ClozeMath: Improving Mathematical Reasoning in Language Models by Learning to Fill Equations

Quang Hieu Pham, Thuy Duong Nguyen, Tung Pham +2

The capabilities of large language models (LLMs) have been enhanced by training on data that reflects human thought processes, such as the Chain-of-Thought format. However, evidenc…