A Generative AI Engineer is developing an LLM to summarize articles into a specific poetic form, like a haiku. The initial outputs from the model do not exhibit the intended tone or style. Which of the following strategies is LEAST likely to enhance the LLM's output to meet the desired style?