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
- Opt-in request security middleware for the HTTP server (rate limiting, IP lists, payload screening)
- [Feature] Add optional fastapi-guard security middleware
- [Fix] Make Mamba2 prefill batch invariant in deterministic mode
- [Scheduler] Align deterministic flashinfer chunks to the q-tile grid, not the split-KV tile
- [Perf] Apply min_new_tokens penalties with compact stop-token updates
- [Bug] KDA prefill dispatch sends packed one-token batches with FP32 state into BF16-only FlashInfer recurrence
- [Fix] Gemma 4: skip quantization for ignored fused vision qkv/gate_up layers
- [HiCache] Batch direct DMA transfers and overlap safe restores
- fix(metrics): report per-iteration prefill variance
- [DSA] Split k-pool indexer prefill rows across attention-TP ranks
- Docs
- Python not yet supported