$1: 1 for wiki encoding (200D); 2 for credit-card encoding (50D); 3 for credit-card plain data(30D).
$2: 0 for training and testing model; 1 for showing training process by testing metrics, probabilities from discriminator for benign and vandal users, fm_loss, and f1.
e.g. python oc_gan.py 1 0
train the model on wiki(70%), and then evaluate it(30%).