Reverse instances in a Dataset

I need to make a train_test_split of the dataset (namely, of its train part). I want to split it in such a way so that the instances are not shuffled, but test part is consisted of first test_size elements (instead of last test_size elements, as it is done by default when doing dataset.train_test_split(test_size=test_size, shuffle=False)).

There is certainly a primitive way of doing it with
test_data = dataset.train_test_split(train_size=test_size, shuffle=False)['train']
but since I am writing a public code, it will cause misunderstanding.

To sum up, is there a way of doing it “more beautiful”?

Hi ! Indeed it’s not convenient to see that this is in reverse order compared to scikit-learn

Currently there isn’t a more satisfactory solution, but feel free to to open an issue on github to discuss this behavior and how we can improve it