Enhancing LLM Metacognition via Cognitive Pairwise Training
Weitao Li, Hao Zhou, Xuanyu Lei, Fandong Meng, Yuanhang Liu, Jingyi Ren, Ante Wang, Xiaolong Wang, Yuanchi Zhang, Fuwen Luo, Guangwen Yang, Lin Gan, Weizhi Ma, Yang Liu
cs.LG
Submitted: 2026-08-22
Updated: 2026-08-25
Code: https://github.com/Tsinghua-dhy/CPT
Terminology
Sources
- Kimi K2.5: Visual Agentic Intelligence
- Qwen3 Technical Report
- OpenClaw-RL: Train Any Agent Simply by Talking
- Tongyi DeepResearch Technical Report
- WildClawBench: A Benchmark for Real-World, Long-Horizon Agent Evaluation
- Hallucinations Undermine Trust; Metacognition is a Way Forward
- Language Models (Mostly) Know What They Know
- Are Reasoning Models More Prone to Hallucination?
- Stop Overthinking: A Survey on Efficient Reasoning for Large Language Models
- Auditing Meta-Cognitive Hallucinations in Reasoning Large Language Models
- Inference-Time Scaling for Generalist Reward Modeling
- Agent Learning via Early Experience
- Model Spec Midtraining: Improving How Alignment Training Generalizes
- Large Language Models are not Fair Evaluators
- The False Promise of Imitating Proprietary LLMs
- The Llama 3 Herd of Models
- Olmo 3
- AbstentionBench: Reasoning LLMs Fail on Unanswerable Questions
- Beyond Binary Rewards: Training LMs to Reason About Their Uncertainty
- Rewarding Doubt: A Reinforcement Learning Approach to Calibrated Confidence Expression of Large Language Models
Related papers
- Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
- AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
- Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
- Convergence issues in Relational Concept Analysis based on AOC-posets
- Beliefs Beyond Posteriors: Local-Consistency Optimisation for Bayesian Neural Networks
- Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks