LACUNA is a new testbed that injects PII into predefined model parameters to benchmark the localization precision of LLM unlearning methods, revealing that SOTA approaches are imprecise despite strong output performance.
hub
Who’s harry potter? approximate unlearning in llms.arXiv preprint arXiv:2310.02238
33 Pith papers cite this work, alongside 13 external citations. Polarity classification is still indexing.
33
Pith papers citing it
13
external citations · external index
hub tools
citation-role summary
background 3
method 1
citation-polarity summary
representative citing papers
Sequential LLM defense deployment leads to risk exacerbation in 38.9% of cases due to anti-aligned updates in shared critical layers, addressed by conflict-guided layer freezing.
INT4 quantization recovers up to 22 times more forgotten training data in unlearned LLMs, and the proposed DURABLEUN-SAF method is the first to maintain forgetting across BF16, INT8, and INT4 precisions.
NPO enables stable unlearning of 50%+ training data in LLMs on TOFU by making collapse exponentially slower than gradient ascent, preserving sensible outputs where prior methods fail.
Process sidecars use a secant-based two-parameter edit to achieve second-order accurate memory revocation after safety training, outperforming scalar task arithmetic on refusal tasks across three models.
TimeROME-DLM enables training-free knowledge editing in masked diffusion language models via temporal causal tracing and low-rank residual edit memory applied at inference time.
REMEDI is a new benchmark for evaluating machine unlearning in multi-label clinical disease inference on MIMIC-III data that reveals trade-offs in existing methods.
TRACER uses token reassignment for concept-related items plus a coherence regularizer to unlearn specific concepts in generative recommendation while preserving utility better than baselines.
For any ρ>0 there exists a ρ-TV-stable RL algorithm for tabular MDPs supporting exact unlearning at expected cost ρ√(ln T) of retraining from scratch, with regret O(H²√(SAT)+H³S²A+H^{2.5}S²A/ρ) and matching lower bound Ω(H√(SAT)+SAH/ρ).
DivIn samples initial noise from a guidance potential posterior via Langevin dynamics to improve diversity in class-to-image and text-to-image generation.
TC-UMIA is a population-level attack using pre- and post-unlearning predictions to infer membership across forget, retain, and unseen sets, revealing added privacy leakage to retained data.
LLM unlearning is reframed as inadvertently installing backdoor triggers on forget-tokens; Random Noise Augmentation is introduced as a defense that improves robustness with theoretical guarantees.
Min-K% Prob detects pretraining data in LLMs by flagging outlier low-probability words in text, achieving 7.4% better performance than prior methods on the new WIKIMIA benchmark.
MASC achieves competitive forget-retain trade-offs in language model unlearning at lower computational cost via margin self-correction and an online stopping criterion on TOFU, MUSE News, and MUSE Books.
HAMU is a constrained-optimization unlearning method that uses forget-retain data similarity as a hardness measure to guarantee specified forget-quality gains while minimizing retain degradation.
VGID constructs an intervention-induced teacher distribution via visual perturbation plus textual in-context unlearning and distills it into the student MLLM to achieve parameter-level forgetting.
First unified survey formalizing Pretraining Data Exposure across exposure levels and reviewing attack, defense, and contamination methods for LLMs.
Swapping the reasoning trace prefill on unlearned weights can replicate or reverse the parser-split bypass gap, showing that the gap alone does not identify or rule out weight-level memorization.
ZeroUnlearn reformulates machine unlearning as knowledge re-mapping via model editing, using multiplicative updates with closed-form solutions for efficient few-shot removal of sensitive representations while preserving utility.
Probe-geometry alignment erases cross-sequence memorization signatures in LLMs below chance using per-depth rank-one activation interventions with negligible impact on zero-shot capabilities.
TokenUnlearn identifies critical tokens via masking and entropy signals then applies hard selection or soft weighting to unlearn only those tokens, yielding better forgetting and retained utility than sequence-level baselines on TOFU and WMDP.
A separable expert architecture uses base models, LoRA adapters, and deletable per-user proxies to enable privacy-preserving personalization and deterministic unlearning in LLMs.
CiPO removes undesired knowledge from both intermediate reasoning steps and final answers in large reasoning models by iteratively optimizing preferences toward valid counterfactual traces while keeping overall reasoning performance intact.
TRACE is a three-stage optimization framework that realigns LLMs to new policies by categorizing preference conflicts, scoring impact via bi-level optimization, and applying hybrid losses without new human annotations.
citing papers explorer
-
LACUNA: A Testbed for Evaluating Localization Precision for LLM Unlearning
LACUNA is a new testbed that injects PII into predefined model parameters to benchmark the localization precision of LLM unlearning methods, revealing that SOTA approaches are imprecise despite strong output performance.
-
Defenses at Odds: Measuring and Explaining Defense Conflicts in Large Language Models
Sequential LLM defense deployment leads to risk exacerbation in 38.9% of cases due to anti-aligned updates in shared critical layers, addressed by conflict-guided layer freezing.
-
DurableUn: Quantization-Induced Recovery Attacks in Machine Unlearning
INT4 quantization recovers up to 22 times more forgotten training data in unlearned LLMs, and the proposed DURABLEUN-SAF method is the first to maintain forgetting across BF16, INT8, and INT4 precisions.
-
Negative Preference Optimization: From Catastrophic Collapse to Effective Unlearning
NPO enables stable unlearning of 50%+ training data in LLMs on TOFU by making collapse exponentially slower than gradient ascent, preserving sensible outputs where prior methods fail.
-
Revocable Learned State via Process Sidecars
Process sidecars use a secant-based two-parameter edit to achieve second-order accurate memory revocation after safety training, outperforming scalar task arithmetic on refusal tasks across three models.
-
TimeROME-DLM: Temporal Causal Tracing and Low-Rank Inference-Time Knowledge Editing for Masked Diffusion Language Models
TimeROME-DLM enables training-free knowledge editing in masked diffusion language models via temporal causal tracing and low-rank residual edit memory applied at inference time.
-
REMEDI: A Benchmark for Retention and Unlearning Evaluation in Multi-label Clinical Disease Inference
REMEDI is a new benchmark for evaluating machine unlearning in multi-label clinical disease inference on MIMIC-III data that reveals trade-offs in existing methods.
-
TRACER: Token ReAssignment for Concept ERasure in Generative Recommendation
TRACER uses token reassignment for concept-related items plus a coherence regularizer to unlearn specific concepts in generative recommendation while preserving utility better than baselines.
-
Exact Unlearning in Reinforcement Learning
For any ρ>0 there exists a ρ-TV-stable RL algorithm for tabular MDPs supporting exact unlearning at expected cost ρ√(ln T) of retraining from scratch, with regret O(H²√(SAT)+H³S²A+H^{2.5}S²A/ρ) and matching lower bound Ω(H√(SAT)+SAH/ρ).
-
Initialization is Half the Battle: Generating Diverse Images from a Guidance Potential Posterior
DivIn samples initial noise from a guidance potential posterior via Langevin dynamics to improve diversity in class-to-image and text-to-image generation.
-
Revisiting Privacy Leakage in Machine Unlearning: Membership Inference Beyond the Forgotten Set
TC-UMIA is a population-level attack using pre- and post-unlearning predictions to infer membership across forget, retain, and unseen sets, revealing added privacy leakage to retained data.
-
Improving LLM Unlearning Robustness via Random Perturbations
LLM unlearning is reframed as inadvertently installing backdoor triggers on forget-tokens; Random Noise Augmentation is introduced as a defense that improves robustness with theoretical guarantees.
-
Detecting Pretraining Data from Large Language Models
Min-K% Prob detects pretraining data in LLMs by flagging outlier low-probability words in text, achieving 7.4% better performance than prior methods on the new WIKIMIA benchmark.
-
Fast Unlearning at Scale via Margin Self-Correction
MASC achieves competitive forget-retain trade-offs in language model unlearning at lower computational cost via margin self-correction and an online stopping criterion on TOFU, MUSE News, and MUSE Books.
-
How Hard Can It Be? Hardness-Aware Multi-Objective Unlearning
HAMU is a constrained-optimization unlearning method that uses forget-retain data similarity as a hardness measure to guarantee specified forget-quality gains while minimizing retain degradation.
-
Visual-Noise Guided In-Context Distillation for Multimodal Large Language Model Unlearning
VGID constructs an intervention-induced teacher distribution via visual perturbation plus textual in-context unlearning and distills it into the student MLLM to achieve parameter-level forgetting.
-
Pretraining Data Exposure in Large Language Models: A Survey of Membership Inference, Data Contamination, and Security Implications
First unified survey formalizing Pretraining Data Exposure across exposure levels and reviewing attack, defense, and contamination methods for LLMs.
-
Auditing Reasoning-Trace Memorization Claims after Unlearning with Head-Conditioned Canaries
Swapping the reasoning trace prefill on unlearned weights can replicate or reverse the parser-split bypass gap, showing that the gap alone does not identify or rule out weight-level memorization.
-
ZeroUnlearn: Few-Shot Knowledge Unlearning in Large Language Models
ZeroUnlearn reformulates machine unlearning as knowledge re-mapping via model editing, using multiplicative updates with closed-form solutions for efficient few-shot removal of sensitive representations while preserving utility.
-
Probe-Geometry Alignment: Erasing the Cross-Sequence Memorization Signature Below Chance
Probe-geometry alignment erases cross-sequence memorization signatures in LLMs below chance using per-depth rank-one activation interventions with negligible impact on zero-shot capabilities.
-
Unlearning What Matters: Token-Level Attribution for Precise Language Model Unlearning
TokenUnlearn identifies critical tokens via masking and entropy signals then applies hard selection or soft weighting to unlearn only those tokens, yielding better forgetting and retained utility than sequence-level baselines on TOFU and WMDP.
-
Separable Expert Architecture: Toward Privacy-Preserving LLM Personalization via Composable Adapters and Deletable User Proxies
A separable expert architecture uses base models, LoRA adapters, and deletable per-user proxies to enable privacy-preserving personalization and deterministic unlearning in LLMs.
-
CiPO: Counterfactual Unlearning for Large Reasoning Models through Iterative Preference Optimization
CiPO removes undesired knowledge from both intermediate reasoning steps and final answers in large reasoning models by iteratively optimizing preferences toward valid counterfactual traces while keeping overall reasoning performance intact.
-
The Realignment Problem: When Right becomes Wrong in LLMs
TRACE is a three-stage optimization framework that realigns LLMs to new policies by categorizing preference conflicts, scoring impact via bi-level optimization, and applying hybrid losses without new human annotations.
-
TOFU: A Task of Fictitious Unlearning for LLMs
TOFU is a new benchmark with synthetic profiles and metrics demonstrating that existing unlearning algorithms for LLMs fail to achieve effective forgetting of targeted information.
-
Null-Space Constrained Low-Rank Adaptation for Response-Specified Large Language Model Unlearning
NSRU constrains LoRA updates via null-space projection of retain subspaces to jointly optimize safe-target learning, undesired-response suppression, and retention in LLM unlearning.
-
Calibration vs Decision Making: Revisiting the Reliability Paradox in Unlearned Language Models
Unlearned language models retain low calibration error but show increased shortcut reliance on the TOFU benchmark, extending the reliability paradox to machine unlearning.
-
Towards Scalable Lifelong Knowledge Editing with Selective Knowledge Suppression
LightEdit enables scalable lifelong knowledge editing in LLMs via selective knowledge retrieval and probability suppression during decoding, outperforming prior methods on ZSRE, Counterfact, and RIPE while reducing training costs.
-
Position: No Retroactive Cure for Infringement during Training
Post-hoc mitigation cannot retroactively cure infringement that occurred during unauthorized data ingestion and training because liability attaches to data lineage and retained expressive value in model weights.
-
OFMU: Optimization-Driven Framework for Machine Unlearning
OFMU is a penalty-based bi-level optimizer for machine unlearning that alternates between a gradient-ascent forgetting step and a gradient-descent utility-restoration step, with a similarity penalty between forget and retain gradients.
-
Revisiting the Past: Data Unlearning with Model State History
MSA performs data unlearning in LLMs by arithmetic operations on prior model checkpoints to remove targeted datapoint influence, with experiments showing competitive or better results than existing unlearning methods.
-
AI as a Tool for Simulation-Based Experiments in Literary Studies
Proposes AI-driven simulations for literary-historical experiments and reports preliminary text-generation results claiming the first limited in-distribution outputs matching human novels.
-
AI Safety Landscape for Large Language Models: Taxonomy, State-of-the-art, and Future Directions
The paper introduces a taxonomy of AI safety for LLMs organized into Trustworthy AI, Responsible AI, and Safe AI perspectives, accompanied by a review of state-of-the-art methods, challenges, and future directions.