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
- [Perf] [AMD] KDA: Build extend chunk indices from host seq lens (SGLANG_ROCM_NO_EXTEND_SYNC)
- [Fix] routed-experts: Read DP-local rows for every A2A backend
- [Fix] openai: Report the raw byte for SentencePiece byte-fallback pieces in chat logprobs
- [Fix] [AMD] aiter: Reserve the decode workspace in the automatic mem_fraction_static
- [Fix] openai: honor chat template stop strings for input_ids requests
- [Doc] sglang-diffusion guide subclasses ImageSamplingParams, which does not exist (the base class is SamplingParams)
- [CI] [AMD] basic_perf: Relax the MI300X p95 limit for embeddings batch size 10
- [Tracking] Multimodal parity between the Rust and Python frontends
- [Bug] GLM-4.7-Flash ignores ModelOpt FP8 KV-cache scales (self_attn.k_proj.k_scale / v_proj.v_scale)
- [Bug] DeepSeek MLA FP8 KV cache: k/v_scale corrupt decode (ModelOpt names) or are dropped (compressed-tensors)
- Docs
- Python not yet supported