Awesome RL with Human Feedback / 2025 / arXiv

Equilibrate RLHF: Towards Balancing Helpfulness-Safety Trade-off in Large Language Models

作者：Yingshui Tan, Yilei Jiang, Yanshi Li, Jiaheng Liu, Xingyuan Bu, Wenbo Su, Xiangyu Yue, Xiaoyong Zhu, Bo Zheng · 单位：Alibaba Group · 来源：arXiv · 发布日期：2025

安全对齐对齐目标设计框架/系统
新标签打开 AlphaXiv
Equilibrate RLHF: Towards Balancing Helpfulness-Safety Trade-off in Large Language Models figure
AlphaXiv 中文概览（可滚动查看）