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 supported51 Subscribers
View all SubscribersAdd a CodeTriage badge to vllm
Help out
- Issues
- [Bug]: vLLM 0.25.0 crashed
- [ROCm][Bugfix] Return empty UUID string when amdsmi uuid query fails
- Synchronize distributed FlashInfer autotuning and persist cache
- [RFC] TP-wide step-consensus guard with NCCL RAS integration — field-validated, two live deadlock captures
- [Bug]: v0.25.1 int4_per_token_head crashes with SWA models
- [Bugfix]: add spec decode headroom to mrope/xdrope position buffers
- [Perf][Kernel] Fused DSA indexer Top-k kernel (LiteTopk)
- [Performance]: Qwen3-Omni MoE decode 3-6x slower on RTX 5090 (sm_120) — no fused-MoE tuned configs for GeForce Blackwell
- [Kernel] Use fixed BLOCK_SIZE in count_expert_num_tokens to avoid Triton recompiles
- [Perf] Reduce Triton recompiles for multimodal attention ranges
- Docs
- Python not yet supported