MLP#

class MLP(in_dim, out_dim, hidden_units, activation=ReLU(), batch_norm=False, bias=True)[source]#

Standard baseline MLP with no symmetry constraints.

The network defines:

\[\mathbf{f}_{\mathbf{\theta}}: \mathbb{R}^{d_{\mathrm{in}}} \to \mathbb{R}^{d_{\mathrm{out}}}.\]

No equivariance or invariance constraints are imposed.

Constructor of a Multi-Layer Perceptron (MLP) model.

Parameters:

in_dim (int) – Dimension of the input space.
out_dim (int) – Dimension of the output space.
hidden_units (list[int]) – List of number of units in each hidden layer.
activation (Module | list[Module]) – Activation module or list of activation modules.
batch_norm (bool) – Whether to include batch normalization.
bias (bool) – Whether to include a bias term in the linear layers.

forward(input)[source]#

Forward pass of the MLP model.