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
- Add kwargs to optimizer, scheduler and dataloader using function `accelerator().load_state()`
- FSDP2 Unwrap model, still has dtensors as weights
- `max_memory` key can not be `"disk"`
- (fix) remove sampler_is_batch_sampler code in prepare_data_loader(..)
- accelerator.prepare
- Wrong epoch when resuming from checkpoint
- infer_auto_device_map inefficiently allocates GPU memory for models with imbalanced module sizes
- DeepSpeedEngineWrapper.backward() does a bit too much
- Some adjustment for supporting Deepspeed-Ulysses
- Plan to support FSDP2?
- Docs
- Python not yet supported