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 supported31 Subscribers
View all SubscribersAdd a CodeTriage badge to vllm
Help out
- Issues
- [Feature] Support Anthropic search_result content blocks
- [kv_offload] Use Hugepages for kv_offload tiering setup
- [Bug]: CUDA invalid argument in FlashInfer AllReduce norm fusion on MiniMax-M3 MXFP8, 4x H200
- [Core] Release CUDA graphs before sleep() unmap, re-capture on wake
- [V1][PP] Reach cudagraph-mode consensus across PP ranks (fix #45094 split-brain wedge)
- [Bugfix] Skip torch-compile decoration for DiffusionGemma transformers fallback
- [Perf] Bound DiffusionGemma sampler transient via request-tiled logits
- [Bugfix] Cap fastapi < 0.137 to avoid prometheus-fastapi-in strumentator crash on serve startup
- Relax routed_experts capture KV-connector check to a warning for P/D
- [Bugfix] Correct prompt lengths for timed_traces benchmark
- Docs
- Python not yet supported