@article{approximationoflogpartitionfunctioninpol, title = {Approximation of Log-Partition Function in Policy Mirror Descent Induces Implicit Regularization for LLM Post-Training}, author = {Zhenghao Xu and Qin Lu and Changlong Yu and Tuo Zhao}, year = {2026}, eprint = {2602.05933}, archivePrefix = {arXiv}, url = {https://arxiv.org/abs/2602.05933}, }