Milo Q
Results
3
comments of
Milo Q
个人理解，相当于把prefusion模块当成 LLM的前几层，LLM本身当成LLM的后面的层，所以 LLM在后面本身接收到的信息少了也没事
@MonolithFoundation Do you mean Llava-mini or mPLUG-Owl3?
@MonolithFoundation How to explain the speed gain mentioned in these papers, such as Flops ？