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] dp-attention: Take single-node control-plane ports from a free block
- [Fix] cpu: Cap default OpenMP thread binding by the cgroup CPU quota
- [Perf] hicache_nixl: vectorize strided page-offset extraction across v1 and v2 paths
- [Bug] NVFP4 MoE: the Marlin backend applies the gate's global scale to the up projection
- [Fix] marlin: Apply the up projection's own global scale in NVFP4 MoE
- [Feat] sgl-router: Route DeepSeek image chats by prefix cache
- [Perf] radix_cache: Reuse the eviction heap across calls
- [Fix] sgl-router: Honor --cache-pending-prefix-consult-below under reorg routing
- [CI] cuda: uv pip install -e times out waiting for the shared uv editable-build lock
- [Feat] hicache: Allow dynamic storage backends that support DCP
- Docs
- Python not yet supported