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] Prevent rank divergence in multimodal input broadcast
- [Bugfix] Preserve streaming session ownership on rejected turns
- [Bugfix] Validate NGRAM capacity > max_trie_depth at startup (#36495)
- fix(offloader): pass tie_weights=False in functional_call for tied-param modules under --cpu-offload-gb
- fix(layernorm): keep GemmaRMSNorm weight-loader device-consistent under --cpu-offload-gb
- [Feature] GPU offload target for Qwen3.8-Flash-Next N-gram / PLE embeddings
- [Bugfix] Reject corpus removal during pending async load (#36500)
- fix(hicache): isolate overlapping NIXL path registrations
- Triton attention: wider decode KV tile on ROCm and a 2D-grid kv_indices fill
- fix(hicache): bound NIXL hybrid bounce transfers
- Docs
- Python not yet supported