尽管链式思维(CoT)提示方法提升了大语言模型(LLM)的推理能力,但在一致性、准确性和自我纠正方面仍存在挑战,尤其是在复杂或涉及伦理敏感的任务中。现有的单一维度反思方法改进有限。我们提出MyGO Poly-Reflective Chain-of-Thought(PR-CoT),一种新的方法,采用结构化的多视角反思。在初始CoT之后,PR-CoT引导LLM从多个预定义的角度自我评估其推理:逻辑一致性、信息完整性、偏见/伦理以及替代解决方案。该过程完全通过提示工程实现,将初始CoT精炼为更稳健和准确的最终答案,而无需模型重新训练。在算术、常识、伦理决策和逻辑谜题等任务中,使用GPT-three point five和GPT-four模型进行实验,证明了PR-CoT的优越性能。它在逻辑一致性和错误纠正方面显著优于传统CoT和现有反思方法,在伦理决策等细致领域也有明显提升。消融研究、人类评估和定性分析进一步验证了每个反思视角的贡献以及多维反思范式在促进更可靠LLM推理中的整体有效性。
英文摘要:
While Chain-of-Thought (CoT) prompting advances LLM reasoning, challenges persist in consistency, accuracy, and self-correction, especially for complex or ethically sensitive tasks. Existing single-dimensional reflection methods offer insufficient improvements. We propose MyGO Poly-Reflective Chain-of-Thought (PR-CoT), a novel methodology employing structured multi-perspective reflection. After initial CoT, PR-CoT guides the LLM to self-assess its reasoning across multiple predefined angles: logical consistency, information completeness, biases/ethics, and alternative solutions. Implemented purely via prompt engineering, this process refines the initial CoT into a more robust and accurate final answer without model retraining. Experiments across arithmetic, commonsense, ethical decision-making, and logical puzzles, using GPT-three point five and GPT-four models, demonstrate PR-CoT's superior performance. It significantly outperforms traditional CoT and existing reflection methods in logical consistency and error correction, with notable gains in nuanced domains like ethical decision-making. Ablation studies, human evaluations, and qualitative analyses further validate the contribution of each reflection perspective and the overall efficacy of our poly-reflective paradigm in fostering more reliable LLM reasoning.
Instruct-of-Reflection: Enhancing Large Language Models Iterative Reflection Capabilities via Dynamic-Meta Instruction
Beijing University of Posts and Telecommunications(北京邮电大学);AI Lab of Lenovo Research(联想研究院AI实验室);Tianjin University(天津大学);The Hong Kong University of Science and Technology(香港科技大学)
Enhancing Long-Chain Reasoning Distillation through Error-Aware Self-Reflection
School of Computer Science and Engineering, Northeastern University, China(东北大学计算机科学与工程学院);Department of Computer Science and Technology, Institute for AI, Tsinghua University, China(清华大学人工智能研究院计算机科学与技术系);School of Computer Science, Beijing University of Posts and Telecommunications, China(北京邮电大学计算机科学学院)
RoT: Enhancing Large Language Models with Reflection on Search Trees
School of Information Science and Technology, ShanghaiTech(上海科技大学信息科学与技术学院);Shanghai Engineering Research Center of Intelligent Vision and Imaging(上海智能视觉与成像工程技术研究中心)
PerspAct: Enhancing LLM Situated Collaboration Skills through Perspective Taking and Active Vision
University of Milan-Bicocca(米兰-比科卡大学);University of Naples Federico II(那不勒斯费德里科二世大学);Oversonic Robotics(Oversonic机器人公司);University of Essex(埃塞克斯大学);TUM School of Social Sciences and Technology(慕尼黑技术大学社会科学与技术学院)
Look Again, Think Slowly: Enhancing Visual Reflection in Vision-Language Models
Institute of Automation, Chinese Academy of Sciences(中国科学院自动化研究所);School of Artificial Intelligence, University of Chinese Academy of Sciences(中国科学院大学人工智能学院);Wuhan AI Research(武汉人工智能研究院)
Enhancing Textual Personality Detection toward Social Media: Integrating Long-term and Short-term Perspectives
School of Computer Science and Technology, Dalian University of Technology, Dalian, China(计算机科学与技术学院,大连理工大学,大连,中国);Criminal Investigation Police University of China, Shenyang, China(中国刑事侦查警察大学,沈阳,中国);School of Foreign Languages, Dalian University of Technology, Dalian, China(外语学院,大连理工大学,大连,中国)