Benchmark Data Contamination of Large Language Models: A Survey
This addresses a critical issue for researchers and developers in natural language processing, as it is a survey that synthesizes existing knowledge rather than presenting new incremental findings.
The paper tackles the problem of Benchmark Data Contamination (BDC) in Large Language Models, where models inadvertently incorporate evaluation data during training, leading to unreliable performance assessments, and it reviews mitigation strategies and future directions.
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.