计算机科学

计算机科学 ›› 2025, Vol. 52 ›› Issue (1): 56-64.doi: 10.11896/jsjkx.240700172

• 大语言模型技术研究及应用 • 上一篇    下一篇

提示学习中思维链生成和增强方法综述

郑明琪, 陈晓慧, 刘冰, 张兵, 张然   

  1. 信息工程大学数据与目标工程学院 郑州 450000
  • 收稿日期:2024-07-25 修回日期:2024-10-23 出版日期:2025-01-15 发布日期:2025-01-09
  • 通讯作者: 陈晓慧(cxh_vrlab@163.com)
  • 作者简介:(2602770165@qq.com)
  • 基金资助:
    国家自然科学基金(42371438,42401501)

Survey of Chain-of-Thought Generation and Enhancement Methods in Prompt Learning

ZHENG Mingqi, CHEN Xiaohui, LIU Bing, ZHANG Bing, ZHANG Ran   

  1. Institute of Data and Target Engineering,Information Engineering University,Zhengzhou 450000,China
  • Received:2024-07-25 Revised:2024-10-23 Online:2025-01-15 Published:2025-01-09
  • About author:ZHENG Mingqi,born in 2001,postgraduate.Her main research interests include artificial intelligence and big data analytics.
    CHEN Xiaohui,born in 1983,Ph.D,professor.Her main research interests include human-computer interactive intelligence and so on.
  • Supported by:
    National Natural Science Foundation of China(42371438,42401501).

PDF (PC)

267

摘要： 大语言模型凭借其卓越的语言理解和文本生成能力,在多个领域取得了突破性进展。尽管如此,其在处理复杂推理任务时的表现往往不尽如人意,准确率的提升空间依然巨大。针对这一挑战,学术界提出了思维链策略,这是一种创新的方法,通过让模型生成推理过程来增强模型的推理性能。文中通过全面梳理和深入分析现有的思维链研究,不仅总结了其核心概念和结构框架,还详细探讨了推理生成方法和增强方法。进一步对思维链在不同任务场景中的应用进行了广泛探讨,展示了思维链在提升模型性能方面的潜力。同时,也对思维链的局限性进行了批判性分析,指出了思维链方法存在的不足。最后,对思维链的未来发展进行了前瞻性展望,旨在为思维链未来的研究方向提供指导,并为该领域的研究者提供有价值的参考和启示。

关键词: 思维链, 大语言模型, 提示学习, 推理生成, 推理增强

Abstract: Large language models have made breakthroughs in several domains due to their superior language understanding and text generation capabilities.However,their performance in handling complex reasoning tasks is not very good and the accuracy needs to be improved.As a result,academics have proposed chain of thought(CoT),an innovative approach that aims to enhance the reasoning performance of models by allowing them to generate reasoning processes.In this paper,by comprehensively combing and deeply analyzing the existing research on CoT,we not only summarize its concepts and structural framework,but also explore the inference generation method and enhancement method in detail.The application of CoT in different task scenarios is further extensively explored,demonstrating the potential of CoT in enhancing model performance.At the same time,this paper also critically analyzes the limitations of CoT.Finally,this paper provides a prospective outlook on the future development of the chain-of-thinking strategy,aiming to provide guidance on the future research direction of CoT and to provide valuable references and insights for researchers.

Key words: Chain of thought, Large language model, Prompt learning, Reasoning generation, Reasoning enhancement

中图分类号: 

  • TP301
[1]POURKAMALI N,SHARIFI S E.Machine Translation with Large Language Models:Prompt Engineering for Persian,English,and Russian Directions [J].arXiv:2401.08429,2024.
[2]ANAPURE A,DHAMANE S,DHAGE S,et al.EmbodiedEpistemology:A Meta-Cognitive Exploration of Chatbot-Enabled Document Analysis[C]//International Conference on Evolutionary Algorithms and Soft Computing Techniques(EASCT).2023:1-6.
[3]ZHAO H Y,CHEN H J,YANG F,et al.Explainability forLarge Language Models:A Survey[J].ACM Transactions on Intelligent Systems and Technology,2024,15(2):1-38.
[4]LESTER B,AL-RFOU R,CONSTANT N.The Power of Scale for Parameter-Efficient Prompt Tuning[C]//Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing.Online and Punta Cana:Association for Computational Linguistics,2021:3045-3059.
[5]JIN Z,LIU J,LYU Z,et al.Can Large Language Models Infer Causation from Correlation[J].arXiv:2306.05836,2024.
[6]CHU Z,CHEN J,CHEN Q L,et al.A Survey ofChain of Thought Reasoning:Advances,Frontiers and Future[J].arXiv:2309.15402,2023.
[7]XIA Y,WANG R,LIU X,et al.Beyond Chain-of-Thought:A Survey of Chain-of-X Paradigms for LLMs[J].arXiv 2404.15676,2024.
[8]YU Z H,HE L,WU Z,et al.Towards Better Chain-of-Thought Prompting Strategies:A Survey[J].arXiv:2310.04959,2023.
[9]BROWN T B,MANN B,RYDER N,et al.Language Models are Few-Shot Learners[C]//Proceedings of the 34th International Conference on Neural Information Processing Systems.United States:Curran Associates Inc.Curran Associates Inc.,2020:1877-1901.
[10]CHEN B H,ZHANG Z F,LANGRENÉ N,et al.Unleashing the potential of prompt engineering in Large Language Models:a comprehensive review[J].arXiv:2310.14735,2023.
[11]KOJIMA T,SHANE GU S X,REID M,et al.Large Language Models are Zero-Shot Reasoners[J].arXiv:2205.11916,2022.
[12]DEL M,FISHEL M.DETECTIVE T,et al.A ChallengingBenchmark for Deep Abductive Reasoning in Foundation Mo-dels[J].arXiv:2212.10114,2022.
[13]WEI J,WANG X Z,SCHUURMANS D,et al.Chain-of-Thought Prompting Elicits Reasoning in Large Language Mo-dels[J].arXiv:2201.11903,2023.
[14]YAO S Y,YU D,ZHAO J,et al.Tree of Thoughts:Deliberate Problem Solving with Large Language Models[C]//Proceedings of the 37th International Conference on Neural Information Processing Systems.United States:Curran Associates Inc.Curran Associates Inc.2023:11809-11822.
[15]BESTA M,BLACH N,KUBICEK A,et al.Graph of Thoughts:Solving Elaborate Problems with Large Language Model[J].AAAI Conference on Artificial Intelligence,2024,38(16):17682-17690.
[16]WANG L,XU W Y,LAN Y H,et al.Plan-and-Solve Promp-ting:Improving Zero-Shot Chain-of-Thought Reasoning by Large Language Models[C]//Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics(Vo-lume 1:Long Papers).Toronto:Association for Computational Linguistics,2023:2609-2634.
[17]SHUM K,DIAO S,ZHANG T,et al.Automatic Prompt Augmentation and Selection with Chain-of-Thought from Labeled Data[C]//Conference on Empirical Methods in Natural Language Processing,Findings of the Association for Computational Linguistics.EMNLP,2023:12113-12139.
[18]ZHUO S Z,ASTON Z,MU L,et al.Auto-matic Chain ofThought Prompting in Large Language Models[J].arXiv:2210.03493,2023.
[19]DIAO S Z,WANG P,LIN Y,et al.Active-Prompt with Chain-of-Thought for Large Language Model[J].arXiv:2302.12246,2023.
[20]WANG X Z,WEI J,SCHUURMANS D,et al.Self-Consistency Improves Chain of Thought Reasoning in Language Models[J].arXiv:2203.11171,2022.
[21]CHEN W,MA X,WANG X Y,et al.Program of ThoughtsPrompting:Disentangling Computation from Reasoning for Numerical Reasoning Tasks[J].arXiv:2211.12588,2022.
[22]GAO L,MADAAN A,ZHOU S,et al.PAL:Program-aidedLanguage Models[J].arXiv:2211.10435,2023.
[23]KHOT T,TRIVEDI H,FINLAYSON M,et al.DecomposedPrompting:A Modular Approach for Solving Complex Task[J].arXiv:2210.02406,2023.
[24]LING Z,FANG Y H,LI X L,et al.Deductive Verification of Chain-of-Thought Reasoning[C]//Proceedings of the 37th International Conference on Neural Information Processing Systems,United States:Curran Associates Inc.Curran Associates Inc.,2023:36407-36433.
[25]YORAN O,WOLFSON T,BOGIN B,et al.Answering Questions by Meta-Reasoning over Multiple Chains of Thought[C]//Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing.Kalika Bali:Association for Computational Linguistics,2023:5942-5966.
[26]SHAO H,LI D,WEN H W,et al.Language Is Not All YouNeed:Aligning Perception with Language Models[J],arXiv:2302.14045,2023.
[27]ZHANG Z S,ZHANG A,LI M,et al.Multimodal Chain-of-Thought Reasoning in Language Models[J].arXiv:2302,00923,2023.
[28]ZHAO R C,LI X X,JOTY S,et al.Verify-and-Edit:A Know-ledge-Enhanced Chain-of-Thought Framework[C]//Procee-dings of the 61st Annual Meeting of the Association for Computa-tional Linguistics(Volume 1:Long Papers).Toronto:Association for Computational Linguistics,2023:5823-5840.
[29]LUCIE C M,JONATHAN M,JAKUB A,et al.Teaching Small Language Models to Reason[C]//Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics(Volume 2:Short Papers).Toronto:Association for Computational Linguistics,2022:1773-1781.
[30]NAMGYU H,LAURA S,YUN S Y,et al.Large LanguageModels Are Reasoning Teachers[C]//Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics(Volume 1:Long Papers).Toronto:Association for Computational Linguistics,2022:14852-14882.
[31]MILES T,JULIAN M,ETHAN P,et al.Language ModelsDon’t Always Say What They Think:Unfaithful Explanations in Chain-of-Thought Prompting[J].arXiv:2305.04388,2023.
[32]OMAR S,ZHANG H X,WILLIAM H,et al.On SecondThought,Let's Not Think Step by Step! Bias and Toxicity in Zero-Shot Reasoning[C]//Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics(Vo-lume 1:Long Papers).Toronto:Association for Computational Linguistics,2022:4454-4470.
[33]WENHAO Y,CHENGUANG Z,ZAITANG L,et al.A Survey of Knowledge-enhanced Text Generation[J].ACM Computing Surveys,2022,54(11s):1-38.
[34]WU D J,ZHANG J,HUANG X M.Chain of Thought Promp-ting Elicits Knowledge Augmentation[C]//Findings of the Association for Computational Linguistics:ACL 2023.Toronto:Association for Computational Linguistics,2023:6519-6534.
[35]MOMCHIL H,ARNAV A,PRESLAV N,et al.Few-ShotCross-Lingual Stance Detection with Sentiment-Based Pre-trai-ning[J].Proceedings of the AAAI Conference on Artificial Intelligence,2022,36(10):10729-10737.
[36]ALDAYEL A,MAGDY W.Stance detection on social media:State of the art and trends[J].Information Processing and Ma-nagement,2021,58(4):102597.
[37]ZHANG B W,XIANGHUA F,DAIJUN D,et al.Investigating Chain-of-thought with ChatGPT for Stance Detection on Social Media[J].arXiv:2304.03087,2023.
[38]JOSEPH G,OMAR S,SARAH P.Chain-of-Thought Embed-dings for Stance Detection on Social Media.[C]//Findings of the Association for Computational Linguistics:EMNLP 2023.Kalika Bali:Association for Computational Linguistics,2023:4154-4161.
[39]DING D J,FU X H,PENG X J,et al.Leveraging Chain-of-Thought to Enhance Stance Detection with Prompt-Tuning[J].Mathematics,2024,12(4):568.
[40]SOAM D,THAKUR S.Sentiment Analysis Using Deep Lear-ning:A Comparative Study[J].Electronics,2023,9(3):483.
[41]MAROUANE B,MOHAMMED K,ABDERRAHIM B H.Acomprehensive survey on sentiment analysis:Approaches,challenges and trends[J].Knowledge-based Systems,2021,226:107134.
[42]ZHANG W X,DENG Y,LIU B,et al.Sentiment Analysis in the Era of Large Language Models:A Reality Check[C]//Findings of the Association for Computational Linguistics:NAACL 2024.Mexico City:Association for Computational Linguistics,2023:3881-3906.
[43]HAO F,BOBO L,QIAN L,et al.Reasoning Implicit Sentimentwith Chain-of-Thought Prompting[C]//Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics(Volume 2:Short Papers).Toronto:Association for Computational Linguistics,2023:1171-1182.
[44]LI Z J,CHEN G W,SHAO R,et al.Enhancing Emotional Ge-neration Capability of Large Language Models via Emotional Chain-of-Thought[J].arXiv:2401.06836,2024.
[45]XU G,CHEN X L,LU P,et al.AGCVT-prompt for sentiment classification:Automatically generating chain of thought and verbalizer in prompt learning[J].Engineering Applications of Artificial Intelligence,2024,132:107907.
[46]ANUSHKA G,DIKSHA C,ANJUM,et al.Automated NewsSummarization Using Transformers[M]//Sustainable Advanced Computing,2022.
[47]XIAO L,CHEN X L.Enhancing LLM with Evlutionary FineTuning for News Summary Generation[J].arXiv:2307.02839,2023.
[48]WANG Y,ZHANG Z S,WANG R.Element-aware Summarization with Large Language Models:Expert-aligned Evaluation and Chain-of-Thought Method[C]//Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics(Volume 1:Long Papers).Toronto:Association for Computational Linguistics,2023:8640-8665.
[49]ZHANG Y B,GAO S X,HUANG Y X,et al.3A-COT:an attend-arrange-abstract chain-of-thought for multi-document summarization[J/OL].International Journal of Machine Learning and Cybernetics,2024:1-19.https://link.springer.com/article/10.1007/s13042-024-02225-0.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
渝ICP备16013121号-4
地址：重庆市渝北区洪湖西路18号    邮编：401121  
电话：023-63500828（编辑部） 023-67039612（会议/专题） 023-67039623（财务/发票）
E-mail：jsjkx12@163.com
本系统由北京玛格泰克科技发展有限公司设计开发