mlp

MLP & its config.

class cneuromax.fitting.deeplearning.litmodule.nnmodule.mlp.MLPConfig(dims='???', p_dropout=0.0)

Bases: object

Holds MLP config values.

Parameters:

• dims (list[int], default: '???') – List of dimensions for each layer.

• p_dropout (float, default: 0.0) – Dropout probability.

class cneuromax.fitting.deeplearning.litmodule.nnmodule.mlp.MLP(config, activation_fn)

Bases: Module

Multi-layer perceptron (MLP).

Allows for a variable number of layers, activation functions, and dropout probability.

Parameters:

• config (MLPConfig)

• activation_fn (Module) – The singular activation function to use in between each layer.

model

The internal torch.nn.Module that holds the MLP’s layers.

Type:

torch.nn.Sequential

forward(x)

Flattens input’s dimensions and passes it through the model.

Note

This MLP is currently only capable of returning 1D torch.Tensor batches.

TODO: Add support for returning 2D+ torch.Tensor batches.

Parameters:

x (Float[Tensor, 'batch_size *d_input']) – The input data batch.

Return type:

Float[Tensor, 'batch_size output_size']

Returns:

The output batch.