Diverse LLMs or Diverse Question Interpretations? That is the Ensembling Question

Rafael Rosales, Santiago Miret

Effectively leveraging diversity has been shown to improve performance for various machine learning models, including large language models (LLMs). However, determining the most ef…