search

minyoungg / pix2latent

Code for: Transforming and Projecting Images into Class-conditional Generative Networks
https://minyoungg.github.io/pix2latent
Apache License 2.0
194 stars 14 forks source link

Fine-tuning the weights of generative model? #4

Closed Twizwei closed 4 years ago

Twizwei commented 4 years ago

Hi there, excellent job! Thanks for providing the code!

I have a question about fine-tuning. In Section 3.5 of the paper, I think there is a regularization term that can also optimize the weights of the original GAN. Where can I find this part in your code? Currently, I just saw the optimization of z,c, and ф. Did I miss something? Please correct me if I was wrong.

Thank you again!

minyoungg commented 4 years ago

There is a fine-tuning procedure in the paper. I did not include the fine-tuning but it should be fairly straightforward to implement. The main meat of the paper is in the optimization and the fine-tuning is just the icing on the cake. If you think it would be helpful to be integrated into the codebase I can add it in my next push.

Twizwei commented 4 years ago

Yes, that would be helpful! Thank you so much!