do that to just test that the callback works -- they should run a fake training process of some kind.

bfirsh avatar Mar 08 '21 02:03 bfirsh

Maybe, some kind of fake data could be generated with NumPy or PyTorch instead of using a "real" dataset?

gabrielmbmb avatar Mar 08 '21 05:03 gabrielmbmb

Yeah this doesn't need to do anything real.

bfirsh avatar Mar 11 '21 20:03 bfirsh

These files are pretty small. Could just include them in the repo. Ideally we'd have a test that didn't actually do anything on real data though. https://github.com/golbin/TensorFlow-MNIST/tree/master/mnist/data

bfirsh avatar Mar 11 '21 20:03 bfirsh

A pytext fixture with dummy generated train data could be nice, as it could be used in future test cases easily.

gabrielmbmb avatar Mar 11 '21 20:03 gabrielmbmb