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 supported59 Subscribers
View all SubscribersAdd a CodeTriage badge to vllm
Help out
- Issues
- fix(tool_parsers): accept single tool call object in xLAM non-streaming parser
- fix(reasoning): check input_ids directly in Step3ReasoningParser.extract_content_ids
- [Bugfix] Fix fast detokenizer output for adjacent non-special added tokens
- [Bugfix][Rust Frontend] Stream DeepSeek DSML tool-call names and arguments like Python
- [Bug]: Block-FP8 Triton fused_moe_kernel fails to compile with Triton 3.9 since #64760 (`too many indices for tensor of rank 0` at `b_scale[None, :]`)
- [Feature][Rust Frontend] Stream Qwen Coder string arguments
- [Feature][Rust Frontend] Stream Gemma 4 string arguments
- [Bugfix][Model] MiMo-V2 omni: label video grids with the sampled frames' media time
- [Bugfix][MoE] Fix Triton block-FP8 scalar scale indexing
- [Doc][Recipe] Muse Spark Open: DSpark + specialized decode mode; gate Kimi-K3.1 low-latency decode off AMD
- Docs
- Python not yet supported