/rigl

End-to-end training of sparse deep neural networks with little-to-no performance loss.

Primary LanguagePythonApache License 2.0Apache-2.0

Watchers