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
- [Frontend] Support Llama for structured decisions
- [Bug]: MTP speculative decoding with forced-reject diverges from target-only with confident margins on Qwen3.8-27B (GDN hybrid) at K>=5, n>=8, L>=192
- [Bug]: [ROCm][RDNA4] `RDNAHybridW4A16LinearKernel` is still auto-selected for GPTQ and fails with `zp shape mismatch` (follow-up to #51971)
- [Bugfix][Frontend] Exclude hidden stop-token logprobs
- [Bugfix][Frontend] Respect item order when adding generative scoring special tokens
- [RFC]: Attention Backend Test Suites
- [Core][Mamba] Drop the align-mode prefill chunk split
- [Feature][ROCm] Add MoRI-IO data transport for the EC CPU connector
- [Spec Decode] Add DSpine adjacent-causal-injection drafts
- [ROCm][MI350P] Muse Spark MonoKernel decode assumes 256 co-resident CTAs (MI355X); refused on 128-CU MI350P
- Docs
- Python not yet supported