Cheating Automatic LLM Benchmarks: Null Models Achieve High Win Rates.

Xiaosen Zheng, Tianyu Pang, Chao Du, Qian Liu, Jing Jiang, Min Lin

Browse the full ICLR paper archive.