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] scheduler/MiniMax-M3.1-Flash: Bound a chunked prefill continuation by the allocatable KV
- [Fix] profiler: keep trace files inside output_dir
- [Perf] [NVIDIA] muse-spark: TF32 cluster split-K GEMM for the decode router logits
- [Fix] dflash: Stage the draft sampler only for batches its buffers hold
- [Fix] [AMD] aiter: Plan the DCP asm verify shard with the draft tokens under CUDA graph replay
- [Feat] diffusion/model: Support Efficient-Large-Model/Open-H3-Regenerate-2K (MiniMax-H3 2K draft refiner)
- [Feat] qwen4-exp: Add node-sharded Mooncake Engram storage
- [Perf] [NVIDIA] dspark: Stage draft DP-MoE token counts through persistent buffers
- [Fix] multi-tokenizer: Broadcast a non-owner's abort to the owning worker
- [Fix] dspark: Stop compact verify from counting padding rows as accepted drafts
- Docs
- Python not yet supported