VBGE issues

[rsrcoderplay]

Hello, author. I have a question, in your code, why do you use two layers of DGCN convolution in the VBGE model, but this part of the explanation is not seen in your paper (or I ignored), I hope you can answer my doubts.