Enhancing LLM Metacognition via Cognitive Pairwise Training

• URL: http://arxiv.org/abs/2606.00869v1
• Date: Sat, 30 May 2026 19:53:19 GMT
• Title: Enhancing LLM Metacognition via Cognitive Pairwise Training
• Abstract summary: Reinforcement learning with verifiable rewards can make models more willing to give confident answers when evidence or reasoning is unreliable.<n>Existing SFT or RL methods mainly teach LLMs to refuse or express uncertainty at the response level.<n>We propose Cognitive Pairwise Training (CPT), a cognitive mid-training alignment stage that turns pairwise comparisons over reasoning traces into a reusable alignment signal.
• Score: 56.19495984013656
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Reinforcement learning with verifiable rewards (RLVR) has become central to LLM reasoning, but its outcome-level rewards can make models more willing to give confident answers when evidence or reasoning is unreliable. Existing SFT or RL methods mainly teach LLMs to refuse or express uncertainty at the response level, which can overfit abstention behavior rather than improve reasoning reliability. To address this limitation, we propose Cognitive Pairwise Training (CPT), a cognitive mid-training alignment stage that turns pairwise comparisons over reasoning traces into a reusable alignment signal. By learning to distinguish trustworthy from flawed reasoning, CPT encourages the model to internalize a reasoning-quality discrimination boundary rather than memorize surface refusal patterns. Across five model scales and three model families, CPT improves the reasoning--metacognition trade-off. At 14B, CPT+RL outperforms the standard SFT+RL pipeline by +2.2 math-average points and +5.2 abstention-F1 points. Further analyses show that CPT improves trace quality and exhibits strong robustness and scalability across evaluation and training settings. Code and models are released at https://github.com/Tsinghua-dhy/CPT.

Related papers

• Joint Training of Multi-Token Prediction in Reinforcement Learning via Optimal Coefficient Calibration [61.46060073417047]
  Multi-Token Prediction (MTP) has been a widely adopted module in pretraining.<n>We show that the per-step effect of MTP on the RL objective can be decomposed into two terms: a first-order correlation and a second-order penalty.<n>We propose an adaptive scheme that tracks the optimal coefficient online via a log-probability proxy at negligible cost.
  arXiv  Detail & Related papers  (2026-05-27T09:07:06Z)
• Beyond Correctness: Learning Robust Reasoning via Transfer [51.403609251508904]
  We adopt a simple philosophical view, robust reasoning should remain useful beyond the mind that produced it.<n>We introduce Reinforcement Learning with Transferable Reward, which operationalizes robustness via transfer reward.<n>Our approach improves sampling consistency while improving final answer accuracy, and it reaches comparable performance in substantially fewer training steps.
  arXiv  Detail & Related papers  (2026-02-09T10:41:44Z)
• Stop Rewarding Hallucinated Steps: Faithfulness-Aware Step-Level Reinforcement Learning for Small Reasoning Models [59.6715047267181]
  Small reasoning models (SRMs) are prone to hallucinations, especially in intermediate reasoning steps.<n>Existing mitigation methods based on online reinforcement learning rely on outcome-based rewards or coarse-grained chain-of-thought evaluation.<n>We propose Faithfulness-Aware Step-Level Reinforcement Learning (FaithRL), introducing step-level supervision via explicit faithfulness rewards from a process reward model.
  arXiv  Detail & Related papers  (2026-02-05T17:15:12Z)
• When Silence Is Golden: Can LLMs Learn to Abstain in Temporal QA and Beyond? [34.471723373734605]
  Large language models (LLMs) rarely admit uncertainty, often producing fluent but misleading answers, rather than abstaining (i.e., refusing to answer)<n>This weakness is even evident in temporal question answering, where models frequently ignore time-sensitive evidence and facts across different time-periods.<n>Existing approaches such as calibration might be unreliable in capturing uncertainty in complex reasoning.<n>We frame abstention as a teachable skill and introduce a pipeline that couples Chain-of-Thought (CoT) supervision with Reinforcement Learning (RL) guided by abstention-aware rewards.
  arXiv  Detail & Related papers  (2026-02-04T16:54:47Z)
• Confidence as a Reward: Transforming LLMs into Reward Models [54.98336080630691]
  Confidence-as-a-Reward (CRew) is a training-free method that utilizes token-level confidence in the model's final answers as a proxy for reward.<n>We show that CRew outperforms existing training-free reward approaches on the MATH500 and RewardMATH benchmarks.<n>We propose CRew-DPO, a training strategy that constructs preference data from confidence scores combined with correctness signals.
  arXiv  Detail & Related papers  (2025-10-15T12:51:47Z)
• ConfClip: Confidence-Weighted and Clipped Reward for Reinforcement Learning in LLMs [32.13266235550995]
  Reinforcement learning (RL) has become a standard paradigm for refining large language models (LLMs)<n>Inspired by observations from human learning, we introduce a RL technique that integrates verifiable outcomes with the model's own confidence estimates.
  arXiv  Detail & Related papers  (2025-09-22T13:00:35Z)
• Rethinking Reasoning Quality in Large Language Models through Enhanced Chain-of-Thought via RL [19.659532349434418]
  Reinforcement learning (RL) has recently become the dominant paradigm for strengthening the reasoning abilities of large language models.<n>Yet the rule-based reward functions commonly used on mathematical or programming benchmarks assess only answer format and correctness.<n>We propose Dynamic Reasoning Efficiency Reward (DRER) -- a plug-and-play RL reward framework that reshapes both reward and advantage signals.
  arXiv  Detail & Related papers  (2025-09-07T11:52:18Z)
• Post-Training Large Language Models via Reinforcement Learning from Self-Feedback [3.73824942136665]
  Large Language Models (LLMs) often produce plausible but poorly-calibrated answers.<n>We present Reinforcement Learning from Self-Feedback (RLSF), a post-training stage that uses the model's own confidence as an intrinsic reward.
  arXiv  Detail & Related papers  (2025-07-29T15:46:26Z)
• Improve Vision Language Model Chain-of-thought Reasoning [86.83335752119741]
  Chain-of-thought (CoT) reasoning in vision language models (VLMs) is crucial for improving interpretability and trustworthiness. We show that training VLM on short answers does not generalize well to reasoning tasks that require more detailed responses.
  arXiv  Detail & Related papers  (2024-10-21T17:00:06Z)

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.