Hi Qixun,

Thanks for your great efforts in reimplementing all those models! I have a simple question about the implementation details of MincutPool:

For layers.py (184 - 188) in MinCutPool, we are supposed to get the degree matrix of A, but I feel a bit lost about your implementation, could you provide some clarifications about that?

D_flat = torch.sum(A, dim=-1)
D_eye = torch.eye(D_flat.size(1)).type_as(A).to(device)
D_flat = D_flat.unsqueeze(2).expand(*D_flat.size(), D_flat.size(1))
D = D_eye * D_flat

@quqixun thanks for sharing this super code base just wanted to knw what type of input has to be passed to the graph classification modules since when we download directly we have the following files

should we perform any preprocessing ? thanks in advance