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] [NPU] batch-invariant: Enable PyTorch deterministic algorithms
- [Bug] Unbounded compiled-grammar cache in BaseGrammarBackend leaks host memory with per-request JSON schemas
- [Fix] constrained: bound the compiled-grammar cache (LRU) to stop host-memory growth with per-request schemas
- sglang:cache_hit_rate gauge drops to 0.0 on prefill reports with zero tokens
- [CI] workflows: Skip optional test workflows without opt-in labels
- [Feat] spec: Add a registry for draft attention backends
- [Test] [NPU] cover C128 page size configuration and cross-page inference
- [Bug] PP + EAGLE/MTP: --speculative-use-block-verification / --speculative-use-rejection-sampling crash on first sampled request (PP relay drops draft_probs)
- [Fix] speculative: Reject PP rejection sampling at startup
- [Perf] [Apple] mlx: Bound request map allocation by KV capacity
- Docs
- Python not yet supported