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] sgl-model-gateway: accepted client sockets don't get TCP_NODELAY, adds ~40 ms to TTFT on some streaming requests
- [Fix] managers: Keep the reserved token budget after auto-truncation
- [Refactor] openai: drop unreachable request-id code in serving base
- [Fix] server_args: revalidate LMCache after model capability resolution
- [Fix] utils: Preserve quoted subprocess arguments
- [Fix] gateway: Disable Nagle on accepted client connections
- [Fix] kimi-k3: move temporal position weights to the embedding device
- [Fix] model_loader: Hash np_cache weight file names instead of using tensor names
- [Perf] DFLASH: Verify top-k sampled batches on the all-gathered top-k support (opt-in)
- [Perf] trtllm_mha: Split full-attention decode KV across CTAs at large batch (opt-in)
- Docs
- Python not yet supported