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] unlimited-ocr-v2: Add Unlimited-OCR v2 model support
- [Bug] DeepSeek-V4 scheduler crash when multimodal offsets become None under concurrent aborted streams
- [Fix] multimodal: Reject malformed offsets before scheduler execution
- [Fix] penaltylib: Replay retained output tokens after retraction
- docs(amd): document APU memory model and mamba state cache on gfx1151
- [Fix] dflash: Gather candidate top-k over the lm_head's group under DP attention
- [Fix] cuda-graph: Wait for host-cache loads before replaying a captured linear-attention extend
- [Fix] attention: Skip the MSA-v1 QMUL4 prefill case when QMUL4 is unavailable
- [Fix] rust-server: Admit a multimodal model's generations only once its MM processor is installed
- [Fix] rust-server: Let an HTTP layer start a native /generate request's timing
- Docs
- Python not yet supported