English
Related papers

Related papers: FlipGuard: Defending Preference Alignment against …

200 papers

Offline policy improvement faces an inherent conflict between maximizing value and fitting the data distribution. While in-sample weighted regression is stable, it suffers from over-conservatism that suppresses high-value actions in the…

Machine Learning · Computer Science 2026-05-28 Jiaxin Zhao , Weihang Pan , Xun Liang , Binbin Lin

Conventional wisdom in pruning Transformer-based language models is that pruning reduces the model expressiveness and thus is more likely to underfit rather than overfit. However, under the trending pretrain-and-finetune paradigm, we…

Computation and Language · Computer Science 2023-01-18 Shaoyi Huang , Dongkuan Xu , Ian E. H. Yen , Yijue Wang , Sung-en Chang , Bingbing Li , Shiyang Chen , Mimi Xie , Sanguthevar Rajasekaran , Hang Liu , Caiwen Ding

Behavior of deep neural networks can be inconsistent between different versions. Regressions during model update are a common cause of concern that often over-weigh the benefits in accuracy or efficiency gain. This work focuses on…

Computation and Language · Computer Science 2021-05-10 Yuqing Xie , Yi-an Lai , Yuanjun Xiong , Yi Zhang , Stefano Soatto

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

Large language models are typically aligned with human preferences by optimizing $\textit{reward models}$ (RMs) fitted to human feedback. However, human preferences are multi-faceted, and it is increasingly common to derive reward from a…

The success of AI assistants based on Language Models (LLMs) hinges on Reinforcement Learning from Human Feedback (RLHF) to comprehend and align with user intentions. However, traditional alignment algorithms, such as PPO, are hampered by…

Computation and Language · Computer Science 2024-07-03 Songyang Gao , Qiming Ge , Wei Shen , Shihan Dou , Junjie Ye , Xiao Wang , Rui Zheng , Yicheng Zou , Zhi Chen , Hang Yan , Qi Zhang , Dahua Lin

Given a finite set of sample points, meta-learning algorithms aim to learn an optimal adaptation strategy for new, unseen tasks. Often, this data can be ambiguous as it might belong to different tasks concurrently. This is particularly the…

Machine Learning · Computer Science 2024-10-24 Alfredo Reichlin , Gustaf Tegnér , Miguel Vasco , Hang Yin , Mårten Björkman , Danica Kragic

Fine-tuning safety-aligned language models for downstream tasks often leads to substantial degradation of refusal behavior, making models vulnerable to adversarial misuse. While prior work has shown that safety-relevant features are encoded…

Machine Learning · Computer Science 2026-05-05 Sadia Asif , Mohammad Mohammadi Amiri

When machine learning systems meet real world applications, accuracy is only one of several requirements. In this paper, we assay a complementary perspective originating from the increasing availability of pre-trained and regularly…

Recent generative models have significantly advanced speech restoration tasks, yet their training objectives often misalign with human perceptual preferences, resulting in suboptimal quality. While post-training alignment has proven…

Sound · Computer Science 2025-11-18 Junan Zhang , Xueyao Zhang , Jing Yang , Yuancheng Wang , Fan Fan , Zhizheng Wu

Alignment of large language models remains a central challenge in natural language processing. Preference optimization has emerged as a popular and effective method for improving alignment, typically through training-time or prompt-based…

Machine Learning · Computer Science 2025-10-01 Frédéric Berdoz , Luca A. Lanzendörfer , René Caky , Roger Wattenhofer

Prevailing alignment methods induce opaque parameter changes, obscuring what models truly learn. To address this, we introduce Feature Steering with Reinforcement Learning (FSRL), a framework that trains a lightweight adapter to steer model…

Artificial Intelligence · Computer Science 2025-12-02 Jeremias Ferrao , Matthijs van der Lende , Ilija Lichkovski , Clement Neo

Self-improvement through post-training methods such as iterative preference learning has been acclaimed for enhancing the problem-solving capabilities (e.g., mathematical reasoning) of Large Language Models (LLMs) without human…

Computation and Language · Computer Science 2024-07-09 Ting Wu , Xuefeng Li , Pengfei Liu

Recent advances in LLMs have enhanced AI capabilities, but also increased the risk posed by malicious requests, highlighting the need for effective LLM safeguards to detect such queries. Existing approaches largely rely on classifier-based…

Computation and Language · Computer Science 2025-10-14 Zhuowei Chen , Bowei Zhang , Nankai Lin , Tian Hou , Lianxi Wang

Current reinforcement learning objectives for large-model reasoning primarily focus on maximizing expected rewards. This paradigm can lead to overfitting to dominant reward signals, while neglecting alternative yet valid reasoning…

Machine Learning · Computer Science 2026-02-24 Wendi Li , Sharon Li

The emergence of large Vision Language Models (VLMs) has broadened the scope and capabilities of single-modal Large Language Models (LLMs) by integrating visual modalities, thereby unlocking transformative cross-modal applications in a…

Computer Vision and Pattern Recognition · Computer Science 2025-09-23 Shuo Xing , Peiran Li , Yuping Wang , Ruizheng Bai , Yueqi Wang , Chan-Wei Hu , Chengxuan Qian , Huaxiu Yao , Zhengzhong Tu

Detecting AI-generated text is an important but challenging problem. Existing likelihood-based detection methods are often sensitive to content complexity and may exhibit unstable performance. In this paper, our key insight is that modern…

Artificial Intelligence · Computer Science 2026-04-21 Junxi Wu , Kailin Huang , Dongjian Hu , Bin Chen , Hao Wu , Shu-Tao Xia , Changliang Zou

Recent work shows that preference alignment objectives can be interpreted as divergence estimators between aligned (preferred) & unaligned (less-preferred) distributions, yielding a principled recipe for designing alignment losses. However,…

Machine Learning · Computer Science 2026-05-12 Rajdeep Haldar , Lantao Mei , Guang Lin , Yue Xing , Qifan Song

This work studies the challenge of aligning large language models (LLMs) with offline preference data. We focus on alignment by Reinforcement Learning from Human Feedback (RLHF) in particular. While popular preference optimization methods…

Machine Learning · Computer Science 2024-06-07 Xiang Ji , Sanjeev Kulkarni , Mengdi Wang , Tengyang Xie

Quality of datasets plays an important role in large language model (LLM) alignment. In collecting human feedback, however, preference flipping is ubiquitous and causes corruption in data annotation; the issue necessitates the alignment…

Artificial Intelligence · Computer Science 2025-12-02 Yifan Xu , Xichen Ye , Yifan Chen , Qiaosheng Zhang