Guanhua Zhang

One Simple Fix That Makes LLM Benchmark Rankings Actually Agree

Model A wins on MMLU. Model B wins on ARC-Challenge. Model C wins on HellaSwag.