Do we need to permute dimension in CNN?

vision
2

Hi,

What does permute the dimensions means?
In my shallow view, between conv layer and fc layer, there should be a flatten operation in tensorflow, and i think it is similar in pytorch.