Tag
This paper proposes Dynamic Retrieval-based Policy Generation (DRPG), a framework that uses memory retrieval and environment feedback to dynamically generate policies for continual improvement of large language models across various tasks.