1 paper
Zhanke Zhou, Zhaocheng Zhu, Xuan Li +5
Numerous applications of large language models (LLMs) rely on their ability to perform step-by-step reasoning. However, the reasoning behavior of LLMs remains poorly understood, po…