batch_size

[pp072]

Hi! Where I can change batch_size? I can't compute on gtx 960 4GB with vgg16 model ( !

[fzliu]

If memory is becoming a limitation for you, try reducing the image size (a smaller model like GoogleNet would also work).

With that being said, I'm surprised that 4GB wasn't enough for VGG16. What options are you using to generate the final image?

[pp072]

Now I use resolution 512 and 1024 interations and it's work. But I want to use resolution 1024X1024. (style.py -s s7.jpg -c 3.jpg - m vgg16 -g 0 -l 512)

[fzliu]

If you want a higher-res image, try using a smaller model, or run it on the CPU.