Responses are random, although the plan was to make it so the pure input data sentences would map 1:1 to answers, with the rest being random. I even had this data filled in in the input, but I didn't have the time to fix it (although I realize now it would've only taken a few minutes to fix)