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
- [Fix] pd: Size decode Mamba spec-state scratch by running batch, not decode extra slots
- [Feat] [XPU] support speculative decoding for GDN
- feat(diffusion): add optional TinyVAE image and video decoding
- [Feat] [CPU] model: Support LFM2.5 models on CPU
- [KDA] Speed up the ReplaySSM fold with narrower tiles and input prefetch
- [Bug][NPU] GPT-OSS FIA sink attention uses incompatible prefill and eager decode paths
- [Fix] [NPU] attention: preserve GPT-OSS FIA sinks in prefill and eager decode
- [Feat][NVIDIA] DeepSeek-V4.1: Engram precompute on GPU (10/N)
- [Feat] [AMD] DeepSeek-V4.1: FP8 unified KV for the ratio-1/2 layers, with DSpark, PD and the UMBP linker
- [ModelOpt] Support explicit Qwen vision and MTP NVFP4 metadata
- Docs
- Python not yet supported