Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

A few questions #42

Open
aravind598 opened this issue Aug 20, 2022 · 1 comment
Open

A few questions #42

aravind598 opened this issue Aug 20, 2022 · 1 comment

Comments

@aravind598
Copy link

aravind598 commented Aug 20, 2022

Hi there, much thanks for your work on MobileStyleGAN I have a few questions to ask.

  1. How long should the training last?
    I have been training on my custom StyleGAN2 Model for 70+ epochs and the original loss seems to be maintaining at around 7. How many epoch was the original MobileNet FFHQ student network trained for?

  2. What are the losses to look out for?

Epoch 2:  18%|| 2546/14000 [59:37<4:28:15,  1.41s/it, loss=10.041, v_num=1234, l1=5.73, l2=4.59, loss_p=2.04, loss_g=7.92, d_reg=0.0189, loss_d=0.00057, kid_val=0.375, loss_val=18.9]

For the losses above what would be their target value roughly speaking and which of the losses should I take note of or is the most important in this case? Also what are the kid_val target for the FFHQ model in the paper?

  1. Does the kid_val was not in the top True result affect training in any way?

I would much appreciate your reply. Thanks

Edit: Apologies for tagging @bes-dev

@RahulBhalley
Copy link

@aravind598 could you please share the pip freeze and python version you're using?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants