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] mini-lb: Fail /health and /health_generate while a backend is down
- fix(kimi-k3): validate native output and propagate terminal failures
- [Feat] disaggregation: Add NIXL CPU-buffer PD support for TDX/CC
- [Feat] [Intel] Use SYCL kernels for mxfp4 indexer on XPU
- [Feat] [NPU] multimodal: Add opt-in DVPP JPEG image decoding
- [Fix] scheduler: Never schedule a late request after an abort-mode pause with multiple tokenizer workers
- [Bug] scheduler: Priority preemption requeues victims in set order, which can differ across TP ranks
- [Fix] memory_pool: Preserve FP4 MHA block scales during PD and CPU retraction
- [Perf] [NVIDIA] sgl-kernel: Build FlashMLA natively for sm_107a
- [Fix] moe: Broadcast scalar block scales correctly with swap_ab on Triton 3.9
- Docs
- Python not yet supported