English
Related papers

Related papers: KL Penalty Control via Perturbation for Direct Pre…

200 papers

Post-training processes are essential phases in grounding pre-trained language models to real-world tasks, with learning from demonstrations or preference signals playing a crucial role in this adaptation. We present a unified theoretical…

Machine Learning · Computer Science 2025-07-08 Bo Wang , Qinyuan Cheng , Runyu Peng , Rong Bao , Peiji Li , Qipeng Guo , Linyang Li , Zhiyuan Zeng , Yunhua Zhou , Xipeng Qiu

Aligning large language models (LLMs) with human preferences is commonly done via reinforcement learning from human feedback (RLHF) with Proximal Policy Optimization (PPO) or, more simply, via Direct Preference Optimization (DPO). While DPO…

Artificial Intelligence · Computer Science 2026-05-04 Abdulhady Abas Abdullah , Fatemeh Daneshfar , Seyedali Mirjalili , Mourad Oussalah

The alignment of large language models with human values presents a critical challenge, particularly when balancing conflicting objectives like helpfulness and harmlessness. Existing approaches, such as Reinforcement Learning from Human…

Computation and Language · Computer Science 2025-03-04 Yuxuan Liu

As development of large language models (LLM) progresses, aligning them with human preferences has become increasingly important. We propose stepwise DPO (sDPO), an extension of the recently popularized direct preference optimization (DPO)…

Computation and Language · Computer Science 2024-10-08 Dahyun Kim , Yungi Kim , Wonho Song , Hyeonwoo Kim , Yunsu Kim , Sanghoon Kim , Chanjun Park

Aligning Large Language Models (LLMs) with human preferences is crucial for safe and effective AI interactions. While popular methods like Direct Preference Optimization (DPO) have simplified alignment, they remain sensitive to data noise…

Artificial Intelligence · Computer Science 2026-03-03 Ning Yang , Hai Lin , Yibo Liu , Baoliang Tian , Guoqing Liu , Haijun Zhang

The alignment of language models with human preferences is vital for their application in real-world tasks. The problem is formulated as optimizing the model's policy to maximize the expected reward that reflects human preferences with…

Computation and Language · Computer Science 2024-06-06 Haozhe Ji , Cheng Lu , Yilin Niu , Pei Ke , Hongning Wang , Jun Zhu , Jie Tang , Minlie Huang

Direct Preference Optimization (DPO) has been widely adopted for preference alignment of Large Language Models (LLMs) due to its simplicity and effectiveness. However, DPO is derived as a bandit problem in which the whole response is…

Computation and Language · Computer Science 2025-04-16 Aiwei Liu , Haoping Bai , Zhiyun Lu , Yanchao Sun , Xiang Kong , Simon Wang , Jiulong Shan , Albin Madappally Jose , Xiaojiang Liu , Lijie Wen , Philip S. Yu , Meng Cao

Direct Preference Optimization (DPO) has been demonstrated to be highly effective in mitigating hallucinations in Large Vision Language Models (LVLMs) by aligning their outputs more closely with human preferences. Despite the recent…

Computer Vision and Pattern Recognition · Computer Science 2025-09-24 Jihao Gu , Yingyao Wang , Meng Cao , Pi Bu , Jun Song , Yancheng He , Shilong Li , Bo Zheng

Direct Preference Optimization (DPO) and its variants have become standard for aligning Large Language Models due to their simplicity and offline stability. However, we identify two fundamental limitations. First, the optimal policy depends…

Artificial Intelligence · Computer Science 2026-02-10 Yu Li , Tian Lan , Zhengling Qi

Adapting large language models (LLMs) for specific tasks usually involves fine-tuning through reinforcement learning with human feedback (RLHF) on preference data. While these data often come from diverse labelers' groups (e.g., different…

Computation and Language · Computer Science 2024-05-31 Shyam Sundhar Ramesh , Yifan Hu , Iason Chaimalas , Viraj Mehta , Pier Giuseppe Sessa , Haitham Bou Ammar , Ilija Bogunovic

Direct Preference Optimization is an offline post-SFT method for aligning language models from preference pairs, with strong results in instruction following and summarization. However, DPO's sequence-level implicit reward can be brittle…

Computation and Language · Computer Science 2026-03-03 Samah Fodeh , Linhai Ma , Ganesh Puthiaraju , Srivani Talakokkul , Afshan Khan , Ashley Hagaman , Sarah R. Lowe , Aimee Kendall Roundtree

Direct Preference Optimization (DPO) has proven effective at improving the performance of large language models (LLMs) on downstream tasks such as reasoning and alignment. In this work, we propose Step-Controlled DPO (SCDPO), a method for…

Computation and Language · Computer Science 2024-07-16 Zimu Lu , Aojun Zhou , Ke Wang , Houxing Ren , Weikang Shi , Junting Pan , Mingjie Zhan , Hongsheng Li

Normative theories allow one to elicit key parts of a ML algorithm from first principles, which is crucial at a time of championed scrutiny for ML work. Direct Preference Optimization (DPO) cleverly bypasses reward modeling by making an…

Machine Learning · Computer Science 2026-02-05 Wenxuan Zhou , Shujian Zhang , Brice Magdalou , John Lambert , Ehsan Amid , Richard Nock , Andrew Hard

Reinforcement Learning from Human Feedback (RLHF) has become central to aligning large language models with human values, typically by first learning a reward model from preference data which is then used to update the model with…

Machine Learning · Computer Science 2025-10-21 Keertana Chidambaram , Karthik Vinay Seetharaman , Vasilis Syrgkanis

Aligning large language models with pointwise absolute rewards has so far required online, on-policy algorithms such as PPO and GRPO. In contrast, simpler methods that can leverage offline or off-policy data, such as DPO and REBEL, are…

Machine Learning · Computer Science 2025-12-02 Simon Matrenok , Skander Moalla , Caglar Gulcehre

Direct Preference Optimization (DPO) and related methods align large language models from pairwise preferences by regularizing updates against a fixed reference policy. As the policy drifts, a static reference, however, can become…

Machine Learning · Computer Science 2026-05-18 Youngjae Cho , Jongsuk Kim , Ji-Hoon Kim

Two divergence regimes dominate modern alignment practice. Supervised fine-tuning and many distillation-style objectives implicitly minimize the forward KL divergence KL(q || pi_theta), yielding stable mode-covering updates but often…

Machine Learning · Computer Science 2025-12-30 Wang Zixian

Direct Preference Optimization (DPO) has shown strong potential for mitigating hallucinations in Multimodal Large Language Models (MLLMs). However, existing multimodal DPO approaches often suffer from overfitting due to the difficulty…

Artificial Intelligence · Computer Science 2026-01-05 Longtian Qiu , Shan Ning , Chuyu Zhang , Jiaxuan Sun , Xuming 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

Reinforcement Learning from Human Feedback (RLHF) has become central to aligning large language models with human values, typically by first learning a reward model from preference data which is then used to update the model with…

Artificial Intelligence · Computer Science 2025-10-20 Keertana Chidambaram , Karthik Vinary Seetharaman , Vasilis Syrgkanis