1 paper
Felipe Maia Polo, Ronald Xu, Lucas Weber +6
Most popular benchmarks for comparing LLMs rely on a limited set of prompt templates, which may not fully capture the LLMs' abilities and can affect the reproducibility of results…