The run script is a little easier to deal with in terms of details—just add the necessary variables into the script and run the train method:

#!/usr/bin/env python3
from train import Trainer

HEIGHT  = 28
WIDTH   = 28
CHANNEL = 1
LATENT_SPACE_SIZE = 100
EPOCHS = 50001
BATCH = 32
CHECKPOINT = 500
MODEL_TYPE = -1
trainer = Trainer(height=HEIGHT,
                 width=WIDTH,
                 channels=CHANNEL,
                 latent_size=LATENT_SPACE_SIZE,
                 epochs =EPOCHS,
                 batch=BATCH,
                 checkpoint=CHECKPOINT,
                 model_type=MODEL_TYPE)
                 
trainer.train()

Due to the width, height, and channel being derived from the MNIST data, you might wonder why we define it here. Well, another one of those pesky exercise problems will be to implement the ability for the class to take in different datasets. Given the structure of the code, it should be straightforward to allow the class to take different datasets.