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 supported26 Subscribers
Add a CodeTriage badge to vllm
Help out
- Issues
- [BUG] Port-allocation race between ApiServer processes in hybrid-LB mode (ZMQError: Address already in use)
- [Bug]: Per-attention-head quantization is currently available only with the Flash Attention backend and requires the calibration pathway provided by llm-compressor.
- [RFC]: Convert Triton kernels from raw pointers to block pointers
- [Bug]: Streaming output incorrectly mapped to `reasoning` field instead of `content` when `enable_thinking=False`
- [Waiting on v4 deprecation] Update fast image processing files and classes names after upgrade to Transformers v5
- [Bug]: Gemma 4 fails to initialize with per-token-head KV cache quantization
- [bugfix] [metrics] [ux] fix sparse index_topk MFU accounting
- [vLLM IR] Port non-silu activations
- [EC Connector] Fix ECExampleConnector load device under TP>1
- [Bug]: `token_capacity_kv_cache_groups` (#40384) should also exclude `SlidingWindowSpec` / `ChunkedLocalAttentionSpec`
- Docs
- Python not yet supported