Issam H. Laradji

pytorch version would probably be even better because then it would support GPU and automatic differentiation.

sounds good, thx.

looks amazing @anshul3899 , we can start testing on some applications, thanks!

Ah it looks like the USPS dataset got deleted from the original servers, but I found this one which I think is the same dataset. https://github.com/mingyuliutw/CoGAN/blob/master/cogan_pytorch/data/uspssample/usps_28x28.pkl

Are you using Pytorch 0.4? if you have Pytorch 0.4 or higher `self.y[index]` will be a tensor instead of an integer, which then allows you to apply the function `.clone()`...

Awesome, did it work after upgrading to 0.4?

Thanks for the interest! are you referring to the plot in Figure 9? I used `PrettyPlots` as uploaded here: https://github.com/IssamLaradji/PrettyPlots

I generated Figure 8 using the `visEmbed(exp_dict)` function in `addons/vis.py`. I generated Figure 3 using `visdom` scatter plot `https://github.com/facebookresearch/visdom` Cheers.

@redhat12345 are you using Python 3? if you are using Python 2 then you get this error you mentioned because it does not use `absolute` import (it uses relative import...

Hii, which python version are you using?