vllm
https://github.com/vllm-project/vllm
Python
A high-throughput and memory-efficient inference and serving engine for LLMs
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 supported62 Subscribers
View all SubscribersAdd a CodeTriage badge to vllm
Help out
- Issues
- [Bug]: GLM-5.3-Flash JIT-compiles DeepGEMM `tf32_hc_prenorm_gemm` during serving (20 mHC split-K variants are never warmed)
- [XPU] Compile passes porting llm-scaler decode optimizations, multi-row decode fixes (QKV/norm/GEMV/MoE router/all-reduce)
- [RFC]: TPSP Fusion Design
- [Docs] Add host kernel tuning guidance for engine-core polling and KV offload
- [Usage]: ROCm gfx950: MXFP4 activation quant not fused into GEMM; Quark INT4 W4A16 uses a very slow Triton kernel
- [Bugfix][Frontend] Honor allowed_tools tool_choice in the Responses API
- [Bug]: Exllama GPTQ and RDNA3 W4A16 kernels silently produce wrong output when `group_size % 32 != 0` (reproduced on CUDA sm_120 and ROCm gfx1100)
- [Bugfix][Structured Output] Flag combinators with sibling constraints…
- [Bug]: FP8 KV cache startup can OOM because CUDA graph memory is omitted from the cache budget
- [ROCm] [Feat] Enable HiSparse on ROCm
- Docs
- Python not yet supported