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
- [HiCache] Align LayerSplit CP shards with unified L3 layer chunks
- [Perf] pd: Publish decode KV destinations without host syncs under overlap
- [Fix] [AMD] quark: Bind the MXFP4 MoE AITER layout on gfx95 and gfx1250 without SGLANG_USE_AITER
- [Fix] [NPU] Correct ModelSlim W4A8 MoE scale-bias TP sharding
- [Feat] disaggregation: Support prefill_complete decode allocation with pipeline-parallel prefill
- [Bug] --log-requests-target <dir> loses request logs when --tokenizer-worker-num > 1 (workers share one hourly-rotated file)
- [Fix] logging: Use per-process request log files to avoid rotation races
- [Fix] kda: Pin prefill autotune configs under deterministic inference
- [Fix] rust-server: Support slash-containing model IDs in model retrieval
- [Fix] dsv4: Use type identity in maybe_copy_inplace assert
- Docs
- Python not yet supported