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] Select MXFP4 MoE backend for MiMo V2.6 on SM120
- [Bug] Leaked sgl_shm_mm_* segments fill /dev/shm -> permanent crashloop (load snapshot SIGBUS / NCCL ENOSPC) after a scheduler crash
- [Fix] Reclaim leaked /dev/shm segments at scheduler startup
- feat: support load balance when retrieving kv from multiple MLA candi…
- [Bug] `input_embeds` requests through `/open_session` crash the scheduler
- [Bug] PD decode scheduler crashes with NotImplementedError when retracting a request under --enable-hisparse
- [Bug] PD retract rebootstrap is not pinned to the bound prefill DP rank unless the client supplied disagg_prefill_dp_rank
- [PD] Pin a retract rebootstrap to the prefill DP rank the receiver is bound to
- [HiSparse] Rebootstrap retracted PD decode requests instead of offloading KV
- [mem_cache][Spec] Size the MHA accepted-token KV move by chunk area, not tile width
- Docs
- Python not yet supported