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] mem-cache: support GLM DSA cache layer-split on Ascend
- [Perf] gdn: read strided q/k/v in the Triton chunked prefill and skip the dead output zero fill
- [Chore] [AMD] wip: triton3.9 for rocm10/10.1 gfx1250
- [Fix] sampler: Avoid a host read in the torch top-k / top-p / min-p sampler
- [Test] Add CPU unit coverage for MTP weight filtering
- [CI] [AMD] ci: Name AMD suites, jobs and runners as CUDA does
- [Chore] sgl-router: decouple preprocessing from app state
- [Feat] embedding: Accept chat-style messages on /v1/embeddings
- [Chore] sgl-router: move tokenizer prompt logic out of policies
- [Feat] speculative: Build adaptive spec states within an opt-in memory budget
- Docs
- Python not yet supported