def forward(self, g, node_feat, edge_feat=None):
if self.embed_size > 0:
dgl_warning("The embedding for node feature is used, and input node_feat is ignored, due to the provided embed_size.", norepeat=True)
h = self.embed.weight
else:
h = node_feat
return self.sgc(g, h)
def forward(self, graph, node_feat, edge_feat=None):
if self.embed_size > 0:
dgl_warning(
"The embedding for node feature is used, and input node_feat is ignored, due to the provided embed_size.", norepeat=True)
h = self.embed.weight
else:
h = node_feat
for i in range(self.num_layers):
h = self.conv_list[i](graph, h)
h = self.out_mlp(h)
return h
def forward(self, graph, node_feat, edge_feat = None):
if self.embed_size > 0:
dgl_warning("The embedding for node feature is used, and input node_feat is ignored, due to the provided embed_size.")
h = self.embed.weight
else:
h = node_feat
h = self.dropout(h)
for l, layer in enumerate(self.layers):
h = layer(graph, h, edge_feat)
if l != len(self.layers) - 1:
h = self.activation(h)
h = self.dropout(h)
return h
def forward(self, graph, node_feat, edge_feat=None):
if self.embed_size > 0:
dgl_warning(
"The embedding for node feature is used, and input node_feat is ignored, due to the provided embed_size.",
norepeat=True)
h = self.embed.weight
else:
h = node_feat
for l in range(self.num_layers - 1):
h = self.gat_layers[l](graph, h).flatten(1)
# output projection
logits = self.gat_layers[-1](graph, h).mean(1)
return logits
def forward(self, g, node_feat, edge_feat=None):
if self.embed_size > 0:
dgl_warning(
"The embedding for node feature is used, and input node_feat is ignored, due to the provided embed_size.",
norepeat=True)
h = self.embed.weight
else:
h = node_feat
edge_weight = edge_feat if self.use_edge_weight else None
for l, layer in enumerate(self.layers):
h = layer(g, h, edge_weight=edge_weight)
if l != len(self.layers) - 1:
h = self.act(h)
h = self.dropout(h)
return h