accelerate
https://github.com/huggingface/accelerate
Python
🚀 A simple way to train and use PyTorch models with multi-GPU, TPU, mixed-precision
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 supported1 Subscribers
Add a CodeTriage badge to accelerate
Help out
- Issues
- Feat: Support dynamic batch size in BatchSamplerShard with even_batches
- fix: pass step parameter in TrackioTracker.log()
- Security: Unsafe torch.load in FSDP2 scaler path (CWE-502)
- [Docs] notebook tutorial for distributed inference (fixes #787)
- Security: Fix unsafe torch.load in FSDP2 scaler path (#3964)
- [Bug] prepare() silently double-wraps models and double-serializes checkpoints when called twice on the same object
- [Bug] TrackioTracker.log() ignores step parameter
- [Bug] RNG states from multiple backends (e.g. CUDA + HPU) are saved but only one is restored on load_state
- Fix grammar, spelling, and consistency issues across docs and examples
- `infer_auto_device_map` does not place submodule buffers on `device_map` when submodule is split
- Docs
- Python not yet supported