1 paper
Sang Truong, Yuheng Tu, Michael Hardy +8
Benchmarks are pivotal in driving AI progress, and invalid benchmark questions frequently undermine their reliability. Manually identifying and correcting errors among thousands of…