Thanks for the cool response. Would this use a lot more input token if I’m understanding this correctly because you are stopping the generation after a single fill and then generating again and inputing that for another token?