Princeton and Google AI researchers propose ReAct: an effective artificial intelligence method to synergize reasoning and action in large language models
Although large language models (LLMs) have shown amazing performance on tasks involving interactive decision-making and language comprehension, their reasoning (such as chain-of-thought prompting) and action skills (such as the generation of action plans) have mainly been studied as separate topics. Recent work focuses on translating textual contexts into textual actions using internal knowledge of the …