MADE

Implementation of the Masked Autoencoder model (MADE) proposed in (https://arxiv.org/abs/1502.03509).

class pythae.models.normalizing_flows.MADEConfig(input_dim=None, output_dim=None, hidden_sizes=<factory>, degrees_ordering='sequential')

This is the MADE model configuration instance.

Parameters

• input_dim (tuple) – The input data dimension. Default: None.

• output_dim (tuple) – The output data dimension. Default: None.

• hidden_sizes (list) – The list of the number of hidden units in the Autoencoder. Default: [128].

• degrees_ordering (str) – The ordering to use for the mask creation. Can be either sequential or random. Default: sequential.

class pythae.models.normalizing_flows.MADE(model_config)

Masked Autoencoder model

Parameters

model_config (MADEConfig) – The MADE model configuration setting the main parameters of the model

forward(x, **kwargs)

The input data is transformed toward the prior

Parameters

inputs (torch.Tensor) – An input tensor

Returns

An instance of ModelOutput containing all the relevant parameters

Return type

ModelOutput