Tian, C., Lu, Z., Qian, S., Liu, N., Li, P., Jin, L., … Cas, G. (2026). Rectify Evaluation Preference: Improving LLMs’ Critique on Math Reasoning via Perplexity-aware Reinforcement Learning. Proceedings of the AAAI Conference on Artificial Intelligence, 40(39), 33241–33249. https://doi.org/10.1609/aaai.v40i39.40609