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 supported59 Subscribers
View all SubscribersAdd a CodeTriage badge to vllm
Help out
- Issues
- [ROCm][DSV4.1] Honor the multi-stream GEMM token threshold in the CSA attention path
- [ROCm] Run SiluAndMulWithClamp through AITER's fused clamped SwiGLU
- [Bugfix][Model] Gemma 4: support checkpoints without a vision tower
- [ROCm][DSV4.1] Quantize the shared query once with AITER's fused dual RMSNorm
- [Perf][MiniMax-M3] Port cross-layer index top-k sharing to the NVIDIA path
- [Bugfix][Core] Compare all attention fields in SlidingWindowMLASpec.merge
- [Feature]: Support n > 1 sampling, multimodal processor kwargs, and output_mode: derender in vllm.Inference
- [Bugfix][MoE] Only divide shared expert output in FP16 for models that compensate
- [Perf][Qwen4Exp] Add opt-in SM120 GDN norm/MXFP8 decode fusion
- [ROCm][Perf][MiniMax-M3] Fuse prefill all-reduce + GemmaRMSNorm via FlyDSL QuickReduce
- Docs
- Python not yet supported