Section 04: "PyTorch Custom Datasets": Compare Our Custom Dataset Class to the original ImageFolder Class- TypeError

[marvelousxmax]

Hello! I am having trouble on video "Compare Our Custom Dataset Class to the original ImageFolder Class" (Video point 03:22). I am receiving this TypeError message after writing the code:

I have rewritten the Custom Dataset Class from scratch and still receiving the error.

Any explanation would be helpful. Thank you.

[sametucuncu]

How did you create the imagefoldercustom class? Did you use Dataset (from torch.utils.data import Dataset)
as parent class?

[marvelousxmax]

[sametucuncu]

Your initialization function has typing error. It should be:
def __init__(self, ... )

[marvelousxmax]

Gotcha. I see the it now. Thank you for the second pair of eyes! I rewrote it and sorted the code.

Now, I got this new error TypeError: cannot unpack non-iterable NoneType object.

Does the line ( `self.classes, self.class_to_idx = find_classes(targ_dir)' ) needs to be created in a new list?

[sametucuncu]

Could you check your train_dir definition?