API Documentation

Core classes

Implementation of a variational autoencoder in TensorFlow

Dataset(*arrays, batch_size[, shuffle, to_dense]) Class for generating data mini-batches
Model(n_inputs[, n_labels, optimizer, …]) Base abstract class for models
UnsupervisedModel(n_inputs[, n_labels, …]) Base abstract class for unsupervised models
VAE(n_inputs, n_latent[, n_encoder, …]) Variational Autoencoder

Layers

TensorFlow layers

layers.dense(inputs, units[, activation, …]) Dense layer with optional weight normalization

Operations

TensorFlow operations

ops.gaussian_log_likelihood(x[, mean, …]) Compute the log-likelihood for independent Gaussian variables
ops.log_eluplusone(x[, name]) Compute log(elu(x) + 1) in a numerically stable manner
ops.reduce_logmeanexp(x[, axis, keep_dims, name]) Compute log(sum(exp(x))) in a numerically stable manner