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
- [Feat] [NPU] glm: Support DSA prefill-CP and DSA cache layer-split
- [Fix] metrics: don't publish cache_hit_rate from empty prefill reports
- [Perf] kimi-k3: Shard the MoE front projection and gather quantized activations
- [Fix] spec: Share draft embed/head before target KV pool sizing
- [Fix] moe: Gate expert location per FusedMoE for every draft layer
- [Feat] constrained: Drop structural tag patterns XGrammar cannot compile (opt-in)
- [Bug] Startup OOM with 32 tokenizer workers and auto-selected CUDA VMM transport on multi-node GB300
- [Fix] dcp: Disable NCCL graph buffer registration by default for a2a
- [Spec-XPU] Integrate sycl kernel for build tree and verify tree
- [Feat] hicache: Deduplicate MLA/DSA host KV with broadcast L2 restore
- Docs
- Python not yet supported