English
Related papers

Related papers: Unintentional Unalignment: Likelihood Displacement…

200 papers

Aligning generative models with human preference via RLHF typically suffers from overoptimization, where an imperfectly learned reward model can misguide the generative model to output undesired responses. We investigate this problem in a…

Machine Learning · Computer Science 2024-12-05 Zhihan Liu , Miao Lu , Shenao Zhang , Boyi Liu , Hongyi Guo , Yingxiang Yang , Jose Blanchet , Zhaoran Wang

Direct Preference Optimization (DPO) has become a popular method for fine-tuning large language models (LLMs) due to its stability and simplicity. However, it is also known to be sensitive to noise in the data and prone to overfitting.…

Machine Learning · Computer Science 2025-10-28 Cheol Woo Kim , Shresth Verma , Mauricio Tec , Milind Tambe

The field of preference optimization has made outstanding contributions to the alignment of language models with human preferences. Despite these advancements, recent methods still rely heavily on substantial paired (labeled) feedback data,…

Machine Learning · Computer Science 2026-02-20 Seonggyun Lee , Sungjun Lim , Seojin Park , Soeun Cheon , Kyungwoo Song

Iterative preference optimization has recently become one of the de-facto training paradigms for large language models (LLMs), but the performance is still underwhelming due to too much noisy preference data yielded in the loop. To combat…

Computation and Language · Computer Science 2024-09-18 Jianing Wang , Yang Zhou , Xiaocheng Zhang , Mengjiao Bao , Peng Yan

Direct Preference Optimization (DPO) simplifies reinforcement learning from human feedback (RLHF) for large language models (LLMs) by directly optimizing human preferences without an explicit reward model. We find that during DPO training,…

Computation and Language · Computer Science 2026-01-01 Junshu Pan , Wei Shen , Shulin Huang , Qiji Zhou , Yue Zhang

Direct Preference Optimization (DPO) aligns text-to-image (T2I) generation models with human preferences using pairwise preference data. Although substantial resources are expended in collecting and labeling datasets, a critical aspect is…

Computer Vision and Pattern Recognition · Computer Science 2025-06-09 Yunhong Lu , Qichao Wang , Hengyuan Cao , Xiaoyin Xu , Min Zhang

Improving the alignment of language models with human preferences remains an active research challenge. Previous approaches have primarily utilized Reinforcement Learning from Human Feedback (RLHF) via online RL methods such as Proximal…

Computation and Language · Computer Science 2024-01-25 Tianqi Liu , Yao Zhao , Rishabh Joshi , Misha Khalman , Mohammad Saleh , Peter J. Liu , Jialu Liu

Direct Preference Optimization (DPO) has become the de facto standard for offline preference alignment of large language models, but its reliance on a reference policy introduces a critical tension. DPO weighs each update relative to a…

Machine Learning · Computer Science 2026-02-13 Suqin Yuan , Xingrui Yu , Jiyang Zheng , Lei Feng , Dadong Wang , Ivor Tsang , Tongliang Liu

Post-alignment of large language models (LLMs) is critical in improving their utility, safety, and alignment with human intentions. Direct preference optimisation (DPO) has become one of the most widely used algorithms for achieving this…

Machine Learning · Computer Science 2025-01-06 Rasul Tutnov , Antoine Grosnit , Haitham Bou-Ammar

Generative AI can now synthesize strikingly realistic images from text, yet output quality remains highly sensitive to how prompts are phrased. Direct Preference Optimization (DPO) offers a lightweight, off-policy alternative to RL for…

Computation and Language · Computer Science 2025-07-30 Anas Mohamed , Azal Ahmad Khan , Xinran Wang , Ahmad Faraz Khan , Shuwen Ge , Saman Bahzad Khan , Ayaan Ahmad , Ali Anwar

In this paper, we introduce \emph{refined Direct Preference Optimization} (rDPO), a method for improving the behavioral alignment of Large Language Models (LLMs) without the need for human-annotated data. The method involves creating…

Computation and Language · Computer Science 2024-02-14 Víctor Gallego

Direct Preference Optimization (DPO) is an effective technique that leverages pairwise preference data (usually one chosen and rejected response pair per user prompt) to align LLMs to human preferences. In practice, multiple responses can…

Computation and Language · Computer Science 2024-11-11 Pulkit Pattnaik , Rishabh Maheshwary , Kelechi Ogueji , Vikas Yadav , Sathwik Tejaswi Madhusudhan

Machine unlearning aims to efficiently eliminate the influence of specific training data, known as the forget set, from the model. However, existing unlearning methods for Large Language Models (LLMs) face a critical challenge: they rely…

Computation and Language · Computer Science 2025-01-23 Anmol Mekala , Vineeth Dorna , Shreya Dubey , Abhishek Lalwani , David Koleczek , Mukund Rungta , Sadid Hasan , Elita Lobo

Direct Preference Optimization (DPO) aligns language models using pairwise preference comparisons, offering a simple and effective alternative to Reinforcement Learning (RL) from human feedback. However, in many practical settings, training…

Machine Learning · Computer Science 2026-05-11 Ning Liu , Chuanneng Sun , Kristina Klinkner , Shervin Malmasi

Direct preference optimization (DPO) is a form of reinforcement learning from human feedback (RLHF) where the policy is learned directly from preferential feedback. Although many models of human preferences exist, the critical task of…

Machine Learning · Computer Science 2025-03-04 Branislav Kveton , Xintong Li , Julian McAuley , Ryan Rossi , Jingbo Shang , Junda Wu , Tong Yu

Fairness in Continual Learning for Large Multimodal Models (LMMs) is an emerging yet underexplored challenge, particularly in the presence of imbalanced data distributions that can lead to biased model updates and suboptimal performance…

Machine Learning · Computer Science 2026-03-31 Thanh-Dat Truong , Huu-Thien Tran , Jackson Cothren , Bhiksha Raj , Khoa Luu

Modern alignment pipelines are increasingly replacing expensive human preference labels with evaluations from large language models (LLM-as-Judge). However, AI labels can be systematically biased compared to high-quality human feedback…

Machine Learning · Statistics 2026-02-10 Xintao Xia , Zhiqiu Xia , Linjun Zhang , Zhanrui Cai

Preference optimization techniques, such as Direct Preference Optimization (DPO), are frequently employed to enhance the reasoning capabilities of large language models (LLMs) in domains like mathematical reasoning and coding, typically…

Computation and Language · Computer Science 2025-02-17 Fangkai Jiao , Geyang Guo , Xingxing Zhang , Nancy F. Chen , Shafiq Joty , Furu Wei

Recently, preference optimization methods such as DPO have significantly enhanced large language models (LLMs) in wide tasks including dialogue and question-answering. However, current methods fail to account for the varying difficulty…

Computation and Language · Computer Science 2024-12-31 Jingyuan Ma , Rui Li , Zheng Li , Lei Sha , Zhifang Sui

When fine-tuning pre-trained Large Language Models (LLMs) to align with human values and intentions, maximizing the estimated reward can lead to superior performance, but it also introduces potential risks due to deviations from the…

Machine Learning · Computer Science 2025-05-30 Lijun Zhang , Lin Li , Yajie Qi , Huizhong Song , Yaodong Yang , Jun Wang , Wei Wei