<ul><li>A new research paper introduces CoLa, a self-guided learning paradigm for training automated language guides.</li><li>CoLa aims to simulate human guides in problem-solving scenarios using large language models (LLMs).</li><li>Empirical results show that CoLa consistently outperforms competitive approaches in various language tasks.</li><li>Automated guides in CoLa demonstrate superior adaptability and reasoning strategies compared to human guides.</li></ul>

CoLa -- Learning to Interactively Collaborate with Large LMs

Discover more