sparseml icon indicating copy to clipboard operation
sparseml copied to clipboard

Published 20 hours ago • verified publisher icon neuralmagic

Implement label smoothing for CrossEntropy Loss

Open dbogunowicz opened this issue 3 years ago • 2 comments

CrossEntropy losses (Categorical) now can take label smoothing are as parameter. In theory, ready to be reviewed, but for clarity, I will open this PR once the PyTorch 1.10.2 upgrade gets merged to main (dependency).

dbogunowicz avatar Mar 01 '22 18:03 dbogunowicz

@dbogunowicz looks like there's still some conflicts/unintended changes. can you take a look?

bfineran avatar Mar 07 '22 20:03 bfineran

LGTM pending merge conflicts

rahul-tuli avatar Aug 03 '22 14:08 rahul-tuli