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
- [Perf] unified-memory: Clear Mamba slots without a host sync
- [Perf] speculative: Sample the multi-layer EAGLE draft proposal inside per-step draft-extend graphs
- [Perf] unified-memory: Stage a fused draft's draft-extend metadata once per draft cycle
- [Perf] moe: Defer finalize for MXFP8 experts on the routed TRT-LLM runner
- [Fix] function_call: stop dropping text before a partial tool-call marker
- [Bug] multimodal_gen: cross-node pure-Ulysses H3 serve (1 GPU/node x 4 nodes) freezes at the first distributed generation - /health never binds with --warmup-mode off, 0% GPU on all ranks, zero errors
- [Fix] Kimi-K3: Reject flashinfer_mxfp4 SiTU experts outside the SM100 family
- [CI] [Intel] ci/xpu: Rename rolling Docker tag and add release-cut results CSV
- [Feat] disagg: Write PD decode KV to the linker and skip pages the prefill already stored
- [Fix] [AMD] deterministic: Batch-invariant MiniMax sparse-attention decode
- Docs
- Python not yet supported