Planning by Automatic Prompt Engineering for Large Language Models Agents
The paper proposes a novel method, REPROMPT, for optimizing the step-by-step instructions in the prompt of LLM agents based on the chat history obtained from interactions with LLM agents.
BibTex:
