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(utils): add max reduction type
- fix(tracker): guard init_trackers and log against None kwargs
- [WIP] Make accelerate work end-to-end on AMD ROCm
- fix train deepseek V4 with fsdp2: AttributeError: 'Tensor' object has no attribute 'device_mesh'
- fix(checkpointing): save/restore SeedableRandomSampler for map-style datasets
- fix(modeling): include named_buffers in module split expansion for infer_auto_device_map
- Feature request: disk_offload() that mmaps HF-cache safetensors directly (no duplicate copy, no CPU-RAM materialization)
- fix: compile inner model before DDP wrapping to prevent Dynamo tracing DDP internals
- Unable to infer valid auto device map for gemma-4-E4B-it
- [Checkpointing] Persist DataLoaderShard epoch counter across save/load_state
- Docs
- Python not yet supported