Benchmark Data Contamination of Large Language Models: A Survey

TL;DR

This survey reviews the challenge of Benchmark Data Contamination in large language models, discussing its impact on evaluation reliability and exploring alternative assessment methods to improve model evaluation integrity.

Contribution

It provides a comprehensive overview of BDC issues in LLMs and discusses potential solutions and future directions for more reliable evaluation methods.

Findings

BDC affects the accuracy of LLM performance evaluation

Traditional benchmarks are vulnerable to data contamination

Alternative assessment approaches are needed for reliable evaluation

Abstract

The rapid development of Large Language Models (LLMs) like GPT-4, Claude-3, and Gemini has transformed the field of natural language processing. However, it has also resulted in a significant issue known as Benchmark Data Contamination (BDC). This occurs when language models inadvertently incorporate evaluation benchmark information from their training data, leading to inaccurate or unreliable performance during the evaluation phase of the process. This paper reviews the complex challenge of BDC in LLM evaluation and explores alternative assessment methods to mitigate the risks associated with traditional benchmarks. The paper also examines challenges and future directions in mitigating BDC risks, highlighting the complexity of the issue and the need for innovative solutions to ensure the reliability of LLM evaluation in real-world applications.