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
- split_between_processes crashes on tuples when apply_padding=True
- Warn when checkpoint RNG state cannot be restored
- Fix split_between_processes on tuples with apply_padding=True
- Restore deleted variables in purge_accelerate_environment
- Unscale gradients once when FSDP clipping gets a partial list
- [Bug] skip_first_batches drops _drop_last, _non_blocking, slice_fn, and torch_device_mesh across checkpoint resumption
- Fix skip_first_batches dropping drop_last, non_blocking, slice_fn, and mesh
- `balanced_low_0` puts the whole model on GPU 0 when `max_memory` skips a device id
- Fix `balanced_low_0` reading `max_memory` by position instead of device id
- chore(deps): bump the actions group with 3 updates
- Docs
- Python not yet supported