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
- [Kernel] Tune small-M block-FP8 Triton configs for NVIDIA L20
- [Bugfix] Batch-invariant mode: disable CUDA WNA16 MoE kernel (56887)
- [Bug]: vllm/vllm-openai:v0.30.0-cu129 crashes on import: operator torchvision::nms does not exist
- [KV Cache] Add layer-sharded MLA cache with NCCL broadcast (KVPP)
- [EPLB] Gate expert balancing during weight reload
- [Frontend] Stream continuous usage in /v1/messages via message_delta
- [Feature]: Support Prefix-LM for ROCM_ATTN and ROCM_UNIFIED_ATTN backends
- [Bugfix][MoE] Size the FlashInfer one-sided dispatch for experts that quantize their own inputs
- [RFC][KV Connector] NIXL support for shared in-flight prefix loads
- fix: handle sampling when all tokens are invalid (allowed_token_ids + bad_words contradiction)
- Docs
- Python not yet supported