Releases: Chris-hughes10/pytorch-accelerated
Releases · Chris-hughes10/pytorch-accelerated
v0.1.49
v0.1.48
v0.1.47
v0.1.46
What's Changed
- Change process decorators to PartialState by @bepuca in #50
- Update requirements.txt by @Chris-hughes10 in #51
- Improve reporting of issues with batch size on distributed evaluate by @bepuca in #53
- Fix remove_padding for higher dimensionality tensors by @bepuca in #52
Full Changelog: v0.1.45...v0.1.46
v0.1.45
v0.1.44
Full Changelog: v0.1.43...v0.1.44
v0.1.43
Minor updates to make an edge case error more readable
v0.1.42
What's Changed
- Bump accelerate version to 1.17 by @Chris-hughes10 in #48
Full Changelog: v0.1.41...v0.1.42
v0.1.41
What's Changed
- Update examples format by @Chris-hughes10 in #42
- Refactor loss tracking by @Chris-hughes10 in #44
- bump accelerate version by @Chris-hughes10 in #45
- Enables distributed evaluation on uneven inputs by @Chris-hughes10 in #46
Full Changelog: v0.1.40...v0.1.41
v0.1.40
What's Changed
- Add option to execute callbacks during ModelEma evaluation loop by @Chris-hughes10 in #41
Full Changelog: v0.1.39...v0.1.40