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
- [Bug] [AMD] Mixtral-8x7B FP8 static-scale checkpoint loses ~10 pt GSM8K on MI300X; crashes with SGLANG_USE_AITER=0
- [Perf][DSA]: Cut MTP decode overhead with live-length-bounded replay metadata and a greedy-draft argmax exchange
- [Perf] [NVIDIA] qsa: Share compressed-key reads across the four verify rows of a request
- [Perf] [NVIDIA] qwen3.8: Fuse MoE finalize with the shared-expert gate
- [Perf] [NVIDIA] gdn: Make ReplaySSM verify pipeline stages configurable
- [Feat] rust/grpc: Report the node's Rust listeners in api.v1 discovery
- [Fix] sglang-mm: Accept media payloads at byte-budget boundaries
- [Fix] scheduler: keep chunked prefill within allocatable KV under memory pressure
- [Fix] Kimi-K3: tool_choice required/none grammars
- [Fix] grammar: mask every scheduler stop token until the grammar can end
- Docs
- Python not yet supported