Course project (zero to gna's)

I downloaded a dataset from kaggle for course project.
It doesn’t have different folders for test and train , so I first created a dataset by converetion images into tensors for all images and then split it into training , validation and test using random_split function .
But now I want to apply diffrent transformations for training and validation datasets.
How can I do this ?]
The question’s are:

  1. Can I apply two different transformations for the same dataset while loading it using ImageFolder class?
  2. Or can I first split the dataset in test and val and then apply different transformations for each one of then using any function?

I think I’ve solved this problem by copying the dataset as described in one of the posts on SO