CAP: Data Contamination Detection via Consistency Amplification

Yi Zhao, Jing Li, Linyi Yang

Large language models (LLMs) are widely used, but concerns about data contamination challenge the reliability of LLM evaluations. Existing contamination detection methods are often…