Suppose you are training a GAN to generate images of handwritten digits

$(0-9)$ using the MNIST dataset. If the generator tends to produce only a few

distinct digits, what is this phenomenon called, and how might you adjust

the model architecture or training procedure to mitigate it$?$