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] rust-server: Answer a full tokenizer queue with 503 instead of blocking intake, take extension metrics off the mutex, release the GIL in shutdown
- [Fix] radix-tree: Clamp SWA tombstone recovery to the protected prefix in the Rust tree core
- [Doc] [AMD] GLM-5.3-Flash: add the MI355X MXFP4 recipe with the AMD monokernel
- [Perf] hicache: End each buffer-only backup chain where the insert walk's previous one ended
- [Fix] sgl-router: Skip the worker handoff for small request bodies
- [Feat] dflash: Run DFLASH on a pipelined PD prefill
- [Fix] scheduler: Observe finished storage prefetches before shortest-prefill-first ranks, and count staged prefixes as cached
- [Feat] rust-server: Serve media the MM processor already holds without a fetch slot, and fail fetches at its deadline
- [Fix] rust-server: Answer and log request rejections as Python does, and refuse generate fields it would drop
- [Perf] tp_worker: load the worker's tokenizer on first use
- Docs
- Python not yet supported