@article{evpoexplainedvariancepolicyoptimizationf, title = {EVPO: Explained Variance Policy Optimization for Adaptive Critic Utilization in LLM Post-Training}, author = {Chengjun Pan and Shichun Liu and Jiahang Lin and Dingwei Zhu and Jiazheng Zhang and Shihan Dou and Songyang Gao and Zhenhua Han and Binghai Wang and Rui Zheng and Xuanjing Huang and Tao Gui and Yansong Feng}, year = {2026}, eprint = {2604.19485}, archivePrefix = {arXiv}, url = {https://arxiv.org/abs/2604.19485}, }