TY  - RPRT
TI  - Multi-dimensional Preference Alignment by Conditioning Reward Itself
AU  - Jiho Jang
AU  - Jinyoung Kim
AU  - Kyungjune Baek
AU  - Nojun Kwak
PY  - 2026
UR  - https://arxiv.org/abs/2512.10237
ID  - 2512.10237
ER  -