@inproceedings{shi-etal-2024-ulmr,
title = "{ULMR}: Unlearning Large Language Models via Negative Response and Model Parameter Average",
author = "Shi, Shaojie and
Tan, Xiaoyu and
Qiu, Xihe and
Qu, Chao and
Nie, Kexin and
Cheng, Yuan and
Chu, Wei and
Yinghui, Xu and
Qi, Yuan",
editor = "Dernoncourt, Franck and
Preo{\c{t}}iuc-Pietro, Daniel and
Shimorina, Anastasia",
booktitle = "Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing: Industry Track",
month = nov,
year = "2024",
address = "Miami, Florida, US",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.emnlp-industry.57",
pages = "755--762",
abstract = "In recent years, large language models (LLMs) have attracted significant interest from the research community due to their broad applicability in many language-oriented tasks, and are now widely used in numerous areas of production and daily life. One source of the powerful capabilities of LLMs is the massive scale of their pre-training dataset. However, these pre-training datasets contain many outdated, harmful, and personally sensitive information, which inevitably becomes memorized by LLM during the pre-training process. Eliminating this undesirable data is crucial for ensuring the model{'}s safety and enhancing the user experience. However, the cost of extensively cleaning the pre-training dataset and retraining the model from scratch is very high. In this work, we propose ULMR , a unlearning framework for LLMs , which first uses carefully designed prompts to rewrite the instructions in the specified dataset, and generate corresponding negative responses. Subsequently, to ensure that the model does not excessively deviate post-training, we perform model parameter averaging to preserve the performance of the original LLM. We conducted experiments on two public datasets, TOFU and RWKU, demonstrating that our method can effectively forget specified information while retaining the capabilities of the original LLM.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="shi-etal-2024-ulmr">
<titleInfo>
<title>ULMR: Unlearning Large Language Models via Negative Response and Model Parameter Average</title>
</titleInfo>
<name type="personal">
<namePart type="given">Shaojie</namePart>
<namePart type="family">Shi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiaoyu</namePart>
<namePart type="family">Tan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xihe</namePart>
<namePart type="family">Qiu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chao</namePart>
<namePart type="family">Qu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kexin</namePart>
<namePart type="family">Nie</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yuan</namePart>
<namePart type="family">Cheng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wei</namePart>
<namePart type="family">Chu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xu</namePart>
<namePart type="family">Yinghui</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yuan</namePart>
<namePart type="family">Qi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2024-11</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing: Industry Track</title>
</titleInfo>
<name type="personal">
<namePart type="given">Franck</namePart>
<namePart type="family">Dernoncourt</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Daniel</namePart>
<namePart type="family">Preoţiuc-Pietro</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Anastasia</namePart>
<namePart type="family">Shimorina</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Miami, Florida, US</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>In recent years, large language models (LLMs) have attracted significant interest from the research community due to their broad applicability in many language-oriented tasks, and are now widely used in numerous areas of production and daily life. One source of the powerful capabilities of LLMs is the massive scale of their pre-training dataset. However, these pre-training datasets contain many outdated, harmful, and personally sensitive information, which inevitably becomes memorized by LLM during the pre-training process. Eliminating this undesirable data is crucial for ensuring the model’s safety and enhancing the user experience. However, the cost of extensively cleaning the pre-training dataset and retraining the model from scratch is very high. In this work, we propose ULMR , a unlearning framework for LLMs , which first uses carefully designed prompts to rewrite the instructions in the specified dataset, and generate corresponding negative responses. Subsequently, to ensure that the model does not excessively deviate post-training, we perform model parameter averaging to preserve the performance of the original LLM. We conducted experiments on two public datasets, TOFU and RWKU, demonstrating that our method can effectively forget specified information while retaining the capabilities of the original LLM.</abstract>
<identifier type="citekey">shi-etal-2024-ulmr</identifier>
<location>
<url>https://aclanthology.org/2024.emnlp-industry.57</url>
</location>
<part>
<date>2024-11</date>
<extent unit="page">
<start>755</start>
<end>762</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T ULMR: Unlearning Large Language Models via Negative Response and Model Parameter Average
%A Shi, Shaojie
%A Tan, Xiaoyu
%A Qiu, Xihe
%A Qu, Chao
%A Nie, Kexin
%A Cheng, Yuan
%A Chu, Wei
%A Yinghui, Xu
%A Qi, Yuan
%Y Dernoncourt, Franck
%Y Preoţiuc-Pietro, Daniel
%Y Shimorina, Anastasia
%S Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing: Industry Track
%D 2024
%8 November
%I Association for Computational Linguistics
%C Miami, Florida, US
%F shi-etal-2024-ulmr
%X In recent years, large language models (LLMs) have attracted significant interest from the research community due to their broad applicability in many language-oriented tasks, and are now widely used in numerous areas of production and daily life. One source of the powerful capabilities of LLMs is the massive scale of their pre-training dataset. However, these pre-training datasets contain many outdated, harmful, and personally sensitive information, which inevitably becomes memorized by LLM during the pre-training process. Eliminating this undesirable data is crucial for ensuring the model’s safety and enhancing the user experience. However, the cost of extensively cleaning the pre-training dataset and retraining the model from scratch is very high. In this work, we propose ULMR , a unlearning framework for LLMs , which first uses carefully designed prompts to rewrite the instructions in the specified dataset, and generate corresponding negative responses. Subsequently, to ensure that the model does not excessively deviate post-training, we perform model parameter averaging to preserve the performance of the original LLM. We conducted experiments on two public datasets, TOFU and RWKU, demonstrating that our method can effectively forget specified information while retaining the capabilities of the original LLM.
%U https://aclanthology.org/2024.emnlp-industry.57
%P 755-762
Markdown (Informal)
[ULMR: Unlearning Large Language Models via Negative Response and Model Parameter Average](https://aclanthology.org/2024.emnlp-industry.57) (Shi et al., EMNLP 2024)
ACL
- Shaojie Shi, Xiaoyu Tan, Xihe Qiu, Chao Qu, Kexin Nie, Yuan Cheng, Wei Chu, Xu Yinghui, and Yuan Qi. 2024. ULMR: Unlearning Large Language Models via Negative Response and Model Parameter Average. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing: Industry Track, pages 755–762, Miami, Florida, US. Association for Computational Linguistics.