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 supported51 Subscribers
View all SubscribersAdd a CodeTriage badge to vllm
Help out
- Issues
- [Doc] Design docs link to deleted files via permalinks pinned to Nov 2024 commits
- [Bugfix] Exclude DSpark draft KV-cache group from OffloadingConnector lookup poisoning
- [Bug]: xlam and hunyuan_a13b tool parsers drop the arguments of any tool call that follows an empty-argument call (streaming)
- [Bug]: A tool call truncated by max_tokens diverges between streaming and non-streaming, and leaks raw <tool_call> markup into content (non-streaming)
- [Bugfix] Fix Gemma4 engine parser returning bare values as strings
- [Bug]: DFlash/DSpark draft acceptance collapses with automatic prefix caching enabled
- [MoE] Add flashinfer.moe_ep (NCCL-EP) all2all backends: flashinfer_ep_low_latency / flashinfer_ep_high_throughput
- [LoRA] Skip empty-LoRA path in eager mode when no adapter is active
- [Bugfix][Frontend] Anchor empty-args detection to the current tool in xlam and hunyuan_a13b streaming parsers
- [Bugfix] Report finish_reason='length' for tool calls truncated by max_tokens in streaming
- Docs
- Python not yet supported