English
Related papers

Related papers: Understanding Likelihood Over-optimisation in Dire…

200 papers

While large-scale unsupervised language models (LMs) learn broad world knowledge and some reasoning skills, achieving precise control of their behavior is difficult due to the completely unsupervised nature of their training. Existing…

Machine Learning · Computer Science 2024-07-31 Rafael Rafailov , Archit Sharma , Eric Mitchell , Stefano Ermon , Christopher D. Manning , Chelsea Finn

Most Video Large Language Models (Video-LLMs) adopt preference alignment techniques, e.g., DPO~\citep{rafailov2024dpo}, to optimize the reward margin between a winning response ($y_w$) and a losing response ($y_l$). However, the likelihood…

Computer Vision and Pattern Recognition · Computer Science 2025-06-06 Xiaodong Wang , Jinfa Huang , Li Yuan , Peixi Peng

The widespread application of large language models (LLMs) raises increasing demands on ensuring safety or imposing constraints, such as reducing harmful content and adhering to predefined rules. While there have been several works studying…

Machine Learning · Computer Science 2026-02-13 Yihan Du , Seo Taek Kong , R. Srikant

Preference learning is a key technology for aligning language models with human values. Reinforcement Learning from Human Feedback (RLHF) is a model-based algorithm to optimize preference learning, which first fits a reward model for…

Machine Learning · Computer Science 2024-03-26 Zaifan Jiang , Xing Huang , Chao Wei

The prevalent deployment of learning from human preferences through reinforcement learning (RLHF) relies on two important approximations: the first assumes that pairwise preferences can be substituted with pointwise rewards. The second…

Artificial Intelligence · Computer Science 2023-11-23 Mohammad Gheshlaghi Azar , Mark Rowland , Bilal Piot , Daniel Guo , Daniele Calandriello , Michal Valko , Rémi Munos

Aligning large language models (LLMs) with human preferences has gained significant attention, with Proximal Policy Optimization (PPO) as a standard yet computationally expensive method and Direct Preference Optimization (DPO) as a more…

Artificial Intelligence · Computer Science 2025-02-10 Yuzi Yan , Yibo Miao , Jialian Li , Yipin Zhang , Jian Xie , Zhijie Deng , Dong Yan

Direct Alignment Algorithms (DAAs), such as Direct Preference Optimization (DPO) and Simple Preference Optimization (SimPO), have emerged as efficient alternatives to Reinforcement Learning from Human Feedback (RLHF) algorithms for aligning…

Computation and Language · Computer Science 2026-04-17 Zeguan Xiao , Yun Chen , Guanhua Chen , Ke Tang

Direct Preference Optimization (DPO) has become a popular approach for aligning language models using pairwise preferences. However, in practical post-training pipelines, on-policy generation typically yields multiple candidate responses…

Machine Learning · Computer Science 2025-06-23 Taneesh Gupta , Rahul Madhavan , Xuchao Zhang , Nagarajan Natarajan , Chetan Bansal , Saravan Rajmohan

How can Large Language Models (LLMs) be aligned with human intentions and values? A typical solution is to gather human preference on model outputs and finetune the LLMs accordingly while ensuring that updates do not deviate too far from a…

Computation and Language · Computer Science 2024-05-28 Hung Le , Quan Tran , Dung Nguyen , Kien Do , Saloni Mittal , Kelechi Ogueji , Svetha Venkatesh

Direct Preference Optimization (DPO) has emerged as a compelling approach for training Large Language Models (LLMs) to adhere to human preferences. However, the performance of DPO is sensitive to the fine-tuning of its trade-off parameter…

Artificial Intelligence · Computer Science 2024-10-15 Junkang Wu , Yuexiang Xie , Zhengyi Yang , Jiancan Wu , Jinyang Gao , Bolin Ding , Xiang Wang , Xiangnan He

Large language models (LLMs) have shown great potential in natural language processing tasks, but their application to machine translation (MT) remains challenging due to pretraining on English-centric data and the complexity of…

Computation and Language · Computer Science 2025-01-24 Guofeng Cui , Pichao Wang , Yang Liu , Zemian Ke , Zhu Liu , Vimal Bhat

Aligning large language models with human preferences is essential for improving interaction quality and safety by ensuring outputs better reflect human values. A promising strategy involves Reinforcement Learning from Human Feedback…

Information Retrieval · Computer Science 2025-12-17 Jiacong Zhou , Xianyun Wang , Min Zhang , Jun Yu

Direct Preference Optimization (DPO) has emerged as a de-facto approach for aligning language models with human preferences. Recent work has shown DPO's effectiveness relies on training data quality. In particular, clear quality differences…

Machine Learning · Computer Science 2025-01-28 Nirav Diwan , Tolga Ergen , Dongsub Shim , Honglak Lee

Direct Preference Optimization (DPO) has emerged as a lightweight and effective alternative to Reinforcement Learning from Human Feedback (RLHF) and Reinforcement Learning with AI Feedback (RLAIF) for aligning large language and…

Artificial Intelligence · Computer Science 2025-12-16 Zihui Zhao , Zechang Li

Efficient preference optimization algorithms such as Direct Preference Optimization (DPO) have become a popular approach in aligning large language models (LLMs) with human preferences. These algorithms implicitly treat the LLM as a reward…

Computation and Language · Computer Science 2025-07-29 Tong Liu , Xiao Yu , Wenxuan Zhou , Jindong Gu , Volker Tresp

Direct alignment methods typically train large language models (LLMs) by contrasting the likelihoods of preferred and dispreferred responses. While effective at capturing relative preferences, these methods are widely observed to suppress…

Computation and Language · Computer Science 2025-12-04 Kaiyang Guo , Yinchuan Li , Zhitang Chen

Aligning large language models (LLMs) with human values and intentions is crucial for their utility, honesty, and safety. Reinforcement learning from human feedback (RLHF) is a popular approach to achieve this alignment, but it faces…

Machine Learning · Computer Science 2025-07-22 Junkang Wu , Xue Wang , Zhengyi Yang , Jiancan Wu , Jinyang Gao , Bolin Ding , Xiang Wang , Xiangnan He

In this paper, we take a step towards a deeper understanding of learning from human preferences by systematically comparing the paradigm of reinforcement learning from human feedback (RLHF) with the recently proposed paradigm of direct…

Machine Learning · Computer Science 2024-06-06 Andi Nika , Debmalya Mandal , Parameswaran Kamalaruban , Georgios Tzannetos , Goran Radanović , Adish Singla

Traditional RLHF-based LLM alignment methods explicitly maximize the expected rewards from a separate reward model. More recent supervised alignment methods like Direct Preference Optimization (DPO) circumvent this phase to avoid problems…

Machine Learning · Computer Science 2025-02-03 Abhijnan Nath , Changsoo Jung , Ethan Seefried , Nikhil Krishnaswamy

We present a fine-grained theoretical analysis of the performance gap between two-stage reinforcement learning from human feedback~(RLHF) and direct preference optimization~(DPO). Our study decomposes this gap into two sources: the explicit…

Machine Learning · Computer Science 2026-05-13 Ruizhe Shi , Minhak Song , Runlong Zhou , Zihan Zhang , Maryam Fazel , Simon S. Du