English
Related papers

Related papers: Mitigating Length Bias in RLHF through a Causal Le…

200 papers

Reinforcement Learning from Human Feedback (RLHF) is increasingly used to align large language models (LLMs) with human preferences. However, the effectiveness of RLHF in addressing underlying biases remains unclear. This study investigates…

Computation and Language · Computer Science 2025-03-13 Logan Barnhart , Reza Akbarian Bafghi , Stephen Becker , Maziar Raissi

Reinforcement learning with human feedback (RLHF) is an emerging paradigm to align models with human preferences. Typically, RLHF aggregates preferences from multiple individuals who have diverse viewpoints that may conflict with each…

Machine Learning · Computer Science 2024-03-11 Huiying Zhong , Zhun Deng , Weijie J. Su , Zhiwei Steven Wu , Linjun Zhang

With the development of large language models (LLMs), striking a balance between the performance and safety of AI systems has never been more critical. However, the inherent tension between the objectives of helpfulness and harmlessness…

Artificial Intelligence · Computer Science 2023-10-20 Josef Dai , Xuehai Pan , Ruiyang Sun , Jiaming Ji , Xinbo Xu , Mickel Liu , Yizhou Wang , Yaodong Yang

Pre-trained Language Models (LMs) exhibit strong zero-shot and in-context learning capabilities; however, their behaviors are often difficult to control. By utilizing Reinforcement Learning from Human Feedback (RLHF), it is possible to…

Computation and Language · Computer Science 2024-05-31 Avelina Asada Hadji-Kyriacou , Ognjen Arandjelovic

Recent advances at the intersection of reinforcement learning (RL) and visual intelligence have enabled agents that not only perceive complex visual scenes but also reason, generate, and act within them. This survey offers a critical and…

Computer Vision and Pattern Recognition · Computer Science 2025-12-24 Weijia Wu , Chen Gao , Joya Chen , Kevin Qinghong Lin , Qingwei Meng , Yiming Zhang , Yuke Qiu , Hong Zhou , Mike Zheng Shou

Current reinforcement learning from human feedback (RLHF) pipelines for large language model (LLM) alignment typically assign scalar rewards to sequences, using the final token as a surrogate indicator for the quality of the entire…

Machine Learning · Computer Science 2025-04-24 Ryan Koo , Ian Yang , Vipul Raheja , Mingyi Hong , Kwang-Sung Jun , Dongyeop Kang

Personalizing large language models (LLMs) to accommodate diverse user preferences is essential for enhancing alignment and user satisfaction. Traditional reinforcement learning from human feedback (RLHF) approaches often rely on monolithic…

Machine Learning · Computer Science 2025-04-22 Avinandan Bose , Zhihan Xiong , Yuejie Chi , Simon Shaolei Du , Lin Xiao , Maryam Fazel

Reinforcement learning from human feedback (RLHF) offers a promising approach to aligning large language models (LLMs) with human preferences. Typically, a reward model is trained or supplied to act as a proxy for humans in evaluating…

Computation and Language · Computer Science 2025-09-12 Jiahui Li , Lin Li , Tai-wei Chang , Kun Kuang , Long Chen , Jun Zhou , Cheng Yang

Reinforcement Learning with Verifiable Rewards (RLVR) replaces costly human labeling with automated verifiers. To reduce verifier hacking, many RLVR systems binarize rewards to $\{0,1\}$, but imperfect verifiers inevitably introduce…

Machine Learning · Computer Science 2026-05-25 Xin-Qiang Cai , Wei Wang , Feng Liu , Tongliang Liu , Gang Niu , Masashi Sugiyama

To ensure that large language model (LLM) responses are helpful and non-toxic, a reward model trained on human preference data is usually used. LLM responses with high rewards are then selected through best-of-$n$ (BoN) sampling or the LLM…

Machine Learning · Computer Science 2024-07-04 Adam X. Yang , Maxime Robeyns , Thomas Coste , Zhengyan Shi , Jun Wang , Haitham Bou-Ammar , Laurence Aitchison

Recent advancements in Large Language Models (LLMs) have garnered wide attention and led to successful products such as ChatGPT and GPT-4. Their proficiency in adhering to instructions and delivering harmless, helpful, and honest (3H)…

Machine Learning · Computer Science 2023-10-11 Hao Sun

Reinforcement Learning from Human Feedback (RLHF) is a crucial technique for aligning language models with human preferences, playing a pivotal role in the success of conversational models like GPT-4, ChatGPT, and Llama 2. A core challenge…

Computation and Language · Computer Science 2025-06-04 Chenghua Huang , Zhizhen Fan , Lu Wang , Fangkai Yang , Pu Zhao , Zeqi Lin , Qingwei Lin , Dongmei Zhang , Saravan Rajmohan , Qi Zhang

Large language models~(LLMs) are expected to be helpful, harmless, and honest. In different alignment scenarios, such as safety, confidence, and general preference alignment, binary preference data collection and reward modeling are…

Computation and Language · Computer Science 2025-10-14 Shuai Zhao , Yunqiu Xu , Linchao Zhu , Yi Yang

We provide a theoretical framework for Reinforcement Learning with Human Feedback (RLHF). Our analysis shows that when the true reward function is linear, the widely used maximum likelihood estimator (MLE) converges under both the…

Machine Learning · Computer Science 2024-02-09 Banghua Zhu , Jiantao Jiao , Michael I. Jordan

Process Reward Models (PRMs) play a central role in evaluating and guiding multi-step reasoning in large language models (LLMs), especially for mathematical problem solving. However, we identify a pervasive length bias in existing PRMs:…

Computation and Language · Computer Science 2026-05-20 Congmin Zheng , Jiachen Zhu , Jianghao Lin , Xinyi Dai , Weiwen Liu , Haoxuan Li , Yong Yu , Weinan Zhang , Mengyue Yang

Reward models (RMs) are a crucial component in the alignment of large language models' (LLMs) outputs with human values. RMs approximate human preferences over possible LLM responses to the same prompt by predicting and comparing reward…

Machine Learning · Computer Science 2025-02-27 Junqi Jiang , Tom Bewley , Saumitra Mishra , Freddy Lecue , Manuela Veloso

One of the challenges of aligning large models with human preferences lies in both the data requirements and the technical complexities of current approaches. Predominant methods, such as RLHF, involve multiple steps, each demanding…

Machine Learning · Computer Science 2025-03-19 Siliang Zeng , Yao Liu , Huzefa Rangwala , George Karypis , Mingyi Hong , Rasool Fakoor

Single-axis mitigations of reward-model biases (e.g., reducing proxy reliance on length, sycophancy, or style) can rotate optimization pressure onto correlated proxies rather than eliminate it, a failure mode we call reward bias…

Artificial Intelligence · Computer Science 2026-05-28 Max Lamparth , Daniel Fein , Andreas Haupt , Marcel Hussing , Mykel J. Kochenderfer

Reward models are critical in techniques like Reinforcement Learning from Human Feedback (RLHF) and Inference Scaling Laws, where they guide language model alignment and select optimal responses. Despite their importance, existing reward…

Computation and Language · Computer Science 2024-10-22 Yantao Liu , Zijun Yao , Rui Min , Yixin Cao , Lei Hou , Juanzi Li

Reinforcement Learning with Verifiable Rewards (RLVR) has significantly advanced the reasoning capabilities of Large Language Models (LLMs) by optimizing them against factual outcomes. However, this paradigm falters in long-context…

Computation and Language · Computer Science 2026-03-03 Guanzheng Chen , Michael Qizhe Shieh , Lidong Bing