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
- [Refactor] kimi-k3.1: Move the DCP prefill KV gather out of the DSA backend
- [Fix] qwen4_exp: decode the fp8 PLE table on SM80
- [Perf] rotary_embedding: Short-circuit absent modality token scans in M-RoPE get_rope_index
- [Fix] cache: Register UMBP GPU buffers with device metadata
- [Fix] anthropic: Preserve Kimi Messages stream usage
- [Feature] Support SID-based generative retrieval with constrained decoding and optimized beam search
- [Perf] [AMD] diffusion: Mount the Wan FFN linear+GELU site on gfx1250 with a bit-exact GELU
- [Perf] [AMD] diffusion: Let aiter RMSNorm read strided Q/K rows instead of copying them
- [Perf] [AMD] diffusion: Build the ROCm batched-Conv2D VAE input in one Triton pass
- [Perf] [AMD] DeepSeek-V4.1: Exchange only the indexer selection part attention reads under the HIP row shard
- Docs
- Python not yet supported