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] [AMD] dsv4: Read preshuffled indexer keys in TileLang
- [Fix] model_runner: Cast input_embeds to the model dtype
- [Bug] Kimi-K3.1-Pro: PP output relay stalls while pickling nested tensor metadata
- [Perf] [NPU] attention: reuse eager sparse KV lengths across layers
- [Perf] glm: Bound Triton variants on decode replay
- [Perf] [NVIDIA] kda: Skip the row-streaming packed decode kernel on SM90 and SM10x
- [Perf] [AMD] extend-attention: Use 8 warps for head_dim 256 prefill on RDNA3.5 (gfx115x)
- [Perf] [NVIDIA] attention: Skip unused V4 FlashMLA schedule preparation
- [Feat] linker: Expose external cache linker lookup, load, offload and wait metrics
- [Bug] macOS: /v1/loads returns no ranks because load snapshots use /dev/shm
- Docs
- Python not yet supported