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 supported43 Subscribers
View all SubscribersAdd a CodeTriage badge to vllm
Help out
- Issues
- [Spec Decode][V1][V2] Warm spec-decode helper kernels at startup
- fix(spec decode): suppress EOS at draft positions in rejection sampler
- [PP][Perf] Consolidate residual tensors at PP boundaries.
- [Perf] Batch Weight Prefetching via cuMemcpyBatchAsync to Reduce Latency
- [Bugfix][Gemma 4] Allow loading fine-tuned E2B/E4B checkpoints
- [Performance]: Llama-Nemotron embedding is slower than Transformers for offline batch-32 pooling after compile-cache warmup
- [Tracking] TurboQuant + Gemma 4 multimodal: 5-gate blocker stack
- [Bug]: macOS arm64 build broken — `cpu_attn_vec.hpp` calls `FP32Vec16(const BF16Vec32&, int)` constructor that doesn't exist on ARM (regression from #39445)
- [Bug]: TurboQuant fails on non-power-of-2 head_dim (Phi-2, MSE-K presets)
- [Attention][TurboQuant] Sparse V tile-skip (opt-in)
- Docs
- Python not yet supported