[kimjune01]

this kind of benchmark is particularly susceptible to a narrow grader, where the input request is vague and therefore a significant range of SQL output would be valid, yet graded incorrect.