What condition is essential for stable training in Generative Adversarial Networks (GANs)?
Answer options
A
Generator loss must always be zero
B
Discriminator must outperform the generator
C
Both networks reaching a balanced learning state
D
Running only the generator
Correct answer: Both networks reaching a balanced learning state
Explanation
Quick AnswerThe correct answer is Both networks reaching a balanced learning state because it directly addresses the core logic of Practice Set 1.
GAN training is stable when the generator and discriminator learn in balance. If one dominates the other, training can collapse or fail.