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
- [Bugfix][Quantization] Set orig_dtype in ModelOpt FP8 channel/block schemes
- [ROCm][CI] Run distributed vision tests in the test_areas Distributed…
- [Bugfix][NIXL] Roll back incomplete remote-agent handshakes
- [Core] Move FlashInfer MLA decode workspaces into WorkspaceManager
- [Core] Reuse one CUDA graph capture stream per device
- [Core] Take the top-k/top-p Triton scratch from WorkspaceManager
- [RFC]: Opt-in return of the last hidden state at sampled positions, per request (Model Runner V2)
- [KV Offload] Give each rank its own CPU region outside the replicated layout
- [Performance]: spec-decode CUDA graph throughput deficit on 0.29.0 under gVisor (reproduces off Modal; fixed in 0.30.0)
- [Bug]: Qwen3.6-35B-A3B model with TP 2 and DP 2 returns gibberish output on Intel B70 cards
- Docs
- Python not yet supported