Investigating Data Contamination in Modern Benchmarks for Large Language Models

arXiv:2311.09783v2 Announce Type: replace
Abstract: Recent observations have underscored a disparity between the inflated benchmark scores and the actual performance of LLMs, raising concerns about potential contamination of evaluation benchmarks. This issue is especially critical for closed-source models and certain open-source models where training data transparency is lacking. In this paper we study data contamination by proposing two methods tailored for both open-source and proprietary LLMs. We first introduce a retrieval-based system to explore potential overlaps between evaluation benchmarks and …

abstract arxiv benchmark benchmarks concerns cs.ai cs.cl data data transparency evaluation issue language language models large language large language models llms modern open-source models performance training training data transparency type