Get statically binarized MNIST. Code partially taken from https://github.com/altosaar/proximity_vi/blob/master/get_binary_mnist.py

Log of the probability mass of the values x under the logistic distribution with parameters mean and scale. The sum is taken over all dimensi

Returns the smallest size (H, W) of the image with actual size given as input, such that H and W are powers of 2. :param size

log-likelihood for mixture of discretized logistics, assumes the data has been rescaled to [-1,1] interval Code taken from pytorch adapt

One-sample estimation of element-wise KL between two diagonal multivariate normal distributions. Any number of dimensions, broadcasting s

Log of the probability density of the values x untder the Normal distribution with parameters mean and logvar. The sum is taken over all

Returns a sample from Logistic with specified mean and log scale. :param mu_ls: a tensor containing mean and log scale along dim=1,

Pads input x so that its sizes are powers of 2 :param x: :return: Padded tensor

Code taken from pytorch adaptation of original PixelCNN++ tf implementation https://github.com/pclucas14/pixel-cnn-pp