mlsanityData to 7 Oct 2026

A General Framework to Enhance Fine-tuning-based LLM Unlearning

arXiv 2502.17823 ↗First posted 25 Feb 2025cs.LG cs.CLcited by 12

On the map: LLM safety and security → Machine unlearning

Abstract (arXiv)

Unlearning has been proposed to remove copyrighted and privacy-sensitive data from Large Language Models (LLMs). Existing approaches primarily rely on fine-tuning-based methods, which can be categorized into gradient ascent-based (GA-based) and suppression-based methods. However, they often degrade model utility (the ability to respond to normal prompts). In this work, we aim to develop a general framework that enhances the utility of fine-tuning-based unlearning methods. To achieve this goal, we first investigate the common property between GA-based and suppression-based methods. We unveil that GA-based methods unlearn by distinguishing the target data (i.e., the data to be removed) and suppressing related generations, which is essentially the same strategy employed by suppression-based methods. Inspired by this finding, we introduce Gated Representation UNlearning (GRUN) which has two components: a soft gate function for distinguishing target data and a suppression module using Representation Fine-tuning (ReFT) to adjust representations rather than model parameters. Experiments show that GRUN significantly improves the unlearning and utility. Meanwhile, it is general for fine-tuning-based methods, efficient and promising for sequential unlearning.

Key facts. Cited by 12 papers in the map (data through 7 Oct 2026). Placed in Machine unlearning, one of 12 directions in LLM safety and security. Builds on 33 of its 57 references that are in the map. First posted on arXiv 25 Feb 2025.

Builds on

33 of 57 references.

Closest ancestors · 22

Same direction, newest first

  1. 2024Does Unlearning Truly Unlearn? A Black Box Evaluation of LLM Unlearning Methods18 Nov
  2. 2024Unlearning as multi-task optimization: A normalized gradient difference approach with an adaptive learning rate29 Oct
  3. 2024UnStar: Unlearning with Self-Taught Anti-Sample Reasoning for LLMs22 Oct
  4. 2024Catastrophic Failure of LLM Unlearning via Quantization21 Oct
  5. 2024When Machine Unlearning Meets Retrieval-Augmented Generation (RAG): Keep Secret or Forget Knowledge?20 Oct
  6. 2024LLM Unlearning via Loss Adjustment with Only Forget Data14 Oct
  7. 2024Do Unlearning Methods Remove Information from Language Model Weights?11 Oct
  8. 2024Simplicity Prevails: Rethinking Negative Preference Optimization for LLM Unlearning9 Oct
  9. 2024Position: LLM Unlearning Benchmarks are Weak Measures of Progress3 Oct
  10. 2024On Effects of Steering Latent Representation for Large Language Model Unlearning12 Aug
  11. 2024MUSE: Machine Unlearning Six-Way Evaluation for Language Models8 Jul
  12. 2024Reversing the Forget-Retain Objectives: An Efficient LLM Unlearning Framework from Logit Difference12 Jun
10 more
  1. 2024Large Language Model Unlearning via Embedding-Corrupted Prompts12 Jun
  2. 2024Negative Preference Optimization: From Catastrophic Collapse to Effective Unlearning8 Apr
  3. 2024Guardrail Baselines for Unlearning in LLMs5 Mar
  4. 2024Eight Methods to Evaluate Robust Unlearning in LLMs26 Feb
  5. 2024Towards Safer Large Language Models through Machine Unlearning15 Feb
  6. 2024TOFU: A Task of Fictitious Unlearning for LLMs11 Jan
  7. 2023In-Context Unlearning: Language Models as Few Shot Unlearners11 Oct
  8. 2023Who's Harry Potter? Approximate Unlearning in LLMs3 Oct
  9. 2022Knowledge Unlearning for Mitigating Privacy Risks in Language Models4 Oct
  10. 2021Machine Unlearning of Features and Labels26 Aug

Other references · 11

  1. 2023GPT-4 Technical Reportcited by 18,132
  2. 2024The Llama 3 Herd of Modelscited by 14,192
  3. 2023LLaMA: Open and Efficient Foundation Language Modelscited by 13,958
  4. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 7,916
  5. 2020Measuring Massive Multitask Language Understandingcited by 7,768
  6. 2019Machine Unlearningcited by 1,020
  7. 2023An Empirical Study of Catastrophic Forgetting in Large Language Models During Continual Fine-tuningcited by 622
  8. 2024The WMDP Benchmark: Measuring and Reducing Malicious Use With Unlearningcited by 488
  9. 2023Regulating ChatGPT and other Large Generative AI Modelscited by 102
  10. 2024Copyright Protection in Generative AI: A Technical Perspectivecited by 35
  11. 2020Online Forgetting Process for Linear Regression Modelscited by 21

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 6; 2026: 6.

12 papers.

  1. 2025A Comprehensive Survey in LLM(-Agent) Full Stack Safety: Data, Training and Deploymentcited by 110
  2. 2025SoK: Machine Unlearning for Large Language Modelscited by 9
  3. 2025SUA: Stealthy Multimodal Large Language Model Unlearning Attackcited by 4
  4. 2025Generative Model Unlearning: A Survey through Target Events, Unlearning Operators, and Evaluation Protocolscited by 4
  5. 2025The Erasure Illusion: Stress-Testing the Generalization of LLM Forgetting Evaluationcited by 3
  6. 2025A Survey on Unlearning in Large Language Modelscited by 2
  7. 2026Beyond Forgetting: Representation Misdirection Elicits Controllable Side Behaviors and Capabilitiescited by 1
  8. 2026From Anchors to Supervision: Memory-Graph Guided Corpus-Free Unlearning for Large Language Modelscited by 1
  9. 2026Training data attribution in diffusion models via mirrored unlearning and noise-consistent skewcited by 1
  10. 2026CBD: API-Only LLM Black-Box Unlearning through Controlled Behavioral Divergencecited by 0
  11. 2026Efficient LLM Adversarial Training via Low-Rank Defense and Circuit-Guided Surrogatescited by 0
  12. 2026The Tokens Remember: When Tokenization Bypasses Knowledge Editing and Unlearningcited by 0