pytorch-lightning
https://github.com/pytorchlightning/pytorch-lightning
Python
The lightweight PyTorch wrapper for high-performance AI research. Scale your models, not the boilerplate.
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 supported12 Subscribers
Add a CodeTriage badge to pytorch-lightning
Help out
- Issues
- Mixed precision uses `torch.autocast("cuda")` for custom accelerators and `Accelerator` instances, silently running fp32
- Fix CSVLogger automatic step restarting at 0 after every flush
- Fix `ckpt_path="last"` not finding last checkpoints with a templated `CHECKPOINT_NAME_LAST`
- Add overloads to `Fabric.setup`, `setup_optimizers`, and `setup_dataloaders` for precise return types
- Fix FitLoop leaving _restart_stage in RESTARTED_MID_EPOCH after fast-forwarding an epoch
- fix: AMP autocast device for accelerator instances and custom registered accelerators
- Defer logger flushing to logger policies
- docs: fix CodeFactor badge and 1-bit Adam links
- fix: set the sampler epoch before the training iterator is created on resume
- CI: update transformer engine & 1-bit adam DeepSpeed links in documentation and code comments
- Docs
- Python not yet supported