How Reliable Are Automatic Evaluation Methods for Instruction-Tuned LLMs?

arXiv:2402.10770v1 Announce Type: new
Abstract: Work on instruction-tuned Large Language Models (LLMs) has used automatic methods based on text overlap and LLM judgments as cost-effective alternatives to human evaluation. In this paper, we study the reliability of such methods across a broad range of tasks and in a cross-lingual setting. In contrast to previous findings, we observe considerable variability in correlations between automatic methods and human evaluators when scores are differentiated by task type. Specifically, the widely-used ROUGE-L metric strongly …

abstract arxiv cost cross-lingual cs.ai cs.cl evaluation human instruction-tuned language language models large language large language models llm llms paper reliability study tasks text type work