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
- [Chore] [AMD] docker: Build uTLX into the gfx950 rocm10 images
- [Perf] hicache: Publish prefill read-done when offload copies are fenced on the forward stream
- [Perf][DSA]: Capture DSA prefill attention and the NextN draft prefill in the prefill CUDA graph
- [CI] diffusion: Run test_image_preprocessing in the diffusion unit suite
- [Feat] dsa: Extend interleave-CP breakable prefill graphs to GlmMoeDs…
- [Perf] disaggregation: Stop device admission in a decode transfer pass at the first refusal
- [Bug] HiCache Mooncake backend: with DP attention each DP rank mounts the full MOONCAKE_GLOBAL_SEGMENT_SIZE (README says total = global segment size)
- [Fix] minimax-h3: Swap the fc1 halves of the Acc LoRA in build_minimax_h3_pdd_weights
- [Perf] [NVIDIA] qwen3: Reuse fused QK-norm/RoPE kernel for decode
- [Fix] [AMD] attention: size the Lean decode KV tile to the device LDS budget
- Docs
- Python not yet supported