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
- [Bug]: /v1/systemone returns noise for thinking-only Qwen3 checkpoints
- [Bugfix][Frontend] systemone: refuse thinking-only chat templates
- [Bugfix][Parser] Forward skipped tool spans verbatim in the reasoning pass
- [Bugfix][Kimi-K3] Cover persistent scheduling in vision FA4 warmup
- [DCP] Make multicast/unicast epoch ordering explicit
- [Kernel] CUTLASS blockwise FP8 GEMM for SM89 (Ada) with per-token activation scales
- [Bug]: nvfp4 KV cache silently corrupt under token-major (BLNHC) KV-cache layout — writer/reader stride mismatch
- [Bug]: EAGLE/EAGLE-3 drafter silently uses RoPE base 10000 when the draft config omits rope_theta
- [MRV2][spec-decode] Full-graph replay with >=2 concurrent sequences kills the engine on SM75; logits also diverge from PIECEWISE
- [Bugfix][HiSparse] Order shared host writes with a step counter
- Docs
- Python not yet supported