原创力文档- 1、原创力文档（book118）网站文档一经付费（服务费），不意味着购买了该文档的版权，仅供个人/单位学习、研究之用，不得用于商业用途，未经授权，严禁复制、发行、汇编、翻译或者网络传播等，侵权必究。。
- 2、本站所有内容均由合作方或网友上传，本站不对文档的完整性、权威性及其观点立场正确性做任何保证或承诺！文档内容仅供研究参考，付费前请自行鉴别。如您付费，意味着您自己接受本站规则且自行承担风险，本站不退款、不进行额外附加服务；查看《如何避免下载的几个坑》。如果您已付费下载过本站文档，您可以点击 这里二次下载。
- 3、如文档侵犯商业秘密、侵犯著作权、侵犯人身权等，请点击“版权申诉”（推荐），也可以打举报电话：400-050-0827(电话支持时间：9:00-18:30)。
- 4、该文档为VIP文档，如果想要下载，成为VIP会员后，下载免费。
- 5、成为VIP后，下载本文档将扣除1次下载权益。下载后，不支持退款、换文档。如有疑问请联系我们。
- 6、成为VIP后，您将拥有八大权益，权益包括：VIP文档下载权益、阅读免打扰、文档格式转换、高级专利检索、专属身份标志、高级客服、多端互通、版权登记。
- 7、VIP文档为合作方或网友上传，每下载1次， 网站将根据用户上传文档的质量评分、类型等，对文档贡献者给予高额补贴、流量扶持。如果你也想贡献VIP文档。上传文档
CausalSufficiencyandNecessityImproves
Chain-of-ThoughtReasoning
1,6∗2∗34
XiangningYu,ZhuohanWang,LinyiYang,HaoxuanLi,
51†3†6
AnjieLiu,XiaoXue,JunWang,MengyueYang
1TianjinUniversity2CityUniversityofHongKong
3UniversityCollegeLondon4PekingUniversity
55TheHongKongUniversityofScienceandTechnology(Guangzhou)6UniversityofBristol
2
0
2
nAbstract
u
J
Chain-of-Thought(CoT)promptingplaysanindispensableroleinendowinglarge
1languagemodels(LLMs)withcomplexreasoningcapabilities.However,CoT
1currentlyfacestwofundamentalchallenges:(1)Sufficiency,whichensuresthat
thegeneratedintermediateinferencestepscomprehensivelycoverandsubstantiate
]
Lthefinalconclusion;and(2)Necessity,whichidentifiestheinferencestepsthatare
Ctrulyindispensableforthesoundnessoftheresultinganswer.Weproposeacausal
.frameworkthatcharacterizesCoTreasoningthroughtheduallensesofsufficiency
sandnecessity.IncorporatingcausalProbabilityofSufficiencyandNecessityallows
cusnotonlytodeterminewhichstepsarelogicallysufficientornecessary
您可能关注的文档
- 人工智能论文英文版-Purifying Shampoo：Investigating Shampoo’s Heuristics by Decomposing its Preconditioner.pdf
- 人工智能论文英文版-Hive is PSPACE-Hard.pdf
- 人工智能论文英文版-DiagNet：Detecting Objects using Diagonal Constraints on Adjacency Matrix of Graph Neural Network.pdf
- 人工智能论文英文版-Attention Head Embeddings with Trainable Deep Kernels for Hallucination Detection in LLMs.pdf
- 人工智能论文英文版-The Emergence of Abstract Thought in Large Language Models Beyond Any Language.pdf
- 人工智能论文英文版-PersonaLens：A Benchmark for Personalization Evaluation in Conversational AI Assistants.pdf
- 人工智能论文英文版-VERIF：Verification Engineering for Reinforcement Learning in Instruction Following.pdf
- 人工智能论文英文版-Query-Focused Retrieval Heads Improve Long-Context Reasoning and Re-ranking.pdf
- 人工智能论文英文版-Resa：Transparent Reasoning Models via SAEs.pdf
- 人工智能论文英文版-When Detection Fails：The Power of Fine-Tuned Models to Generate Human-Like Social Media Text.pdf
- 人工智能论文英文版-Step-by-step Instructions and a Simple Tabular Output Format Improve the Dependency Parsing Accuracy of LLMs.pdf
- 人工智能论文英文版-Large Language Models for Toxic Language Detection in Low-Resource Balkan Languages.pdf
- 人工智能论文英文版-From Judgment to Interference：Early Stopping LLM Harmful Outputs via Streaming Content Monitoring.pdf
文档评论（0）