gammagl.models.VGAEModel

class VGAEModel(feature_dim, hidden1_dim, hidden2_dim, drop_rate=0.0, num_layers=2, norm='none', name=None)

Applications of Variational Encoders on Graphs proposed in Variational Graph Auto-Encoders paper.

Parameters:

• feature_dim (int) – input feature dimension.

• hidden1_dim (int) – hidden dimension.

• hidden2_dim (int) – output dimension.

• drop_rate (float, optional) – dropout rate.

• num_layers (int, optional) – number of layers.

• norm (str, optional) – apply the normalizer.

• name (str, optional) – model name.

encode(x, edge_index, edge_weight, num_nodes)

reparameterize(mu, logstd)

forward(x, edge_index, edge_weight, num_nodes)