aitorzip/PyTorch-CycleGAN

about identity loss

karim-moon opened this issue · 1 comments

Hi,
thank you for making this repository and it makes me more easier to learn CycleGAN
But seeing this train code, there is a wandering thing about identity loss

At the paper, there is no identity loss.
So what kind of role does identity loss have in this code?

Thanks

I asked something stupid,
To make sure the coordinate real image and converted image at the latent space, it used real image identity mapping.
I found original repository about identity mapping code
Thanks