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
- [AMD] Use HIP batched copies for HiCache write-back
- [AMD] Enable staged HiCache write-back for DeepSeek V4
- fix(attention): convert flashinfer base-2 LSE to natural log in chunked-prefix MLA merge
- fix(logprobs): guard tolist on mixed token-ids/top logprobs batches (#34719)
- [Bug][Diffusion] Spectrum CFG parallel skips negative-branch initialization and can reuse forecaster state across requests
- [Bug][Diffusion] TeaCache CFG lifecycle double resets serial CFG and leaks state with CFG parallel
- [AMD] Fix ROCm router GEMM/correction-bias dtype (fp32, not bf16)
- [Fix] Support GLM ModelOpt NVFP4 with Humming on Hopper
- docs: add MatrixHub as a model-loading feature
- MXFP4 (E2M1+E8M0) KV cache for Ampere sm86: quantization + fused decode kernels [WIP]
- Docs
- Python not yet supported