-i Location for the dataset
-o Location for parameter checkpoints and samples
-p how many iterations between print statements
-t Every how many epochs to write checkpoint/samples?
-r Restore training from previous model checkpoint?
-k Number of components in the mixture.
-z Dimension of the latent variable x
-w Dimension of the latent variable w
-hd Number of neurons of each fc layer
-hl Number of dense layers in each network
-lr Batch size during training per GPU
-x How many epochs to run in total?
-s Random seed to use