TY  - RPRT
TI  - How Sampling Shapes LLM Alignment: From One-Shot Optima to Iterative Dynamics
AU  - Yurong Chen
AU  - Yu He
AU  - Michael I. Jordan
AU  - Fan Yao
PY  - 2026
UR  - https://arxiv.org/abs/2602.12180
ID  - 2602.12180
ER  -