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] vision: Run SDPA per sequence instead of with a patch mask
- [Fix] scheduler: Don't idle-sleep while an EPD request waits for encoder embeddings
- [Bug] HiCache on hybrid GDN model: host-tier hits drop to zero once evicted working set exceeds ~3 long prefixes (host mamba pool exhausted while KV stays on host)
- [AMD] Speed up DeepSeek-V4 DSPARK decode on gfx1250
- [Fix] dsv41: Escape literal image placeholder in user text instead of rejecting
- [Chore] deps: Upgrade the CUDA PyTorch stack to 2.15
- [Feat] [NVIDIA] megakernels: Expose hybrid runtime and DSpark verification views
- [Fix] marlin: Prevent scale underflow in NVFP4 fallback on SM80-SM90 (#44998)
- [Fix] speculative: Refresh shared_read_done_event after post-verify KV cache compaction
- [Fix] diffusion: Refuse CUDA JIT builds below SM80
- Docs
- Python not yet supported