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
- [Bug] RuntimeError: Failed at /sgl-workspace/sglang/python/sglang/kernels/jit/csrc/gemm/marlin/gptq_marlin_repack.cuh:311: size_n = 8608 is not divisible by tile_n_size = 64
- [CPU] add fused-moe-router-shim kernel
- [HiCache] Unify PP prefetch hit queries and buffer allocation
- [XPU] Enable fused QK-norm + RoPE by default
- fix(spec): assemble --speculative-token-map hot lm_head rows across vocab shards at TP>1
- [MiniMax] Key sparse-decode autotune on real kernel args; skip the one-chunk merge
- [Feature] Support quality choices for Ascend NPU Diffusion
- [Bug] runai_streamer: CPU-path tensors alias the streamer's staging buffers, so loaders that hold a tensor read overwritten bytes
- [AMD] Fix wrong MoE output at small batch with aiter fp8 block-scale (split-K with one K tile per split)
- fix: make FlashInfer NVFP4 speculative decoding graph-safe
- Docs
- Python not yet supported