@article{debiasingrewardmodelsviacausallymotivate, title = {Debiasing Reward Models via Causally Motivated Inference-Time Intervention}, author = {Kazutoshi Shinoda and Kosuke Nishida and Kyosuke Nishida}, year = {2026}, eprint = {2604.27495}, archivePrefix = {arXiv}, url = {https://arxiv.org/abs/2604.27495}, }