CiPO: Counterfactual Unlearning for Large Reasoning Models through Iterative Preference Optimization

• URL: http://arxiv.org/abs/2604.15847v1
• Date: Fri, 17 Apr 2026 08:56:36 GMT
• Title: CiPO: Counterfactual Unlearning for Large Reasoning Models through Iterative Preference Optimization
• Abstract summary: Large Reasoning Models (LRMs) emphasize long chain-of-thought (CoT) reasoning to address complex questions.<n>Existing methods either struggle to completely eliminate undesired knowledge from the CoT traces or degrade the reasoning performances due to the interference with the reasoning process.<n>We introduce Counterfactual Unlearning through iterative Preference Optimization (CiPO), a novel framework that redefines unlearning as the targeted intervention of the CoT reasoning in LRMs.
• Score: 54.774620283208776
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Machine unlearning has gained increasing attention in recent years, as a promising technique to selectively remove unwanted privacy or copyrighted information from Large Language Models that are trained on a massive scale of human data. However, the emergence of Large Reasoning Models (LRMs), which emphasize long chain-of-thought (CoT) reasoning to address complex questions, presents a dilemma to unlearning: existing methods either struggle to completely eliminate undesired knowledge from the CoT traces or degrade the reasoning performances due to the interference with the reasoning process. To this end, we introduce Counterfactual Unlearning through iterative Preference Optimization (CiPO), a novel framework that redefines unlearning as the targeted intervention of the CoT reasoning in LRMs. More specifically, given a desired unlearning target answer, CiPO instructs LRMs to generate a logically valid counterfactual reasoning trace for preference tuning. As the LRM adjusts to the counterfactual trace, CiPO iteratively updates the preference learning data to increase the discrepancy from the original model. This iterative loop ensures both desirable unlearning and smooth optimization, effectively mitigating the dilemma. Experiments on challenging benchmarks demonstrate that CiPO excels at unlearning, completely removing knowledge from both the intermediate CoT steps and the final answer, while preserving the reasoning abilities of LRMs.

Related papers

• Think2Go: Generative Next POI Recommendation with LLM Reasoning [57.37706163990998]
  Next Point-of-Interest (POI) recommendation task focuses on mining user behavioral preference patterns from historical check-ins to provide personalized suggestions for the next destination.<n>Existing methods rely on shallow contextual information and handcrafted feature interactions to predict the next POI.<n>We propose Think2Go, a novel generative next POI recommendation framework, which enhances the model's comprehension of SID representations and explores diverse spatial-temporal patterns via test-time computational scaling.
  arXiv  Detail & Related papers  (2026-07-31T03:52:20Z)
• WhisperRec: Latent Reasoning for Efficient Foundation Recommendation Models [51.88075626243057]
  We propose WhisperRec, an efficient latent reasoning framework for foundation recommendation models (FRMs)<n>WhisperRec compresses teacher-generated Chain-of-Thought (CoT) into learnable latent reasoning tokens.<n>We show that WhisperRec consistently outperforms explicit-CoT methods and conventional baselines.
  arXiv  Detail & Related papers  (2026-07-29T08:48:35Z)
• EvoThink: Evolving Thinking in Large Reasoning Models via Self-Pruning and Aha-Moment Preference Optimization [23.37325724250601]
  Large Reasoning Models (LRMs) often suffer from overthinking due to redundant verification steps.<n>We propose EvoThink, a framework that reduces redundant verification and encourages the exploration of new reasoning paths.
  arXiv  Detail & Related papers  (2026-07-22T09:40:22Z)
• Selective Forgetting for Large Reasoning Models [22.21783013812524]
  Large Reasoning Models (LRMs) generate structured chains of thought (CoTs) before producing final answers.<n>The memorization of sensitive information in the training data such as copyrighted and private content has led to ethical and legal concerns.<n>We propose a novel LRM unlearning framework that selectively removes sensitive reasoning components while preserving general reasoning capabilities.
  arXiv  Detail & Related papers  (2026-04-04T03:47:12Z)
• Accordion-Thinking: Self-Regulated Step Summaries for Efficient and Readable LLM Reasoning [62.680551162054975]
  We introduce an end-to-end framework where LLMs learn to self-regulate the granularity of the reasoning steps through dynamic summarization.<n>We apply reinforcement learning to incentivize this capability further, uncovering a critical insight: the accuracy gap between the highly efficient Fold mode and the exhaustive Unfold mode progressively narrows.<n>Our Accordion-Thinker demonstrates that with learned self-compression, LLMs can tackle complex reasoning tasks with minimal dependency token overhead.
  arXiv  Detail & Related papers  (2026-02-03T08:34:20Z)
• CALM Before the STORM: Unlocking Native Reasoning for Optimization Modeling [60.55856973678002]
  Large Reasoning Models (LRMs) have demonstrated strong capabilities in complex multi-step reasoning.<n>Existing domain adaptation methods, originally designed for earlier instruction-tuned models, often fail to exploit the advanced reasoning patterns of modern LRMs.<n>We propose textbfCALM, a framework that progressively refines LRMs within their native reasoning modes for optimization modeling tasks.
  arXiv  Detail & Related papers  (2025-10-05T13:38:31Z)
• Reasoning Model Unlearning: Forgetting Traces, Not Just Answers, While Preserving Reasoning Skills [42.1825027925353]
  Large reasoning models (LRMs) have enabled strong chain-of-thought (CoT) generation through test-time computation.<n>We show that conventional unlearning algorithms, originally designed for non-reasoning models, are inadequate for LRMs.<n>We propose Reasoning-aware Representation Misdirection for Unlearning ($R2MU$), a novel method that effectively suppresses sensitive reasoning traces.
  arXiv  Detail & Related papers  (2025-06-15T20:54:23Z)
• Reinforced Latent Reasoning for LLM-based Recommendation [92.56166822197919]
  Large Language Models (LLMs) have demonstrated impressive reasoning capabilities in complex problem-solving tasks.<n>Existing methods typically rely on fine-tuning with explicit chain-of-thought (CoT) data.<n>In this work, we explore an alternative approach that shifts from explicit CoT reasoning to compact, information-dense latent reasoning.
  arXiv  Detail & Related papers  (2025-05-25T11:03:45Z)
• R$^2$ec: Towards Large Recommender Models with Reasoning [59.32598867813266]
  We propose R$2$ec, a unified large recommender model with intrinsic reasoning capability.<n>R$2$ec introduces a dual-head architecture that supports both reasoning chain generation and efficient item prediction in a single model.<n>To overcome the lack of annotated reasoning data, we design RecPO, a reinforcement learning framework.
  arXiv  Detail & Related papers  (2025-05-22T17:55:43Z)
• R-TOFU: Unlearning in Large Reasoning Models [5.116399056871577]
  We introduce Reasoning-TOFU, the first benchmark tailored to this setting.<n>R-TOFU augments existing unlearning tasks with realistic CoT annotations.<n>We propose Reasoned IDK, a preference-optimization variant that preserves coherent yet inconclusive reasoning.
  arXiv  Detail & Related papers  (2025-05-21T07:44:30Z)
• Reinforcing Thinking through Reasoning-Enhanced Reward Models [6.636512424910708]
  Large Language Models (LLMs) exhibit great potential in complex multi-step reasoning through inference-time thinking.<n>LLMs struggle with deciding when to stop thinking due to limited self-awareness about their knowledge boundaries.<n>This work addresses these challenges by distilling the LLM's own reasoning processes into synthetic behavioral data.
  arXiv  Detail & Related papers  (2024-12-31T04:50:15Z)
• Let's reward step by step: Step-Level reward model as the Navigators for Reasoning [64.27898739929734]
  Process-Supervised Reward Model (PRM) furnishes LLMs with step-by-step feedback during the training phase. We propose a greedy search algorithm that employs the step-level feedback from PRM to optimize the reasoning pathways explored by LLMs. To explore the versatility of our approach, we develop a novel method to automatically generate step-level reward dataset for coding tasks and observed similar improved performance in the code generation tasks.
  arXiv  Detail & Related papers  (2023-10-16T05:21:50Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.