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
- [NPU] Fix fused LoRA ranks and add overlap-loading regression tests
- [NPU] Fix Qwen shared-expert routing and FuseEP accuracy checks
- fix(spec): prefer decode and default EAGLE chunked prefill to 1024 (#1093)
- revise triton tiling
- [DSV4.1] Use CUDA candidate selection in dense prefill
- [RFC] Decouple the SWA sidecar page size from the full-attention page size (DSv4, Hybrid Models)
- [Spec] Add opt-in DSpark Markov candidate pruning
- # [AMD] Qwen3-Next: fused paged-decode attention with output gate on gfx950
- [tracking] thinking models + stop_sequences: five related defects — content loss, misreported stop_reason, enable_thinking ignored (PRs #33151–#33155)
- [MoE] Add FlashInfer SM90 w4a8 cutlass moe
- Docs
- Python not yet supported