TY  - RPRT
TI  - Not All Preferences are What You Need for Post-Training: Selective Alignment Strategy for Preference Optimization
AU  - Zhijin Dong
PY  - 2025
UR  - https://arxiv.org/abs/2507.07725
ID  - 2507.07725
ER  -