Steering, or direct manipulation of internal activations to guide LLM responses toward specific semantic concepts, is emerging as a promising avenue for both understanding how semantic concepts are stored within LLMs and advancing LLM capabilities. Yet, existing steering methods are remarkably brittle, with seemingly non-steerable concepts becoming completely steerable based on subtle algorithmic choices in how concept-related features are extracted. In this work, we introduce an attention-guided steering framework that overcomes three core challenges associated with steering: (1) automatic selection of relevant token embeddings for extracting concept-related features; (2) accounting for heterogeneity of concept-related features across LLM activations; and (3) identification of layers most relevant for steering. Across a steering benchmark of 512 semantic concepts, our framework substantially improved steering over previous state-of-the-art (nearly doubling the number of successfully steered concepts) across model architectures and sizes (up to 70 billion parameter models). Furthermore, we use our framework to shed light on the distribution of concept-specific features across LLM layers. Overall, our framework opens further avenues for developing efficient, highly-scalable fine-tuning algorithms for industry-scale LLMs.
Interpretable Steering of Large Language Models with Feature Guided Activation Additions
Raffles Science Institute(莱佛士科学学院);Raffles Institution(莱佛士书院);Nous Research(努斯研究机构);Centre for Frontier AI Research (CFAR)(前沿人工智能研究中心);Agency for Science Technology and Research (A*STAR)(新加坡科技研究局)
Euphonium: Steering Video Flow Matching via Process Reward Gradient Guided Stochastic Dynamics
Huazhong University of Science and Technology(华中科技大学);Shanghai Jiao Tong University(上海交通大学);Tencent Hunyuan(腾讯混元);University of Chinese Academy of Sciences(中国科学院大学)
Locate-then-Sparsify: Attribution Guided Sparse Strategy for Visual Hallucination Mitigation
State Key Lab. of AI Safety, Institute of Computing Technology, Chinese Academy of Sciences(中国科学院人工智能安全国家重点实验室,计算技术研究所);School of Advanced Interdisciplinary Sciences, University of Chinese Academy of Sciences(中国科学院大学先进交叉科学学院);School of Computer Science and Technology, University of Chinese Academy of Sciences(中国科学院大学计算机科学与技术学院)
本文提出了一种名为Locate-Then-Sparsify for Feature Steering (LTS-FS)的框架,通过定位和稀疏化策略,根据每层与幻觉的相关性调整特征引导强度,从而有效缓解视觉语言模型中的幻觉问题,同时保持良好的性能。
ILRR: Inference-Time Steering Method for Masked Diffusion Language Models
School of Computer Science, Tel Aviv University, Israel(特拉维夫大学计算机科学学院);Department of Electrical and Computer Engineering, Ben Gurion University, Beer Sheva, Israel(本· Gurion大学电气与计算机工程系)
AttentionGAN: Unpaired Image-to-Image Translation using Attention-Guided Generative Adversarial Networks
ETH Zurich(苏黎世联邦理工学院);Shenzhen Graduate School, Peking University(北京大学深圳研究生院);Hong Kong University of Sciences and Technology (HKUST)(香港科技大学);University of Oxford(牛津大学);University of Trento(特伦托大学)