How the calculate the BiGAN discriminator loss after the training

After the training, I want to feed the generated images into the discriminator. This way seems correct? Please find in those pictures the discriminator architecture and the function to calculate the discriminator loss. What do you think?
Capture d’écran 2021-10-04 à 21.49.31

I want to do that in order to keep the fake images that seems really like the real ones.
Thank you,

Based on your code snippet I guess you are using TensorFlow?
If so, I would recommend to post the question in their discussion board as you would find the experts there. :wink: