I posted a question, What does np.random.seed(2) do which is somewhat on similar lines with this specific question. Unlike Scikit-learn’s train_test_split way of creating a validation set, fastai doesn’t seem to have a way to do it, or at least that’s how it looks like to me.
But this answer to my previous question got me thinking, is np.random.seed(2) the right way to select a random block of validation set?