Chain-of-Thought (CoT) has substantially empowered Large Language Models (LLMs) to tackle complex reasoning tasks, yet the verbose nature of explicit reasoning steps incurs prohibitive inference latency and computational costs, limiting real-world deployment. While existing compression methods - ranging from self-training to Reinforcement Learning (RL) with length constraints - attempt to mitigate this, they often sacrifice reasoning capability for brevity. We identify a critical failure mode in these approaches: explicitly optimizing for shorter trajectories triggers rapid entropy collapse, which prematurely shrinks the exploration space and stifles the discovery of valid reasoning paths, particularly for challenging questions requiring extensive deduction. To address this issue, we propose Compress responses for Easy questions and Explore Hard ones (CEEH), a difficulty-aware approach to RL-based efficient reasoning. CEEH dynamically assesses instance difficulty to apply selective entropy regularization: it preserves a diverse search space for currently hard questions to ensure robustness, while permitting aggressive compression on easier instances where the reasoning path is well-established. In addition, we introduce a dynamic optimal-length penalty anchored to the historically shortest correct response, which effectively counteracts entropy-induced length inflation and stabilizes the reward signal. Across six reasoning benchmarks, CEEH consistently reduces response length while maintaining accuracy comparable to the base model, and improves Pass@k relative to length-only optimization.
The Easy, the Hard, and the Learnable: Confidence and Difficulty-Adaptive Policy Optimization for LLM Reasoning
TMLR Group, Department of Computer Science, Hong Kong Baptist University(香港 Baptist 大学计算机科学系 TMLR 组);Stanford University(斯坦福大学);Sydney AI Centre, The University of Sydney(悉尼大学人工智能中心)
Easy2Hard-Bench: Standardized Difficulty Labels for Profiling LLM Performance and Generalization
University of Maryland(马里兰大学);University of Waterloo(滑铁卢大学);Carnegie Mellon University(卡内基梅隆大学);California Institute of Technology(加州理工学院);Microsoft(微软);Capital One
CHUCKLE -- When Humans Teach AI To Learn Emotions The Easy Way
New York Institute of Technology, Department of Computer Science(纽约理工学院计算机科学系);New York University, Electrical and Computer Engineering Department(纽约大学电气与计算机工程系)
Learning How Much to Think: Difficulty-Aware Dynamic MoEs for Graph Node Classification
Institute of Cyberspace Security, Zhejiang University of Technology(浙江工业大学网络空间安全研究院);Binjiang Cyberspace Security Institute of ZJUT(浙江工业大学滨江网络空间安全研究院);Soovar Technologies Co., Ltd.(Soovar科技有限公司);Department of Electronic and Computer Engineering, Hong Kong University of Science and Technology(香港科技大学电子及计算机工程学系)
Predicting respondent difficulty in web surveys: A machine-learning approach based on mouse movement features
Humboldt-Universität zu Berlin(柏林洪堡大学);University of Mannheim(曼海姆大学);University of Koblenz-Landau(科布伦茨-兰道大学);University of Maryland(马里兰大学);Institute for Employment Research(就业研究所)