Dear DGL community,
I’m trying to classify two groups of whole graphs. Previously I ran the graph classification with a GCN Classifier:
class GCNClassifier(nn.Module):
def __init__(self, in_dim, hidden_dim, n_classes, hidden_layers, allow_zero_in_degree):
super(GCNClassifier, self).__init__()
self.layers = nn.ModuleList()
# input layers
self.layers.append(GraphConv(in_dim, hidden_dim, allow_zero_in_degree=allow_zero_in_degree))
# hidden layers
for _ in range(1, hidden_layers):
self.layers.append(GraphConv(hidden_dim, hidden_dim, allow_zero_in_degree=allow_zero_in_degree))
self.classify = nn.Linear(hidden_dim, n_classes)
def forward(self, g):
# Use node degree as the initial node feature. For undirected graphs, the in-degree
# is the same as the out_degree.
h = g.in_degrees().view(-1, 1).float()
# Perform graph convolution and activation function.
for i, gnn in enumerate(self.layers):
h = F.relu(gnn(g, h))
g.ndata['h'] = h
# Calculate graph representation by averaging all the node representations.
hg = dgl.mean_nodes(g, 'h')
return self.classify(hg)
Now I would like to utilize node features stored in g.ndata[‘feat’]. I add my node features earlier by running g.ndata['feat'] = torch.tensor(graphs[i, :, :, 1]), where this tensor is a 25x25 feature matrix (25 nodes with 25 features each).
Changing h to h = g.ndata['feat'] leads to an error Expected object of scalar type Double but got scalar type Float for argument #3 'mat2' in call to _th_addmm_out - it’s been a long time since I last used dgl. Do you have any suggestions where my error could be?