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
- [Feat] deepseek: Support mid-conversation tool loading
- [Bug] ModelOpt NVFP4 fused linear: unwritten parts decide the global scales (Gemma 4 KV-shared layers store only q_proj)
- [Fix] ModelOpt NVFP4: Do not let unwritten fused parts decide the global scales
- [Bug] scheduler_0 crashes with exit code -3 (SIGQUIT) during normal decode operation — no Python exception logged, likely C++ layer crash
- [Perf] dsv4: Size expand_prefill_causally's token block by batch size
- [Fix] [AMD] dsv4: Use the gfx942 FP4 indexer kernels
- [Perf] [NPU] DCP: Optimize DSA Verify preparation and communication
- [Perf] [AMD] fp8: Opt-in per-channel requant of block-FP8 dense linears on gfx950
- [Perf] [AMD] gfx1250: fused all-reduce + add + RMSNorm for custom all-reduce v2
- [WIP] [Feat] [NPU] dspark: Support DeepSeek-V4 tree verification
- Docs
- Python not yet supported