exact match

LLM judges outperform traditional metrics in extractive QA evaluations

Researchers have evaluated the effectiveness of using large language models (LLMs) as judges for extractive question-answering tasks. Their study found that LLM-as-a-judge methods correlate much more strongly with human…