Training on own train, val and test folders

[jiconi]

I have a dataset of the form
train/

• class1/
• class2/
  val/
  -class1/
  -class2/
  test/
  -class1/
  -class2/

How do I use the classification pipeline script to train this dataset? The current script takes a single folder and split it into 5 folds creating a 1/5 val set on the fly. But, I would like to train this specific set of dataset that already has these three folder splits.