@article{bandpobridgingtrustregionsandratioclippi,
  title = {BandPO: Bridging Trust Regions and Ratio Clipping via Probability-Aware Bounds for LLM Reinforcement Learning},
  author = {Yuan Li and Bo Wang and Yufei Gao and Yuqian Yao and Xinyuan Wang and Zhangyue Yin and Xipeng Qiu},
  year = {2026},
  eprint = {2603.04918},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2603.04918},
}