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
- [Core][KV Connector] Keep the connector's head order when its KV layout is unsupported
- [KV Offload] Publish block extra keys in self-describing offload events
- [Bugfix][Spec Decode] Restore ShortConv drafter state in MRV2
- [Bug]: --moe-backend b12x on SM121 (DGX Spark): illegal memory access in CUDA-graph capture with padded rows, worker crash on prefill
- [Bug]: Xid 31 MMU fault under load in multi-node Expert-Parallel (EP16) CUDA-graph replay (fused-MoE FP8 reduce_scatterv) — --enforce-eager clean
- [ROCm][Perf] Fold the MXFP4 dense activation quant into SiLU-mul and RMSNorm with AITER
- Migrate mistralai/Mistral-7B-v0.1 model to transformers backend
- [Frontend] Report model_context_window_exceeded on /v1/messages
- [Frontend] Return Anthropic's prompt-too-long error on /v1/messages
- [Frontend] Replace images with a note for text-only models on /v1/messages
- Docs
- Python not yet supported