we do tokenize inputs and outputs. sorry if that is not clear in the post