Rethinking KL Regularization in RLHF: From Value Estimation to Gradient Optimization

@misc{liuMon Oct 06 2025 11:59:12 GMT+0000 (Coordinated Universal Time)rethinkingklregularization,
      title={Rethinking KL Regularization in RLHF: From Value Estimation to Gradient Optimization},
      author={Kezhao Liu and Jason Klein Liu and Mingtao Chen and Yiming Liu},
      year={Mon Oct 06 2025 11:59:12 GMT+0000 (Coordinated Universal Time)},
      eprint={2510.01555},
      archivePrefix={arXiv},
      primaryClass={cs.AI},
      url={https://arxiv.org/abs/2510.01555},
}