Steering vectors have emerged as a promising approach to controlled text generation, offering interpretable, training-free mechanisms for shaping model outputs. However, their practical generality remains poorly understood. We study the limits of steering vector generalization along three dimensions: trait expressibility, task transfer, and multi-trait composition. Using the PLUME writing personalization benchmark, we extract steering vectors for a range of preferences and evaluate them on summarization and email-writing tasks across two open-source models (Qwen2.5-7B-Instruct and Llama3.1-8B-Instruct). We find that steering effectiveness varies substantially across traits. We further show that steering effectiveness can degrade when vectors extracted from positive and negative style examples are transferred to downstream writing personalization tasks. Finally, we compare common methods for composing multiple steering vectors and find that all methods suffer significant drops in trait expression as more vectors are added, with a tradeoff between coherence and expressibility that requires per-setting hyperparameter tuning. Taken together, our results suggest that steering vectors face meaningful limits as a general-purpose tool for preference alignment.
Personalized Text Generation with Contrastive Activation Steering
Institute of Automation, Chinese Academy of Sciences（中国科学院自动化研究所）；School of Artificial Intelligence, University of Chinese Academy of Sciences（中国科学院大学人工智能学院）；Northeastern University（东北大学）；University of Science and Technology of China（中国科学技术大学）；National University of Singapore（新加坡国立大学）
Capturing Nuanced Preferences: Preference-Aligned Distillation for Small Language Models
The Hong Kong University of Science and Technology (Guangzhou)（香港科技大学（广州））；Soochow University（苏州大学）；University of Technology Sydney（悉尼科技大学）；The Hong Kong University of Science and Technology（香港科技大学）
针对成对偏好蒸馏忽视回复差异程度的问题，本文提出 PAD 框架，将师生内在偏好建模为分布并进行对齐，在多个基准上显著提升 SLM 的人类偏好对齐效果。
StoryLens: Preference-Aligned Story Rewriting via Context-Aware Narrative Enrichment
Beijing University of Posts and Telecommunications（北京邮电大学）；AIM3 Lab, Renmin University of China（中国人民大学AIM3实验室）；Nanyang Technological University（南洋理工大学）
From Volume to Value: Preference-Aligned Memory Construction for On-Device RAG
Department of Computer Science and Engineering, Ulsan National Institute of Science and Technology (UNIST), Ulsan, Republic of Korea（计算机科学与工程系，全州国立科学与技术研究所（UNIST），全州，韩国）
Same Values, Different Languages? From Multilingual Probing to Steering LLMs Toward Chinese Social Values
School of Information Science and Technology, Beijing Foreign Studies University（北京外国语大学信息科学与技术学院）；Beijing Academy of Artificial Intelligence（北京人工智能研究院）；Beihang University（北京航空航天大学）
Compiling Activation Steering into Weights via Null-Space Constraints for Stealthy Backdoors
Zhejiang University（浙江大学）；Palo Alto Networks（帕洛阿尔托网络）；Ningbo Global Innovation Center, Zhejiang University（宁波全球创新中心，浙江大学）；OPPO Research Institute（OPPO研究院）；Qilu University of Technology（齐鲁大学）
Distribution-Aligned Decoding for Efficient LLM Task Adaptation
Hong Kong JC STEM Lab of Smart City（香港JC智能城市STEM实验室）；City University of Hong Kong（香港城市大学）；University of Sussex（苏塞克斯大学）；Huazhong University of Science and Technology（华中科技大学）；Fudan University（复旦大学）；Lingnan University（岭大大学）
Efficient Preference-based Reinforcement Learning via Aligned Experience Estimation
Shanghai Jiao Tong University（上海交通大学）；Institute for Artificial Intelligence, Peking University（北京大学人工智能研究院）；BIGAI（北京智源人工智能研究院）；Tencent Robotics X（腾讯Robotics X实验室）；University of Alberta（阿尔伯塔大学）；Institute of Automation, Chinese Academy of Sciences（中国科学院自动化研究所）
