REACT：在语言模型中协同推理和行动.pdfVIP

Published as a conference paper at ICLR 2023 REA C T : SYNERGIZING REASONING AND ACTING IN LANGUAGE MODELS *,1 2 2 2 2 1 2 Shunyu Yao , Jeffrey Zhao , Dian Yu , Nan Du , Izhak Shafran , Karthik Narasimhan , Yuan Cao 1Department of Computer Science, Princeton University 2 Google Research, Brain team 1 {shunyuy,karthikn}@ 2 {jeffreyzhao,dianyu,dunan,izhak,yuancao}@ 3 2 0 2 ABSTRACT r a While large language models (LLMs) have demonstrated impressive performance M across tasks in language understanding and interactive decision making, their abilities for reasoning (e.g. chain-of-thought prompting) and acting (e.g. action 0 plan generation) have primarily been studied as separate topics. In this paper, we 1 explore the use of LLMs to generate both reasoning traces and task-specific actions ] in an interleaved manner, allowing for greater synergy between the two: reasoning L traces help the model induce, track, and update action plans as well as handle C exceptions, while actions allow it to interface with and gather additional information . from external sources such as knowledge bases or environments. We apply our s approach, named ReAct, to a diverse set of language and decision making t