Reinforcement Learning from Human Feedback (RLHF) aligns language models to human preferences by employing a singular reward model derived from preference data. However, such an approach overlooks the rich diversity of human preferences inherent in data collected from multiple users. In this work, we first derive an impossibility result of alignment with single reward RLHF, thereby highlighting its insufficiency in representing diverse human preferences. To provide an equitable solution to the problem, we learn a mixture of preference distributions via an expectation-maximization algorithm and propose a MaxMin alignment objective for policy learning inspired by the Egalitarian principle in social choice theory to better represent diverse human preferences. We elucidate the connection of our proposed approach to distributionally robust optimization and general utility RL, thereby highlighting the generality and robustness of our proposed solution. We present comprehensive experimental results on small-scale (GPT-2) and large-scale language models (with Tulu2-7B) and show the efficacy of the proposed approach in the presence of diversity among human preferences. Our algorithm achieves an average improvement of more than 16% in win-rates over conventional RLHF algorithms and improves the win-rate (accuracy) for minority groups by over 33% without compromising the performance of majority groups, showcasing the robustness and fairness of our approach. We remark that our findings in this work are not only limited to language models but also extend to reinforcement learning in general.
Advancing Translation Preference Modeling with RLHF: A Step Towards Cost-Effective Solution
School of Computer Science, Fudan University（复旦大学计算机科学学院）；College of Foreign Languages and Literature, Fudan University（复旦大学外国语言文学学院）；Institute of Modern Languages and Linguistics, Fudan University（复旦大学现代语言与语言学研究所）
S2T-RLHF: Hierarchical Credit Assignment for Stable Preference-Based RLHF
State Key Laboratory for Novel Software Technology, Nanjing University（南京大学计算机软件新技术国家重点实验室）；School of Computer Science and Artificial Intelligence, Zhengzhou University（郑州大学计算机科学与人工智能学院）
Avoiding $\mathbf{exp(R_{max})}$ scaling in RLHF through Preference-based Exploration
Department of Electrical & Computer Engineering（电气与计算机工程系）；Boston University（波士顿大学）；Department of Computer Science（计算机科学系）；Cornell University（康奈尔大学）；Faculty of Computing & Data Sciences（计算与数据科学学院）
Iterative Preference Learning from Human Feedback: Bridging Theory and Practice for RLHF under KL-Constraint
University of Illinois Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校）；Salesforce AI Research（Salesforce AI研究院）；The Hong Kong University of Science and Technology（香港科技大学）
Efficient Preference Poisoning Attack on Offline RLHF
Department of Electrical and Computer Engineering, University of California, Davis, Davis, CA, USA（加州大学戴维斯分校电气与计算机工程系）；Department of Electrical and Computer Engineering, University of Iowa, Iowa City, IA, USA（爱荷华大学电气与计算机工程系）
On the Algorithmic Bias of Aligning Large Language Models with RLHF: Preference Collapse and Matching Regularization
University of Pennsylvania（宾夕法尼亚大学）；The Chinese University of Hong Kong, Shenzhen（香港中文大学（深圳））；National University of Singapore（新加坡国立大学）；Peking University（北京大学）；Joint corresponding authors（联合通讯作者）
UCL Centre for Artificial Intelligence（伦敦大学学院人工智能中心）；Department of Computer Science（计算机科学系）；University College London（伦敦大学学院）；UK AI Security Institute（英国人工智能安全研究所）；Department of Computer Science and Engineering（计算机科学与工程系）；University of Bologna（博洛尼亚大学）
