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
- [Bugfix][NIXL] Preserve block geometry for packed MLA caches
- [Usage]: Significant TTFT degradation and poor QPS scaling with Rust frontend on vLLM 0.25.1 (Qwen122B-A10B, avg input 44k tokens)
- [Feature]: Emit BlockInactive KV events when GPU block ref_cnt drops to zero
- [Bug]: Shared prompt_token_ids causes stale M-RoPE positions
- [RFC]: Dual Batch Overlap (DBO) for Model Runner V2
- [Bug]: runai_streamer multi-node: a separate object-storage tokenizer URI is never pulled on peer nodes
- [Bugfix] Pull a separate Run:ai tokenizer on every node
- [Bugfix] Refresh stale M-RoPE state when streaming re-feeds grow a RUNNING request's prompt
- [Bug]: DiffusionGemma: runtime CUDA OOM under concurrent decode (fp32 canvas×vocab temporaries)
- [Bug]: int8_per_token_head KV + prefix caching corrupts output when the KV pool is pinned at 100% (Gemma-4 hybrid, Triton)
- Docs
- Python not yet supported