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] [AMD] gpt-oss: Mixed W4A8 MoE on gfx1250, Gluon below a token count and FlyDSL above
- [Bug] [NPU] 图模式下走Fia v2算子,torch_npu.npu_fused_infer_attention_score_v2函数入参错误
- [Fix] multimodal: keep images of input_ids chat requests; locate them by placeholder ID
- [Feat] dcp: Support GQA DCP with HiCache L2
- [Fix] qwen-vl: keep the client's input_ids for image requests on sglang-miles
- [Fix] moe: Reuse one FlashInfer MXFP4 CUTLASS MoE workspace per stream
- [Perf] disaggregation: Send replicated KV heads from one prefill rank
- [Fix] deepseek-v4.1: Fix bounded replay with CP and DSpark
- [Fix] benchmark: Skip malformed custom JSONL records
- [Fix] [AMD] quark: Fix the EP>1 crash in the Quark MXFP4 W4A16 MoE path and restrict it to gfx1151
- Docs
- Python not yet supported