@article{demystifyinggrouprelativepolicyoptimizat, title = {Demystifying Group Relative Policy Optimization: Its Policy Gradient is a U-Statistic}, author = {Hongyi Zhou and Kai Ye and Erhan Xu and Jin Zhu and Ying Yang and Shijin Gong and Chengchun Shi}, year = {2026}, eprint = {2603.01162}, archivePrefix = {arXiv}, url = {https://arxiv.org/abs/2603.01162}, }