@article{
wu2026multistep,
title={Multi-Step Alignment as Markov Games: An Optimistic Online Mirror Descent Approach with Convergence Guarantees},
author={Yongtao Wu and Luca Viano and Kimon Antonakopoulos and Yihang Chen and Zhenyu Zhu and Quanquan Gu and Volkan Cevher},
journal={Transactions on Machine Learning Research},
issn={2835-8856},
year={2026},
url={https://openreview.net/forum?id=ZWZKaqZCy0},
note={}
}