Recent advancements in slow thinking reasoning models have shown exceptional performance in complex reasoning tasks. However, these models often exhibit overthinking (generating redundant reasoning steps for simple problems), leading to excessive computational resource usage. While current mitigation strategies uniformly reduce reasoning tokens, they risk degrading performance on challenging tasks that require extended reasoning. This paper introduces Difficulty-Adaptive Slow Thinking (DAST), a novel framework that enables models to autonomously adjust the length of Chain-of-Thought (CoT) based on problem difficulty. We first propose a Token Length Budget (TLB) metric to quantify difficulty, then leverage budget-aware reward shaping and budget preference optimization to implement DAST. DAST penalizes overlong responses for simple tasks while incentivizing sufficient reasoning for complex problems. Experiments on diverse datasets and model scales demonstrate that DAST effectively mitigates overthinking (reducing token usage by over 30\% on average) while preserving reasoning accuracy on complex problems. Our codes and models are available at https://github.com/AnonymousUser0520/AnonymousRepo01.
Mitigating Overthinking in Large Reasoning Models via Difficulty-aware Reinforcement Learning
Laboratory for Artificial Intelligence and New Forms of Education（人工智能与新教育形式实验室）；National Engineering Research Center of Educational Big Data（教育大数据国家工程研究中心）；Faculty of Artificial Intelligence in Education, Central China Normal University（教育人工智能学院，中央财经大学）；College of Electronics and Information Engineering, South-Central Minzu University（电子与信息工程学院，西南民族大学）
School of Computing, National University of Singapore, Singapore（新加坡国立大学计算机学院）；Department of Statistics, University of Oxford, United Kingdom（英国牛津大学统计系）；School of Intelligence Science（智能科学学院）；Technology, Nanjing University, China（技术学院，南京大学，中国）
AdaThinking-E: One-Token Entropy Regulation for Adaptive Thinking
Meituan（美团）；MoE Key Lab of Artificial Intelligence（MoE人工智能重点实验室）；AI Institute（人工智能研究院）；School of Computer Science, Shanghai Jiao Tong University（上海交通大学计算机科学学院）；MAIS&NLPR, Institute of Automation, Chinese Academy of Sciences（中国科学院自动化研究所）
ScaleDiff: Scaling Difficult Problems for Advanced Mathematical Reasoning
Gaoling School of Artificial Intelligence, Renmin University of China（中国人民大学耿丽学校）；OpenDataLab, Shanghai Artificial Intelligence Laboratory（上海人工智能实验室）；Tsinghua University（清华大学）；School of Artificial Intelligence, Wuhan University（武汉大学人工智能学院）
Google Research（谷歌研究）；TReNDS Center, Georgia State University（TReNDS中心，佐治亚州立大学）；Maynooth University（迈诺特大学）；Georgia State University/Georgia Institute of Technology/Emory University Center for Translational Research in Neuroimaging and Data Science (TReNDS Center)（佐治亚州立大学/佐治亚理工学院/埃默里大学神经影像与数据科学转化研究中心（TReNDS中心））；Department of Computer Science, Maynooth University（迈诺特大学计算机科学系）
