Torchvision super(ImageFolder, self).__init__ TypeError

I meet a type error that occurred at the internal function of TorchVision when I try to build ImageNet dataloader. The info is as follows:

test_set = datasets.ImageNet(root=test_data_path, split=‘val’, download=None, transform=transf_test)

File “D:\anaconda3\envs\pytorch-15\lib\site-packages\torchvision\datasets\”, line 57, in init
super(ImageNet, self).init(self.split_folder, **kwargs)

File “D:\anaconda3\envs\pytorch-15\lib\site-packages\torchvision\datasets\”, line 203, in init
super(ImageFolder, self).init(root, loader, IMG_EXTENSIONS, transform=transform, target_transform=target_transform, is_valid_file=is_valid_file)

TypeError: super(type, obj): obj must be an instance or subtype of type

Pytorch version: 1.5, TorchVision version: 0.6, Python: 3.7.7

Anyone knows why? Thanks in advance.