About Dataset Configuration

[SydCS]

Hello. Nice work!
The dataset config when loading datasets like cifar100 and imagenet confuses me a bit:

1. The train val test choice. And dataset file folder names do not correspond to those in the code.
2. The code lists the folders with indices like "imagenet/val/0". However, it is not the case in original cifar100 and imagenet folder.

I suppose that the authors have done some preprocessing steps to the datasets folders. Can you please provide some more detailed dataset config steps? Thanks!