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] [AMD] tilelang: Zero masked KV loads in the BF16 sparse MLA decode kernel
- [Fix] moe: Refresh cached correction bias after in-place weight updates
- [Fix] [AMD] triton-moe: Clamp SwiGLU on HIP
- [Fix] r3: Keep routed experts across retract when opted in
- [Fix] benchmark: handle metadata-only completion chunks
- [sgl-router] Add an opt-in toktier tokenizer backend (session-delta encoding)
- [Fix] sampling: reject out-of-vocab stop_token_ids
- [Fix] sampling: Reject out-of-vocab stop_token_ids in Python and Rust verify
- [Fix] [Intel] qwen4_exp: only import qwen4_decode on CUDA; run the qwen4 PLE tests on XPU
- [Doc] server-args: Sync the --quantization and --kv-cache-dtype choices with the parser
- Docs
- Python not yet supported