Daniel Murfet comments on Some Notes on the mathematics of Toy Autoencoding Problems