An approach that interleaves language-model reasoning with actions in an external environment.
登录后收藏ReAct combines reasoning traces and task-oriented actions in an alternating process. Actions can retrieve evidence or interact with an environment, while subsequent reasoning can update the plan using those observations. The paper evaluates this approach on question answering, fact verification, and interactive decision-making tasks.
Research paper first submitted to arXiv on 2022-10-06.
Relevant when understanding tool-using agents. Distinguish the ReAct research method from the unrelated React interface library. Catalog note: this is a source-verified reference entry, not a measured current-trend signal. No search-volume or growth claim is made.
专业版打开当天的 prioritize / watch / skip 表,包括这里还不能公开的新名字。
导入的日报里还没有这个名字。