Possible to extend to VLMs?

[sterzhang]

Hi Authors,

Thanks for your great work! I really enjoy reading your paper and its pretty inspiring. Right now, I have conducted some experiments on the reasoning tasks mentioned in your paper and the results are pretty good!

I am wondering is it possible to extend LatentMAS to visual reasoning tasks as well. For example, can you directly apply a vision language model e.g., Qwen3-VL-8B as the backbone model for image perception and understanding tasks like VLM2-Bench (https://vlm2-bench.github.io/) … that requires model reasoning and collaboration as well?

[jiaruzouu]

Hi @sterzhang ，

Thanks for your interest in our work and the VLM2-Bench you shared with us! We are actively extending LatentMAS to VLMs for more general multimodal tasks like VLM2-Bench. Stay tuned😃!