Investigating How Pre-training Data Leakage Affects Models' Reproduction and Detection Capabilities
Abstract: Large Language Models (LLMs) are trained on massive web-crawled corpora, often containing personal information, copyrighted text, and benchmark datasets.
This inadvertent inclusion in the training dataset, known as data leakage, poses significant risks and could compromise the safety of LLM outputs.
Despite its criticality, existing studies do not examine how leaked instances in the pre-training data influence LLMs' output and detection capabilities.
In this paper, we conduct an experimental survey to elucidate the relationship between data leakage in training datasets and its effects on the generation and detection by LLMs.
Our experiments reveal that LLMs often generate outputs containing leaked information, even when there is little such data in the training dataset.
Moreover, the fewer the leaked instances, the more difficult it becomes to detect such leakage.
Finally, we demonstrate that enhancing leakage detection through few-shot can help mitigate the impact of the leakage rate in the training data on detection performance.
Paper Type: Long
Research Area: Interpretability and Analysis of Models for NLP
Research Area Keywords: Large language models, leaked dataset, pre-training dataset
Contribution Types: Model analysis & interpretability
Languages Studied: English
Submission Number: 1872
Loading