paper-with-me

홈 › Papers

Pruning as a Defense: Reducing Memorization in Large Language Models

2025-02-18 · Mansi Gupta, Nikhar Waghela, Sarthak Gupta, Shourya Goel, Sanjif Shanmugavelu

Large language models have been shown to memorize significant portions of their training data, which they can reproduce when appropriately prompted. This work investigates the impact of simple pruning techniques on this behavior. Our findings reveal that pruning effectively reduces the extent of memorization in LLMs, demonstrating its potential as a foundational approach for mitigating membership inference attacks.

📄 PDF Abstract BibTeX arXiv:2502.15796

Code (0)

등록된 구현이 없습니다.

Tasks

Memorization

Methods 이 논문이 사용한 방법론

Pruning 설명 없음

Similar Papers 제목 키워드 기반

Private Memorization Editing: Turning Memorization into a Defense to Strengthen Data Privacy in Large Language Models

2025-06-09 · Elena Sofia Ruzzetti, Giancarlo A. Xompero, Davide Venditti, Fabio Massimo Zanzotto

Large Language Models (LLMs) memorize, and thus, among huge amounts of uncontrolled data, may memorize Personally Identifiable Information (PII), which should not be stored and, consequently, not leaked. In this paper, w…

Language ModelingLanguage ModellingMemorization

Data-centric NLP Backdoor Defense from the Lens of Memorization

2024-09-21 · Zhenting Wang, Zhizhi Wang, Mingyu Jin, Mengnan Du 외

Backdoor attack is a severe threat to the trustworthiness of DNN-based language models. In this paper, we first extend the definition of memorization of language models from sample-wise to more fine-grained sentence elem…

Backdoor Attackbackdoor defenseLanguage ModelingLanguage Modelling+2

Preventing Verbatim Memorization in Language Models Gives a False Sense of Privacy

2022-10-31 · Daphne Ippolito, Florian Tramèr, Milad Nasr, Chiyuan Zhang 외

Studying data memorization in neural language models helps us understand the risks (e.g., to privacy or copyright) associated with models regurgitating training data and aids in the development of countermeasures. Many p…

MemorizationOpen-Ended Question AnsweringStyle Transfer

Membership Inference Attacks and Defenses in Neural Network Pruning

2022-02-07 · Xiaoyong Yuan, Lan Zhang

Neural network pruning has been an essential technique to reduce the computation and memory requirements for using deep neural networks for resource-constrained devices. Most existing research focuses primarily on balanc…

Inference AttackMembership Inference AttackMemorizationNetwork Pruning

On the Effect of Pruning on Adversarial Robustness

2021-08-10 · Artur Jordao, Helio Pedrini

Pruning is a well-known mechanism for reducing the computational cost of deep convolutional networks. However, studies have shown the potential of pruning as a form of regularization, which reduces overfitting and improv…

Adversarial Robustness