Ismail Elezi comments

Results 7 comments of


                                            Ismail Elezi

Bad results(Not bad now)

Hi, What are the results do you get after the last changes? Do you get similar results to the paper, or still they are worse?

Why are you padding on the sixth convolutional layer?!

I kind of played with it, but the results seem to be significantly worse than yours (visually at least, the validation error rate seem to be around the same). Which...

Average regression loss and classification loss nan, average regression loss inf when training vgg16 model

Late to the party, but this is a problem with the new version of torchvision 0.5. It is not easy to predict (and at times it happens in places you...

step after training

Sorry for not seeing this before. There is actually nothing fancy about generating images after you have trained the GAN. Simply, load the generator, and feed to it the image...

Reproduce results on CARS dataset.

Can you please try with the pretrained versions (I have included them in the code)? Also, please can you check for the PyTorch version? I had used back then PyTorch...

effect with other augmentation method?

That was something in the to-do list, but we never tried. I think it should improve all the results, but that was not necessarily the main point of the paper.

First cycle difference in result

Hi, Yes, they're randomly sampled with the same seed, so the small differences in the 0th cycles are because of the network optimization (gradient descent). Except for ours where we...