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 supported13 Subscribers
Add a CodeTriage badge to vllm
Help out
- Issues
- fp4 support for triton unified attention
- [Bug]: tensorizer model loading with tensor parallelism increases memory requirements
- [Bug]: vLLM subprocesses remain alive after parent process/server exits (MCP client exit → vLLM still alive)
- [P/D] Support async transfer for P2P NCCL connector
- [Bug]: DeepSeek-V3.1 gives garbage output
- [DO NOT MERGE] Test intelligent targeting demo
- [Bug]: vLLM merges special XML-like tags into single tokens, causing tokenization mismatch with HuggingFace
- [DisaggEverything] `DisaggregatedRequestManager` aka `Coordinator` [1/N]
- [Performance]: Lite-Profiler Framework for Model Analysis
- [Bug]: Since version 0.9.2 comes with nccl built-in, using PCIE causes sys errors. How to disable nccl in vllm for versions after 0.9.2?
- Docs
- Python not yet supported