graphorge.gnn_base_model.model.gnn_architectures.build_fnn

build_fnn(input_size, output_size, output_activation=Identity(), hidden_layer_sizes=[], hidden_activation=Identity())

Build multilayer feed-forward neural network.

Parameters:

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

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

• output_activation (torch.nn.Module, default=torch.nn.Identity) – Output unit activation function. Defaults to identity (linear) unit activation function.

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

• hidden_activation (torch.nn.Module, default=torch.nn.Identity) – Hidden unit activation function. Defaults to identity (linear) unit activation function.

Returns:

fnn – Multilayer feed-forward neural network.

Return type:

torch.nn.Sequential