sglang
https://github.com/sgl-project/sglang
Python
SGLang is a high-performance serving framework for large language models and multimodal models.
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 sglang
Help out
- Issues
- [Perf] srt: Replace torch.nonzero() with torch.where() for index extraction (#9889)
- [Fix] diffusion: keep tied vocab tables on device under layerwise offload
- [Perf] lingbot-video-moe: compute timestep modulation once per sample
- [Fix] [XPU] diffusion: Copy flash-attention inputs under torch.compile
- [Feature] [AMD] RDNA3/RDNA4 (gfx1100/gfx1201) on main: CI and upstreaming plan
- [Fix] [AMD] sgl-kernel: honor an explicit AMDGPU_TARGET over GPU autodetection
- [Perf] hicache: Replace O(k * N) list.pop(0) drain loops with slice deletion in HiCache ack queues
- [Doc] server-args: Document four PD disaggregation flags
- [Feat] [AMD] sgl-kernel: accept gfx1101, gfx1102 and gfx1200 in the ROCm AOT build
- [Fix] [AMD] sgl-kernel: backport wave32 DSA top-k (#41687) to amd_march
- Docs
- Python not yet supported