I tried to gcn-whiten cifar10 dataset,
and a sample image is like this
my training accuracy is also decreased extremely bad when I use the gcn-whitened dataset created using that script.
my question is then, what does that gcn=55. mean in first place and should it not be like 0.55 instead?
and second of all is this image supposed to be like this when gcn whitened ? it looks like a complete noise-made-up image!