Unlocking Memorization in Large Language Models with Dynamic Soft Prompting

TL;DR

This paper introduces a dynamic soft prompting method for large language models that adaptively estimates memorization, improving accuracy over previous static approaches and revealing privacy risks in NLP tasks.

Contribution

The authors propose a transformer-based generator for prefix-dependent soft prompts, enabling more precise memorization measurement in LLMs compared to prior fixed prompt methods.

Findings

Achieved up to 112.75% improvement in memorization detection for text generation.

Achieved up to 32.26% improvement for code generation.

Demonstrated superior performance over state-of-the-art techniques.

Abstract

Pretrained large language models (LLMs) have revolutionized natural language processing (NLP) tasks such as summarization, question answering, and translation. However, LLMs pose significant security risks due to their tendency to memorize training data, leading to potential privacy breaches and copyright infringement. Accurate measurement of this memorization is essential to evaluate and mitigate these potential risks. However, previous attempts to characterize memorization are constrained by either using prefixes only or by prepending a constant soft prompt to the prefixes, which cannot react to changes in input. To address this challenge, we propose a novel method for estimating LLM memorization using dynamic, prefix-dependent soft prompts. Our approach involves training a transformer-based generator to produce soft prompts that adapt to changes in input, thereby enabling more accurate extraction of memorized data. Our method not only addresses the limitations of previous methods but also demonstrates superior performance in diverse experimental settings compared to state-of-the-art techniques. In particular, our method can achieve the maximum relative improvement of 112.75% and 32.26% over the vanilla baseline in terms of discoverable memorization rate for the text generation task and code generation task respectively.