def foo(dataset):
return ResGCN(dataset,
hidden,
num_feat_layers,
num_conv_layers,
num_fc_layers,
gfn=False,
collapse=False,
residual=residual,
res_branch=res_branch,
global_pool=global_pool,
dropout=dropout,
edge_norm=edge_norm)
def forward(self, data):
x, edge_index, edge_weight = data.x, data.edge_index, data.edge_attr
#print('input: {} | norm : {}'.format(x, x.norm()))
print(self.conv1)
x = F.relu(self.conv1(x, edge_index, edge_weight))
#x = F.dropout(x, training=self.training)
x = self.fc(x)
#print('output: {} | norm: {}'.format(x, x.norm()))
return F.log_softmax(x, dim=1)
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
data = data.to(device)
model = ResGCN(dataset.num_features, dataset.num_classes, data.edge_index,
data.edge_attr).to(device)
#model = Net().to(device)
print(model)
# optimizer = torch.optim.Adam([
# dict(params=model.conv1.parameters(), weight_decay=5e-4),
# dict(params=model.conv2.parameters(), weight_decay=0)
# ], lr=0.01) # Only perform weight-decay on first convolution.
optimizer = torch.optim.Adam(model.parameters(), lr=0.1)
scheduler = torch.optim.lr_scheduler.ReduceLROnPlateau(optimizer)
def train():
model.train()
optimizer.zero_grad()
loss = F.nll_loss(model(data)[data.train_mask], data.y[data.train_mask])