Interpret GAN loss

Question

I am currently training the standard DCGAN network on my dataset. After 40 epochs, the loss of both generator and discriminator is 45-50. Can someone please explain the reason and possible solution for this?

Without model specifics, we have no way of interpreting the data. It's your model; you own the loss function. What *specifically* does this metric represent? — Prune, Sep 13 '19 at 18:49

score 1 · Answer 1 · answered Sep 13 '19 at 08:53

This interpretation may be added to unsolved problems.

You cannot interpret the loss of generator and discriminator. Since when one improves it will be harder for the other. When generator improves it will be harder for the critic. When critic improves it will be harder for the generator.

The values totally depend on your loss function. You may expect that numbers should be "about the same" over time.

Interpret GAN loss

1 Answers1