vllm-project
[New Model]: qwen2-audio
🚀 The feature, motivation and pitch
Do we have a plan to support voice models like qwen-audio?
Alternatives
No response
Additional context
No response
Before submitting a new issue...
- [X] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the documentation page, which can answer lots of frequently asked questions.
@fyabc are you interested in implementing this?
Hi, our team are developing on Qwen2-Audio vllm support, please check this branch, and @faychu will take effort on it.
@fyabc are you interested in implementing this?
Hi, our team are developing on Qwen2-Audio vllm support, please check this branch, and @faychu will take effort on it.
this branch last commit is about two month ago, during that period,vllm has had several updates. Please do some new adaptation work
@fyabc are you interested in implementing this?
Hi, our team are developing on Qwen2-Audio vllm support, please check this branch, and @faychu will take effort on it.
@fyabc Very excitied to see this release! Please feel free to ping me for review when the PR's up.
When do we expect to support qwen2 audio
We expect to support qwen2audio within one month.
@faychu, Hi ,pro I follow your branch and have done some adaptation work with the latest version of VLLM. It worked well in a single audio per request scenario, but I am having trouble in the multi-audio per request scenario since the latest VLLM audio batch seems to have two dimensions: batch and sub-batch. Could you please give me some advice?
the latest version of VL
@faychu, Hi ,pro I follow your branch and have done some adaptation work with the latest version of VLLM. It worked well in a single audio per request scenario, but I am having trouble in the multi-audio per request scenario since the latest VLLM audio batch seems to have two dimensions: batch and sub-batch. Could you please give me some advice?
Can you share your adaptation work? Thanks
It seems qwen2-audio has already been well supported.You can just follow the https://github.com/vllm-project/vllm/pull/9248. above.