optimum
https://github.com/huggingface/optimum
Python
🚀 Accelerate training and inference of 🤗 Transformers and 🤗 Diffusers with easy to use hardware optimization tools
Triage Issues!
When you volunteer to triage issues, you'll receive an email each day with a link to an open issue that needs help in this project. You'll also receive instructions on how to triage issues.
Triage Docs!
Receive a documented method or class from your favorite GitHub repos in your inbox every day. If you're really pro, receive undocumented methods or classes and supercharge your commit history.
Python not yet supported2 Subscribers
Add a CodeTriage badge to optimum
Help out
- Issues
- Keep biasless projections biasless through MergeLinears
- Intel Arc GPU via torch_directml — 7.14× speedup on BLAS, missing from optimum coverage
- Fix bias fusion for reused Linear modules
- Fix MergeLinears for root-level linear layers
- MergeLinears fails for linear layers at the graph root
- Preserve input dtype in FuseBiasInLinear
- FuseBiasInLinear fails with float16 and bfloat16 inputs
- Add -v/--version to optimum-cli
- Fix bias arithmetic in Conv2d BatchNorm fusion
- Fix binary integer dummy masks for Decision Transformer
- Docs
- Python not yet supported