Version of CIFAR to run the example

[Aivan2008]

I cloned the code and tried to run it, but now the issue is - which version of CIFAR10 are you use in example? No data was downloaded automatically by the code. No python neither matlab or C++ version from official cite seems to be suitable for paths presented in the config file. Thank you for your answer.

[MidoAssran]

Hi, we use the default python version (which is the same as torchvision.CIFAR10).

In the config:

• root_path is your datasets directory, where you put all your data
• image_folder is the folder inside datasets, where cifar-10-batches-py exists (cifar-10-batches-py is the folder torchvision.CIFAR10 looks for as well)

Here is an example to fix your issue. First download the python version you mentioned, an unzip, you will get a cifar-10-batches-py directory. Then create the following directory structure: |- datasets/ |-- cifar10-data/ |---- cifar-10-batches-py/

In your config specify: root_path: datasets/ image_folder: cifar10-data

I will update the README and configs to make this more clear.

[jmarrietar]

Hi @MidoAssran Thanks for the update in the README very clear.

I assume the python version was 3.8 since there are some operators as return *[self.dataset.transform(img) for _ in range(self.supervised_views)], target that are invalid in python 3.7 (The one that Google Colab uses)