search

github-pengge / PyTorch-progressive_growing_of_gans

PyTorch implementation of Progressive Growing of GANs for Improved Quality, Stability, and Variation.
580 stars 97 forks source link

batch sizes don't match paper #4

Open jcpeterson opened 6 years ago

jcpeterson commented 6 years ago

Why are the default batch sizes used? The original paper uses 16 for sizes 4x4 to 128x128, which should be faster (overall) than what is currently used.

github-pengge commented 6 years ago

I did not use the same batch size as the paper used cause I ran the code on 1080 GPU with only 8GB memory, not P100, which has 16GB memory. Besides, the dataset is also different, I used CelebA dataset(with cropped and aligned), and now I'm switching to CelebA-HQ.

jcpeterson commented 6 years ago

I see. I also noticed the learning rate seems a bit fast to use for both the G and D. Any reason why?

github-pengge commented 6 years ago

I found that learning rate was not fixed in official code, and I'm changing the scheduler of learning rate now.