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] dflash: Give the draft a private RoPE module
- [Perf] [NVIDIA] deepseek-v4.1: Zero the MegaAttention packed query once per forward, not per layer
- [Fix] rust_server: Keep idle attention-DP ranks from parking on their ring
- [Doc] [AMD] GLM-5.3: Enable MTP on MI355X and add the agentic recipe with the AMD monokernel
- [Feat] [AMD] minimax: gfx950 MoE, router GEMV, block-fp8 GEMM and fp8 producer kernels
- [Feat] [AMD] distributed: JIT quick all-reduce for TP4 on gfx950
- [Refactor] multimodal: [sglang-miles] [MM refactor 3/9] Extract a shared multimodal processor mixin
- [Perf] [AMD] diffusion: Replace the gfx1250 bf16 scale-shift torch fallback with a bit-exact Triton kernel
- [Perf] [AMD] moe: Add tuned Triton fused MoE config for gemma-4-26B-A4B on MI355X
- [Perf] dflash: Add opt-in compact CUDA graph buckets
- Docs
- Python not yet supported