AutoPlan: Automatic Planning of Interactive Decision-Making Tasks With Large Language Models
Siqi Ouyang, Lei Li · 2023
Recent large language models (LLMs) are promising for making decisions in grounded environments.However, LLMs frequently fail in complex decision-making tasks due to the misalignment between the pre-trained knowledge in LLMs and the actual rules in the environment.Existing methods require either costly gradient computation or lengthy in-context demonstrations.In this paper, we propose AutoPlan, an approach to guide LLMbased agents to accomplish interactive decisionmaking tasks.AutoPlan augments the LLM prompt with a task-solving plan and optimizes it through iterative experience collection and reflection.Our experiments show that Auto-Plan, though using no in-context demonstrations, achieves success rates on par with the baselines using human-written demonstrations on ALFWorld and even outperforms them by 8% on HotpotQA.The code is available at https://github.com/owaski/AutoPlan.