Source code for cogdl.layers.sgc_layer

import torch.nn as nn

from cogdl.utils import spmm


[docs]class SGCLayer(nn.Module): def __init__(self, in_features, out_features, order=3): super(SGCLayer, self).__init__() self.in_features = in_features self.out_features = out_features self.order = order self.W = nn.Linear(in_features, out_features)
[docs] def forward(self, graph, x): output = self.W(x) for _ in range(self.order): output = spmm(graph, output) return output