chengshunli

Dear Pathak. I am interested in the image inpainting field. I noticed that many researchers are employing Paris Street-View dataset in their inpainting tasks. It would be thankful if you...

你好！我也遇到了這個問題，您現在搞清楚了嗎？我們可以交流一下嗎

me too! did you solve this problem?

> > Interesting, I would expect positional encoding (possibly a different encoding than a simple linear mesh) would have helped. > > So, this suggests a few possible outcomes (1x1...

> I see! Thanks for your reply! I will try again~ Hi, have you done the experiment as suggested by the author? What is the effect of adding the BN...

> I see! Thanks for your reply! I will try again~ When I was training AOTGAN, this problem also occurred. The iterations increased, but the effect became bad. How about...

@glenn-jocher Thank you for your reply, but doesn’t this picture show the loss changes during the training and verification process? In this picture, the loss decreases relatively flatly. Can it...