Vector Quantized VAE

This module is the implementation of the Vector Quantized VAE proposed in (https://arxiv.org/abs/1711.00937).

Available samplers

Normalizing flows sampler to come.

GaussianMixtureSampler

Fits a Gaussian Mixture in the Autoencoder’s latent space.

MAFSampler

Fits a Masked Autoregressive Flow in the Autoencoder’s latent space.

IAFSampler

Fits an Inverse Autoregressive Flow in the Autoencoder’s latent space.

class pythae.models.VQVAEConfig(input_dim=None, latent_dim=10, uses_default_encoder=True, uses_default_decoder=True, commitment_loss_factor=0.25, quantization_loss_factor=1.0, num_embeddings=512, use_ema=False, decay=0.99)[source]

Vector Quantized VAE model config config class

Parameters
  • input_dim (tuple) – The input_data dimension.

  • latent_dim (int) – The latent space dimension. Default: None.

  • commitment_loss_factor (float) – The commitment loss factor in the loss. Default: 0.25.

  • quantization_loss_factor – The quantization loss factor in the loss. Default: 1.

  • num_embedding (int) – The number of embedding points. Default: 512

  • use_ema (bool) – Whether to use the Exponential Movng Average Update (EMA). Default: False.

  • decay (float) – The decay to apply in the EMA update. Must be in [0, 1]. Default: 0.99.

class pythae.models.VQVAE(model_config, encoder=None, decoder=None)[source]

Vector Quantized-VAE model.

Parameters
  • model_config (VQVAEConfig) – The Variational Autoencoder configuration setting the main parameters of the model.

  • encoder (BaseEncoder) – An instance of BaseEncoder (inheriting from torch.nn.Module which plays the role of encoder. This argument allows you to use your own neural networks architectures if desired. If None is provided, a simple Multi Layer Preception (https://en.wikipedia.org/wiki/Multilayer_perceptron) is used. Default: None.

  • decoder (BaseDecoder) – An instance of BaseDecoder (inheriting from torch.nn.Module which plays the role of encoder. This argument allows you to use your own neural networks architectures if desired. If None is provided, a simple Multi Layer Preception (https://en.wikipedia.org/wiki/Multilayer_perceptron) is used. Default: None.

Note

For high dimensional data we advice you to provide you own network architectures. With the provided MLP you may end up with a MemoryError.

forward(inputs, **kwargs)[source]

The VAE model

Parameters

inputs (BaseDataset) – The training dataset with labels

Returns

An instance of ModelOutput containing all the relevant parameters

Return type

ModelOutput