@article{starpostabilityaugmentedreinforcementpol, title = {StaRPO: Stability-Augmented Reinforcement Policy Optimization}, author = {Jinghan Zhang and Fengran Mo and Tharindu Cyril Weerasooriya and Ruimin Dai and Xiaoyan Han and Yanjie Fu and Dakuo Wang and Kunpeng Liu}, year = {2026}, eprint = {2604.08905}, archivePrefix = {arXiv}, url = {https://arxiv.org/abs/2604.08905}, }