graphorge.gnn_base_model.model.gnn_architectures.build_rnn

build_rnn(input_size, hidden_layer_sizes, output_size, num_layers=None, rnn_cell='GRU', bias=True)[source]

Build multilayer recurrent neural network.

Parameters:

  • input_size (int) – Number of neurons of input layer.

  • hidden_layer_sizes (list[int]) – Number of neurons of hidden layers.

  • output_size (int) – Number of neurons of output layer.

  • num_layers (list[int], default=None) – Number of layers in each RNN module. Defaults to None.

  • rnn_cell (str, default='GRU') – RNN architecture cell.

  • bias (bool, default=True) – Whether to use bias within the RNN cell.

Returns:

rnn – Multilayer recurrent neural network with linear output layer.

Return type:

torch.nn.Sequential