levanter
https://github.com/stanford-crfm/levanter
Python
Legible, Scalable, Reproducible Foundation Models with Named Tensors and Jax
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 levanter
Help out
- Issues
- Update transformers requirement from <5.0,>=4.57.1 to >=4.57.1,<6.0
- Support parameter tags in optimizer weight decay
- Tree-of-actors implementation of Ray TPU scheduling
- Add smuggler utilities for JAX transforms
- UpcycleLm script
- Adding PSGD-QUAD to optimizer collection
- Fix: Correct implementation in scion.py
- feat(loss): add pallas kernels for fused cross-entropy loss
- Dataloading should fail if requested Mixture is not feasible
- Fix torch test meshes, run torch tests in CI
- Docs
- Python not yet supported