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: reject incompatible handshakes without killing decode
- [Fix] dsv4: Turn the V4.1 candidate scheme off when candidate_topk_blocks is 0
- [Feat][Intel] Quantization: Service Initialization support for AutoRound INT4 models on CPU
- [Perf] [NPU] sampling: fuse top-p and top-k verify renorm
- [Bug] SM12x: block-fp8-as-MXFP8 gate (k%32) is looser than b12x dispatch (K%128); K%128!=0 layers regress silently
- [Fix] gateway: cancel pending prefill on early decode failures
- [CI] [NPU] accuracy: Add 8-NPU suite for MiniMax-M2.5 W8A8 4p GPQA
- [Feat] [NVIDIA] moe: Add opt-in cuTile NVFP4 support for MiniMax on SM12x
- [Bug] Qwen-VL image preprocessing OOMs on GPU: 2.87x peak from torch.cat(processed_images) plus a cache that never returns memory
- [Feature] Add pipeline-parallel EAGLE support for PD decode
- Docs
- Python not yet supported