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
- [Bug] GLM-5.3-Flash has no usable DSA attention backend on SM121 (DGX Spark): trtllm SM100-only, tilelang exceeds dynamic smem, triton ROCm-only, flashinfer_sparse_mla gated to GlmMoeDsa archs
- [Bug] FlashInfer autotune cache is discarded every boot under MoE expert parallelism (EP>1): per-rank shapes never agree, so the fused-MoE tactic is re-drawn each start
- [Fix] Move weight version fields from response `metadata` to `sglext`, fixing OpenAI SDK validation
- [AMD] Enable DeepSeek-V4.1 on gfx942
- [NPU] Add complete-group C128 Mooncake prefetch
- [Bug] humming MoE runner picks tuning config with a 4x-inflated shape under EP, and the obvious EP-aware fix regresses serving
- [Feature][weight-cache] Support DeepSeek-V4-Flash MXFP4 IPC loading
- [Feature][DeepSeek-V4.1] Decode graph width is pinned to --context-length; add a decode-phase max_seq_len
- [Quantization] Add and extend H200 [32,32] fp8_w8a8 tuned configs
- [DSA] Add opt-in SM90 LiteTopK fused indexer top-k prefill path
- Docs
- Python not yet supported