通知：联邦节假日调休：哥伦布日 / 原住民日（Columbus Day / Indigenous Peoples' Day），今日（13日）无更新。通知：联邦节假日调休：哥伦布日 / 原住民日（Columbus Day / Indigenous Peoples' Day），今日（13日）无更新。
arXiv 2602.11201cs.CL
链式思维推理中忠实度衰减的机制证据
Mechanistic Evidence for Faithfulness Decay in Chain-of-Thought Reasoning
Fordham University（福特汉姆大学）；Algoverse AI Research（Algoverse AI研究）；Rice University（稻子大学）；UC San Diego（圣地亚哥大学）；Santa Clara University（圣克拉拉大学）；LMU Munich（慕尼黑路德维希-马克西米利安大学）
Donald Ye, Max Loffgren, Om Kotadia, Linus Wong, Jonas Rohweder
Chain-of-Thought (CoT) explanations are widely used to interpret how language models solve complex problems, yet it remains unclear whether these step-by-step explanations reflect how the model actually reaches its answer, or merely post-hoc justifications. We propose Normalized Logit Difference Decay (NLDD), a metric that measures whether individual reasoning steps are faithful to the model's decision-making process. Our approach corrupts individual reasoning steps from the explanation and measures how much the model's confidence in its answer drops, to determine if a step is truly important. By standardizing these measurements, NLDD enables rigorous cross-model comparison across different architectures. Testing three model families across syntactic, logical, and arithmetic tasks, we discover a consistent Reasoning Horizon (k*) at 70--85% of chain length, beyond which reasoning tokens have little or negative effect on the final answer. We also find that models can encode correct internal representations while completely failing the task. These results show that accuracy alone does not reveal whether a model actually reasons through its chain. NLDD offers a way to measure when CoT matters.
Investigating the Interplay between Contextual and Parametric Chain-of-Thought Faithfulness under Optimization
University of Copenhagen（哥本哈根大学）；Technische Universität Berlin（柏林技术大学）；German Research Center for Artificial Intelligence (DFKI)（德国人工智能研究中心（DFKI））；BIFOLD – Berlin Institute for the Foundations of Learning and Data（BIFOLD – 柏林学习与数据基础研究院）
GeoFaith: A Spatio-Temporal Dual View of Faithful Chain-of-Thought
Xidian University（西安电子科技大学）；Xi’an Jiaotong University（西安交通大学）；Mohamed bin Zayed University of Artificial Intelligence（穆罕默德·本·扎耶德人工智能大学）；The Hong Kong University of Science and Technology (Guangzhou)（香港科学与技术大学（广州））；University of Science and Technology of China（中国科学技术大学）
Towards Better Chain-of-Thought: A Reflection on Effectiveness and Faithfulness
School of Artificial Intelligence, University of Chinese Academy of Sciences（人工智能学院，中国科学院大学）；The Key Laboratory of Cognition and Decision Intelligence for Complex Systems, Institute of Automation, Chinese Academy of Sciences（复杂系统认知与决策智能重点实验室，中国科学院自动化研究所）；China Merchants Bank（中国民生银行）
Measuring Faithfulness Depends on How You Measure: Classifier Sensitivity in LLM Chain-of-Thought Evaluation
University of Nevada, Las Vegas, Department of Management, Entrepreneurship and Technology, Lee Business School（内华达大学拉斯维加斯分校管理、创业与技术系，Lee商学院）；DeepNeuro AI
From Concept Alignment to Causal Grounding: An Intervention Test of Chain-of-Thought Faithfulness
German Research Center for Artificial Intelligence (DFKI)（德国人工智能研究中心（DFKI））；Saarland Informatics Campus（萨尔兰信息学园区）；Centre for European Research in Trusted AI (CERTAIN)（欧洲可信人工智能研究中心（CERTAIN））；Technische Universität Berlin（柏林工业大学）；IBM Research（IBM研究院）；University of Copenhagen（哥本哈根大学）；University of Groningen（格罗宁根大学）
Measuring Chain-of-Thought Monitorability Through Faithfulness and Verbosity
ML Alignment and Theory Scholars（机器学习对齐与理论学者计划）；University of Delaware（特拉华大学）；AI Safety Argentina（阿根廷AI安全组织）；Buenos Aires AI Safety Hub（布宜诺斯艾利斯AI安全中心）；University of Buenos Aires（布宜诺斯艾利斯大学）；Center for AI Safety（AI安全中心）
SIFT: Robust Meta-Faithfulness Verification of Chain-of-Thought Reasoning Under Distribution Shift
Istanbul Technical University（伊斯坦布尔理工大学）；Pabna University of Science and Technology（帕布纳科技大学）；American International University-Bangladesh（美国国际大学-孟加拉）；Deakin University（迪肯大学）；North South University（南北大学）
Iteration Head: A Mechanistic Study of Chain-of-Thought
FAIR, Meta AI（FAIR，Meta人工智能公司）；Datashape, INRIA（Datashape，法国国家信息与自动化研究所）；Courant University and Center for Data Science, NYU（纽约大学柯朗数学科学研究所与数据科学中心）；Center for Data Science, NYU（纽约大学数据科学中心）；NYU（纽约大学）；Courant University（柯朗大学）
Measuring and curing reasoning rigidity: from decorative chain-of-thought to genuine faithfulness
Indian Institute of Information Technology Allahabad (IIITA)（印度阿拉哈巴德信息技术学院）；National Institute of Electronics and Information Technology (NIELIT)（国家电子与信息技术研究院）
Stop Rewarding Hallucinated Steps: Faithfulness-Aware Step-Level Reinforcement Learning for Small Reasoning Models
Institute of Computing and Intelligence, Harbin Institute of Technology, Shenzhen, China（哈尔滨工业大学深圳研究院）；Institute of Artificial Intelligence (TeleAI), China Telecom Corp Ltd（中国电信人工智能研究院）；College of Integrated Circuits, Zhejiang University, Hangzhou, Zhejiang, China（浙江大学集成电路学院）；Zhongguancun Academy, Beijing, China（中关村学院）
Finite State Automata Inside Transformers with Chain-of-Thought: A Mechanistic Study on State Tracking
Key Laboratory of High Confidence Software Technology (PKU), MOE, China（高性能软件技术关键实验室（PKU），教育部，中国）；School of Computer Science, Peking University, China（北京大学计算机学院，中国）；The University of Hong Kong（香港大学）；Shanghai AI Lab（上海人工智能实验室）
Lie to Me: How Faithful Is Chain-of-Thought Reasoning in Reasoning Models?
University of Nevada, Las Vegas, Department of Management, Entrepreneurship and Technology, Lee Business School（内华达大学拉斯维加斯分校，管理、创业与技术系，李商学院）；DeepNeuro AI
