mlsanityData to 7 Oct 2026

Unlearning as multi-task optimization: A normalized gradient difference approach with an adaptive learning rate

arXiv 2410.22086 ↗First posted 29 Oct 2024cs.LG cs.CLcited by 29

On the map: LLM safety and security → Machine unlearning

Abstract (arXiv)

Machine unlearning has been used to remove unwanted knowledge acquired by large language models (LLMs). In this paper, we examine machine unlearning from an optimization perspective, framing it as a regularized multi-task optimization problem, where one task optimizes a forgetting objective and another optimizes the model performance. In particular, we introduce a normalized gradient difference (NGDiff) algorithm, enabling us to have better control over the trade-off between the objectives, while integrating a new, automatic learning rate scheduler. We provide a theoretical analysis and empirically demonstrate the superior performance of NGDiff among state-of-the-art unlearning methods on the TOFU and MUSE datasets while exhibiting stable training.

Key facts. Cited by 29 papers in the map (data through 7 Oct 2026). Placed in Machine unlearning, one of 12 directions in LLM safety and security. Builds on 41 of its 62 references that are in the map. First posted on arXiv 29 Oct 2024.

Builds on

41 of 62 references.

Closest ancestors · 12

Same direction, newest first

  1. 2024MUSE: Machine Unlearning Six-Way Evaluation for Language Models8 Jul
  2. 2024Negative Preference Optimization: From Catastrophic Collapse to Effective Unlearning8 Apr
  3. 2024Rethinking Machine Unlearning for Large Language Models13 Feb
  4. 2024Machine Unlearning in Large Language Models3 Feb
  5. 2024TOFU: A Task of Fictitious Unlearning for LLMs11 Jan
  6. 2023Unlearn What You Want to Forget: Efficient Unlearning for LLMs31 Oct
  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. 2022Continual Learning and Private Unlearning24 Mar
  11. 2021Adaptive Machine Unlearning8 Jun
  12. 2021Machine Unlearning via Algorithmic Stability25 Feb

Other references · 29

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2023GPT-4 Technical Reportcited by 18,132
  3. 2022Training language models to follow instructions with human feedbackcited by 15,712
  4. 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,084
  5. 2023Llama 2: Open Foundation and Fine-Tuned Chat Modelscited by 12,865
  6. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 7,916
  7. 2023Mistral 7Bcited by 3,065
  8. 2021Convex optimizationcited by 2,479
  9. 2020Extracting Training Data from Large Language Modelscited by 2,144
  10. 2020Gradient Surgery for Multi-Task Learningcited by 1,210
  11. 2019Machine Unlearningcited by 1,020
  12. 2017GradNorm: Gradient Normalization for Adaptive Loss Balancing in Deep Multitask Networkscited by 816
17 more
  1. 2023The RefinedWeb Dataset for Falcon LLM: Outperforming Curated Corpora with Web Data, and Web Data Onlycited by 755
  2. 2023Textbooks Are All You Need II: phi-1.5 technical reportcited by 559
  3. 2021Editing Factual Knowledge in Language Modelscited by 548
  4. 2020Just Pick a Sign: Optimizing Deep Multitask Models with Gradient Sign Dropoutcited by 199
  5. 2020Gradient Vaccine: Investigating and Improving Multi-task Optimization in Massively Multilingual Modelscited by 167
  6. 2023Prodigy: An Expeditiously Adaptive Parameter-Free Learnercited by 129
  7. 2023Learning-Rate-Free Learning by D-Adaptationcited by 113
  8. 2021Reasonable Effectiveness of Random Weighting: A Litmus Test for Multi-Task Learningcited by 100
  9. 2023DoG is SGD's Best Friend: A Parameter-Free Dynamic Step Size Schedulecited by 85
  10. 2021RotoGrad: Gradient Homogenization in Multitask Learningcited by 82
  11. 2021Multi-Task Learning in Natural Language Processing: An Overviewcited by 61
  12. 2022Do Current Multi-Task Optimization Methods in Deep Learning Even Help?cited by 60
  13. 2022A Survey of Multi-task Learning in Natural Language Processing: Regarding Task Relatedness and Training Methodscited by 42
  14. 2024AKEW: Assessing Knowledge Editing in the Wildcited by 35
  15. 2023Fair Enough: Standardizing Evaluation and Model Selection for Fairness Research in NLPcited by 11
  16. 2024Gradient descent with generalized Newton's methodcited by 9
  17. 2023On the Pareto Front of Multilingual Neural Machine Translationcited by 7

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 15; 2026: 14.

29 papers.

Sort:

By area:

  1. 2025BLUR: A Bi-Level Optimization Approach for LLM Unlearningcited by 23
  2. 2025LUME: LLM Unlearning with Multitask Evaluationscited by 21
  3. 2025SAEs Can Improve Unlearning: Dynamic Sparse Autoencoder Guardrails for Precision Unlearning in LLMscited by 21
  4. 2025A General Framework to Enhance Fine-tuning-based LLM Unlearningcited by 12
  5. 2025Constrained Entropic Unlearning: A Primal-Dual Framework for Large Language Modelscited by 10
  6. 2025SoK: Machine Unlearning for Large Language Modelscited by 9
  7. 2025Leak@k: Unlearning Does Not Make LLMs Forget Under Probabilistic Decodingcited by 9
  8. 2026VLA-Forget: Vision-Language-Action Unlearning for Embodied Foundation Modelscited by 9
  9. 2025Not Every Token Needs Forgetting: Selective Unlearning to Limit Change in Utility in Large Language Model Unlearningcited by 7
  10. 2025Machine Unlearning under Overparameterizationcited by 4
  11. 2025OFMU: Optimization-Driven Framework for Machine Unlearningcited by 4
  12. 2025SemEval-2025 Task 4: Unlearning sensitive content from Large Language Modelscited by 3
17 more
  1. 2025A Survey on Unlearning in Large Language Modelscited by 2
  2. 2026Unlearning the Unpromptable: Prompt-free Instance Unlearning in Diffusion Modelscited by 1
  3. 2025Cyber for AI at SemEval-2025 Task 4: Forgotten but Not Lost: The Balancing Act of Selective Unlearning in Large Language Modelscited by 0
  4. 2025A mean teacher algorithm for unlearning of language modelscited by 0
  5. 2025UNO: Unlearning via Orthogonalization in Generative modelscited by 0
  6. 2026UnPII: Unlearning Personally Identifiable Information with Quantifiable Exposure Riskcited by 0
  7. 2026Structural Disentanglement in Bilinear MLPs via Architectural Inductive Biascited by 0
  8. 2026MPU: Towards Secure and Privacy-Preserving Knowledge Unlearning for Large Language Modelscited by 0
  9. 2026Harmonizing Multi-Objective LLM Unlearning via Unified Domain Representation and Bidirectional Logit Distillationcited by 0
  10. 2026How Hard Can It Be? Hardness-Aware Multi-Objective Unlearningcited by 0
  11. 2026LLM Unlearning for Cyber Defense: A Survey on Methods, Challenges, and Emerging Threatscited by 0
  12. 2026The More Popular, The Harder to Forget: Adaptive Popularity for LLM Unlearningcited by 0
  13. 2026SAUL: Sharpness-Aware Augmented-Lagrangian Unlearningcited by 0
  14. 2026Can Scientific Claims Be Removed from Large Language Models? A Systematic Evaluation of Claim-Level Unlearningcited by 0
  15. 2026BLADE: Bilevel Low-rank Augmented-Lagrangian Erasure for LLM Unlearningcited by 0
  16. 2026Neuralyzing the Trace: Selective Representation-Level Unlearning with Contrastive Sparse Autoencoderscited by 0
  17. 2026Do LLMs Really Forget? Hidden-State Leakage in Model Unlearning and How to Fix itcited by 0