Can a GAN Produce Different Inception Scores with the Same Dataset and Noise?

Question

If the dataset, shuffle, and noise are all kept the same, is it possible for the same GAN to give different Inception Scores each time?

Answer 1

if you fix the seed (all random factors) you should get similar results. but if you only fix the shuffle and noise, it might be not enough. e.g if you train a GAN, then starting point (initialization) of the GAN is not the same always, so the trained GAN will be different at the end.