Lost in Benchmarks? Rethinking Large Language Model Benchmarking with Item Response Theory

Zhou, Hongli, Huang, Hui, Zhao, Ziqing · arXiv · 2025