The Reflexion strategy[54] constructs an agent that learns in excess of a number of episodes. At the conclusion of Every single episode, the LLM is presented the document of the episode, and prompted to think up "classes uncovered", which would assistance it execute superior in a subsequent episode. These "classes figured out" are presented for the