About the total parameters #7
Description
Hi, I wonder to ask about the total parameters of this framework. You mentioned in the paper that "We adopt Llama-3.1-8B�Instruct [12] as the base model for text agent, Qwen2-VL-7B-Instruct [38] for other four agents", so if each agent adopts a separate model, it should be 36B. Please let me know, is it correct or not?