Is there a way to split dataset in Specific range?

I have a dataset and i add some data that i dont want to let them split into test data
how should i do

Maybe something like this ?

splits = ds.select(range(1000)).train_test_split()