Cover Image

Reasoning Model Unlearning: Forgetting Traces, Not Just Answers, While Preserving Reasoning Skills

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Wang, Changsheng, Fan, Chongyu, Zhang, Yihua, Jia, Jinghan, Wei, Dennis, Ram, Parikshit, Baracaldo, Nathalie, Liu, Sijia
Format:	Preprint
Published:	2025
Subjects:	Artificial Intelligence Machine Learning
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

WAGLE: Strategic Weight Attribution for Effective and Modular Unlearning in Large Language Models
by: Jia, Jinghan, et al.
Published: (2024)

Invariance Makes LLM Unlearning Resilient Even to Unanticipated Downstream Fine-Tuning
by: Wang, Changsheng, et al.
Published: (2025)

EPiC: Towards Lossless Speedup for Reasoning Training through Edge-Preserving CoT Condensation
by: Jia, Jinghan, et al.
Published: (2025)

Beyond SFT: Reinforcement Learning for Safer Large Reasoning Models with Better Reasoning Ability
by: Jia, Jinghan, et al.
Published: (2025)

Downgrade to Upgrade: Optimizer Simplification Enhances Robustness in LLM Unlearning
by: Lang, Yicheng, et al.
Published: (2025)

Challenging Forgets: Unveiling the Worst-Case Forget Sets in Machine Unlearning
by: Fan, Chongyu, et al.
Published: (2024)

LLM Unlearning on Noisy Forget Sets: A Study of Incomplete, Rewritten, and Watermarked Data
by: Wang, Changsheng, et al.
Published: (2025)

CyclicReflex: Improving Reasoning Models via Cyclical Reflection Token Scheduling
by: Fan, Chongyu, et al.
Published: (2025)

SalUn: Empowering Machine Unlearning via Gradient-based Weight Saliency in Both Image Classification and Generation
by: Fan, Chongyu, et al.
Published: (2023)

Towards LLM Unlearning Resilient to Relearning Attacks: A Sharpness-Aware Minimization Perspective and Beyond
by: Fan, Chongyu, et al.
Published: (2025)

Simplicity Prevails: Rethinking Negative Preference Optimization for LLM Unlearning
by: Fan, Chongyu, et al.
Published: (2024)

One Token Embedding Is Enough to Deadlock Your Large Reasoning Model
by: Zhang, Mohan, et al.
Published: (2025)

Understanding Forgetting in LLM Supervised Fine-Tuning and Preference Learning -- A Convex Optimization Perspective
by: Fernando, Heshan, et al.
Published: (2024)

SEUF: Is Unlearning One Expert Enough for Mixture-of-Experts LLMs?
by: Zhuang, Haomin, et al.
Published: (2024)

Defensive Unlearning with Adversarial Training for Robust Concept Erasure in Diffusion Models
by: Zhang, Yimeng, et al.
Published: (2024)

Model Sparsity Can Simplify Machine Unlearning
by: Jia, Jinghan, et al.
Published: (2023)

Split, Unlearn, Merge: Leveraging Data Attributes for More Effective Unlearning in LLMs
by: Kadhe, Swanand Ravindra, et al.
Published: (2024)

UnlearnCanvas: Stylized Image Dataset for Enhanced Machine Unlearning Evaluation in Diffusion Models
by: Zhang, Yihua, et al.
Published: (2024)

Forget Vectors at Play: Universal Input Perturbations Driving Machine Unlearning in Image Classification
by: Sun, Changchang, et al.
Published: (2024)

LLM Unlearning Under the Microscope: A Full-Stack View on Methods and Metrics
by: Fan, Chongyu, et al.
Published: (2025)

Forgetting to Forget: Attention Sink as A Gateway for Backdooring LLM Unlearning
by: Shang, Bingqi, et al.
Published: (2025)

The Unlearning Mirage: A Dynamic Framework for Evaluating LLM Unlearning
by: Shah, Raj Sanjay, et al.
Published: (2026)

Unlearners Can Lie: Evaluating and Improving Honesty in LLM Unlearning
by: Gu, Renjie, et al.
Published: (2026)

Enhancing In-context Learning via Linear Probe Calibration
by: Abbas, Momin, et al.
Published: (2024)

Selective Forgetting for Large Reasoning Models
by: Le, Tuan, et al.
Published: (2026)

The Chain Holds, the Answer Folds: Trace-Answer Dissociation in Reasoning Models Under Adversarial Pressure
by: Li, Yubo, et al.
Published: (2026)

Auditing Reasoning-Trace Memorization Claims after Unlearning with Head-Conditioned Canaries
by: Li, Yanhang, et al.
Published: (2026)

How Do Answer Tokens Read Reasoning Traces? Self-Reading Patterns in Thinking LLMs for Quantitative Reasoning
by: Chen, Haoyang, et al.
Published: (2026)

ACU: Analytic Continual Unlearning for Efficient and Exact Forgetting with Privacy Preservation
by: Tang, Jianheng, et al.
Published: (2025)

Towards Reasoning-Preserving Unlearning in Multimodal Large Language Models
by: Li, Hongji, et al.
Published: (2025)

Galaxy: A Cognition-Centered Framework for Proactive, Privacy-Preserving, and Self-Evolving LLM Agents
by: Bao, Chongyu, et al.
Published: (2025)

Reversing the Forget-Retain Objectives: An Efficient LLM Unlearning Framework from Logit Difference
by: Ji, Jiabao, et al.
Published: (2024)

LLM Unlearning Reveals a Stronger-Than-Expected Coreset Effect in Current Benchmarks
by: Pal, Soumyadeep, et al.
Published: (2025)

Reasoning While Recommending: Entropy-Guided Latent Reasoning in Generative Re-ranking Models
by: Zhang, Changshuo
Published: (2026)

Safety Mirage: How Spurious Correlations Undermine VLM Safety Fine-Tuning and Can Be Mitigated by Machine Unlearning
by: Chen, Yiwei, et al.
Published: (2025)

To Generate or Not? Safety-Driven Unlearned Diffusion Models Are Still Easy To Generate Unsafe Images ... For Now
by: Zhang, Yimeng, et al.
Published: (2023)

Preserving Cross-Modal Stability for Visual Unlearning in Multimodal Scenarios
by: Li, Jinghan Xu Yuyang Zhang Qixuan Cai Jiancheng Chen Keqiu
Published: (2025)

Beyond Reasoning Gains: Mitigating General Capabilities Forgetting in Large Reasoning Models
by: Phan, Hoang, et al.
Published: (2025)

SOUL: Unlocking the Power of Second-Order Optimization for LLM Unlearning
by: Jia, Jinghan, et al.
Published: (2024)

Sandwich Reasoning: An Answer-Reasoning-Answer Approach for Low-Latency Query Correction
by: Zhang, Chen, et al.
Published: (2026)