mlsanityData to 14 Oct 2026

Dissecting Fine-Tuning Unlearning in Large Language Models

arXiv 2410.06606 ↗First posted 9 Oct 2024cs.CL cs.LGcited by 26

On the map: LLM safety and security → Machine unlearning

Abstract (arXiv)

Fine-tuning-based unlearning methods prevail for preventing targeted harmful, sensitive, or copyrighted information within large language models while preserving overall capabilities. However, the true effectiveness of these methods is unclear. In this work, we delve into the limitations of fine-tuning-based unlearning through activation patching and parameter restoration experiments. Our findings reveal that these methods alter the model's knowledge retrieval process, providing further evidence that they do not genuinely erase the problematic knowledge embedded in the model parameters. Instead, the coefficients generated by the MLP components in the model's final layer are the primary contributors to these seemingly positive unlearning effects, playing a crucial role in controlling the model's behaviors. Furthermore, behavioral tests demonstrate that this unlearning mechanism inevitably impacts the global behavior of the models, affecting unrelated knowledge or capabilities. The code is released at https://github.com/yihuaihong/Dissecting-FT-Unlearning.

Key facts. Cited by 26 papers in the map (data through 14 Oct 2026). Placed in Machine unlearning, one of 12 directions in LLM safety and security. Builds on 21 of its 33 references that are in the map. First posted on arXiv 9 Oct 2024.

Builds on

21 of 33 references.

Closest ancestors · 6

Same direction, newest first

  1. 2024Intrinsic Test of Unlearning Using Parametric Knowledge Traces17 Jun
  2. 2024Digital Forgetting in Large Language Models: A Survey of Unlearning Methods2 Apr
  3. 2024Dissecting Language Models: Machine Unlearning via Selective Pruning2 Mar
  4. 2023Unlearn What You Want to Forget: Efficient Unlearning for LLMs31 Oct
  5. 2023Who's Harry Potter? Approximate Unlearning in LLMs3 Oct
  6. 2022Knowledge Unlearning for Mitigating Privacy Risks in Language Models4 Oct

Other references · 15

  1. 2023Llama 2: Open Foundation and Fine-Tuned Chat Modelscited by 13,028
  2. 2021Evaluating Large Language Models Trained on Codecited by 8,320
  3. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 8,065
  4. 2022Locating and Editing Factual Associations in GPTcited by 2,691
  5. 2020Transformer Feed-Forward Layers Are Key-Value Memoriescited by 1,336
  6. 2022Mass-Editing Memory in a Transformercited by 1,001
  7. 2024OLMo: Accelerating the Science of Language Modelscited by 613
  8. 2023Dissecting Recall of Factual Associations in Auto-Regressive Language Modelscited by 528
  9. 2024Dolma: an Open Corpus of Three Trillion Tokens for Language Model Pretraining Researchcited by 483
  10. 2024A Mechanistic Understanding of Alignment Algorithms: A Case Study on DPO and Toxicitycited by 185
  11. 2023Can Sensitive Information Be Deleted From LLMs? Objectives for Defending Against Extraction Attackscited by 158
  12. 2023Use of LLMs for Illicit Purposes: Threats, Prevention Measures, and Vulnerabilitiescited by 82
3 more
  1. 2024Eraser: Jailbreaking Defense in Large Language Models via Unlearning Harmful Knowledgecited by 51
  2. 2024Localizing Paragraph Memorization in Language Modelscited by 31
  3. 2024Towards Comprehensive Post Safety Alignment of Large Language Models via Safety Patchingcited by 17

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2024: 1; 2025: 10; 2026: 15.

26 papers.

Sort:

By area:

  1. 2025Adaptive Multi-Subspace Representation Steering for Attribute Alignment in Large Language Modelscited by 16
  2. 2025Mechanistic Unveiling of Transformer Circuits: Self-Influence as a Key to Model Reasoningcited by 11
  3. 2025SoK: Machine Unlearning for Large Language Modelscited by 11
  4. 2025LUNE: Efficient LLM Unlearning via LoRA Fine-Tuning with Negative Examplescited by 8
  5. 2026Contrastive Representation Shaping for LLM Unlearningcited by 7
  6. 2024Towards Multi-dimensional Explanation Alignment for Medical Classificationcited by 6
  7. 2026Controlling Repetition in Protein Language Modelscited by 5
  8. 2025Towards Reasoning-Preserving Unlearning in Multimodal Large Language Modelscited by 4
  9. 2026Alignment midtraining for animalscited by 4
  10. 2025KnowledgeSmith: Uncovering Knowledge Updating in LLMs with Model Editing and Unlearningcited by 3
  11. 2026Understanding and Enforcing Weight Disentanglement in Task Arithmeticcited by 3
  12. 2025ZK-APEX: Zero-Knowledge Approximate Personalized Unlearning with Executable Proofscited by 2
14 more
  1. 2026KUDA: Knowledge Unlearning by Deviating Representation for Large Language Modelscited by 2
  2. 2026Multilingual Unlearning in LLMs: Transfer, Dynamics, and Reversibilitycited by 2
  3. 2025Understanding Machine Unlearning Through the Lens of Mode Connectivitycited by 1
  4. 2025The Rise of Parameter Specialization for Knowledge Storage in Large Language Modelscited by 1
  5. 2025Forgetting-MarI: LLM Unlearning via Marginal Information Regularizationcited by 1
  6. 2026Faithful-Patchscopes: Understanding and Mitigating Model Bias in Hidden Representations Explanation of Large Language Modelscited by 1
  7. 2026Behemoth: Benchmarking Unlearning in LLMs Using Fully Synthetic Datacited by 0
  8. 2026EvoMU: Evolutionary Machine Unlearningcited by 0
  9. 2026Structural Disentanglement in Bilinear MLPs via Architectural Inductive Biascited by 0
  10. 2026Fast Unlearning at Scale via Margin Self-Correctioncited by 0
  11. 2026Understanding Machine Unlearning Through the Lens of Mode Connectivitycited by 0
  12. 2026CleanVideo: Adaptive Concept Erasure for Text-to-Video Diffusion Modelscited by 0
  13. 2026The Tokens Remember: When Tokenization Bypasses Knowledge Editing and Unlearningcited by 0
  14. 2026Learning What to Forget: Distributional Unlearning for LLM Representation Spacescited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Anatomy of Unlearning: The Dual Impact of Fact Salience and Model Fine-Tuning
  2. 2025A Survey on Unlearning in Large Language Models
  3. 2025Understanding the Dilemma of Unlearning for Large Language Models
  4. 2026UnlearningSoup: Is Repeated Tuning Necessary for Large Language Model Unlearning?
  5. 2024To Forget or Not? Towards Practical Knowledge Unlearning for Large Language Models
  6. 2025Existing Large Language Model Unlearning Evaluations Are Inconclusive
  7. 2025Does Localization Inform Unlearning? A Rigorous Examination of Local Parameter Attribution for Knowledge Unlearning in Language Models
  8. 2026Model Unlearning Objectives Vary for Distinct Language Functions