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 supported62 Subscribers
View all SubscribersAdd a CodeTriage badge to vllm
Help out
- Issues
- [RFC]: NixlConnector: convert standard KV pages to ROCm AITER shuffled pages on the decode side
- [Feature]: Add lifecycle management hooks for the worker extension class (--worker-extension-cls)
- [ROCm][Bug] mimo_audio.py imports CUDA-only vllm.vllm_flash_attn, killing the engine core on audio input
- [Bugfix][Distributed] Guard cross-node symmetric-memory rendezvous
- [Bug]: Mistral tokenizer rejects or merges OpenAI-valid tool call IDs (truncate_tool_call_ids keeps the last 9 chars)
- [Bug]: MoRIIO WRITE: after the producer dies, decode requests hang until client timeout and KV usage stays elevated
- [Frontend][Core] Add unified readiness checks
- [RFC]: [Spec decode] KV-cache pool collapse when a draft model introduces a new cache spec type (hybrid GDN + DFlash): diagnosis, mitigation dead-ends, and a working dedicated-draft-pool POC
- [Metrics] Export the prefill token backlog as vllm:num_queued_tokens
- [Bugfix][Frontend] Reject chat turns the template silently drops
- Docs
- Python not yet supported