Large Language Models are Inconsistent and Biased Evaluators

arXiv:2405.01724v1 Announce Type: new
Abstract: The zero-shot capability of Large Language Models (LLMs) has enabled highly flexible, reference-free metrics for various tasks, making LLM evaluators common tools in NLP. However, the robustness of these LLM evaluators remains relatively understudied; existing work mainly pursued optimal performance in terms of correlating LLM scores with human expert scores. In this paper, we conduct a series of analyses using the SummEval dataset and confirm that LLMs are biased evaluators as they: (1) exhibit familiarity …

abstract arxiv capability cs.ai cs.cl free however language language models large language large language models llm llm evaluators llms making metrics nlp performance reference robustness tasks terms tools type work zero-shot