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] speculative: Restore DFLASH draft graph replay metadata
- [Perf] [XPU] mrope: Keep M-RoPE decode deltas on device and drop per-layer GDN fills
- [Feat] [NPU] eplb: Enable dynamic expert load balancing on Ascend
- [Fix] olmoe: apply q/k norm over all heads under tensor parallelism
- [Fix] glm5_next: clone RunAI-streamed q_a/kv_a projections before fusing (#37369)
- [Perf] [AMD] dsa: Faster gfx950 fused indexer logits and top-k for MTP verify rows
- [Fix] mooncake: Bind the transfer engine init thread to gpu_id
- [Perf] [AMD] dsa: gfx950 split-K decode plan for 64-head fp8 sparse MLA
- [Feat] dflash: Support Nemotron Bolt E4B BF16 GGUF drafters
- [Fix] sampling: Count the stop_regex window bound in UTF-8 bytes
- Docs
- Python not yet supported