Awesome RL with Human Feedback / 2026 / arXiv

Pretrain Value, Not Reward: Decoupled Value Policy Optimization

作者：Chenghua Huang, Lu Wang, Fangkai Yang, Pu Zhao, Qingwei Lin, Dongmei Zhang, Saravan Rajmohan · 单位：School of Computer Science, Fudan University · 来源：arXiv · 发布日期：2025-02-24

奖励规范奖励建模偏好学习训练稳定性策略优化
新标签打开 AlphaXiv
Pretrain Value, Not Reward: Decoupled Value Policy Optimization figure
AlphaXiv 中文概览（可滚动查看）