Trouble with larger image datasets in project testing

[RealTomNook]

I was studying your project, and was able to run your sample compiled code against the sample images provided, and that worked fine. Then I created my own sample dataset of about 10 images and that also worked. But when I tried it with 1000 pictures, it didn't work. I got a lot of weird results that didn't make sense. I'm not sure how to fix it so that it can work with more pictures. From debugging the code, it seems when running this number of images, I get a lot of NaN values for softmax. Was wondering if there's anything that can be tweaked to improve this?