Awesome RL with Human Feedback / 2024 / arXiv

Zeroth-Order Policy Gradient for Reinforcement Learning from Human Feedback without Reward Inference

作者：Qining Zhang, Lei Ying · 单位：University of Michigan, Ann Arbor · 来源：arXiv · 发布日期：2024

奖励建模偏好学习策略优化无奖励对齐策略梯度人类反馈强化学习
新标签打开 AlphaXiv
Zeroth-Order Policy Gradient for Reinforcement Learning from Human Feedback without Reward Inference figure
AlphaXiv 中文概览（可滚动查看）