spamosaic.train_utils.graph_decode

spamosaic.train_utils.graph_decode(z, edge_index)[source]

Compute edge-wise similarity scores using dot product of node embeddings.

Parameters:

z (torch.Tensor) – Node embeddings of shape [num_nodes, embedding_dim].
edge_index (torch.Tensor) – Edge index tensor with shape [2, num_edges].

Returns:

Edge probabilities computed via sigmoid(dot(z_i, z_j)).

Return type:

torch.Tensor