Efficient Reinforcement Learning for Mathematical Theorem Proving with Compact Language Models. JAAIR. 2026;5(1). Accessed October 27, 2026. https://www.jaair.org/index.php/home/article/view/249