Evaluating Creativity in Large Language Models - arxiv.org