Pragmatic inference is inherently graded. Different lexical items give rise to pragmatic enrichment to different degrees. Scalar implicature exemplifies this property through scalar diversity, where implicature strength varies across scalar items. However, evaluations of pragmatic inference in large language models (LLMs) often rely on prompt-based manipulations. Beyond prompt-level effects, this study introduces Continuous Interpretive Steering (CIS), a method that probes graded pragmatic interpretation by treating activation-level steering strength as a continuous experimental variable. To support this analysis, this study introduces a new dataset, GraSD, which encodes graded scalar diversity. Experiments on four LLMs show that uniform activation steering increases pragmatic interpretations globally but collapses item-level variation, whereas graded activation steering yields differentiated interpretive shifts aligned with scalar diversity grades. It indicates that graded sensitivity is encoded in the representation space and can be systematically recovered through controlled intervention. Together, CIS and GraSD provide a principled framework for evaluating graded pragmatic sensitivity in LLMs.
Steering Language Generation: Harnessing Contrastive Expert Guidance and Negative Prompting for Coherent and Diverse Synthetic Data Generation
Mathematical Sciences Institute, Australian National University(澳大利亚国立大学数学科学研究所);Research School of Astronomy & Astrophysics, Australian National University(澳大利亚国立大学天文与天体物理研究学院);School of Computing, Australian National University(澳大利亚国立大学计算学院);Department of Astronomy, The Ohio State University(俄亥俄州立大学天文学系)
Style-Specific Neurons for Steering LLMs in Text Style Transfer
School of Computation, Information and Technology, Technical University of Munich(慕尼黑工业大学计算、信息与技术学院);Munich Center for Machine Learning(慕尼黑机器学习中心);Center for Information and Language Processing, LMU Munich(慕尼黑大学信息与语言处理中心)
Less is Enough: Synthesizing Diverse Data in LLM Feature Space with Sparse Autoencoders
Department of Computing, University of Georgia, Georgia, United States(佐治亚大学计算机系);Computer Engineering, University of California San Diego, California, United States(加州大学圣地亚哥分校计算机工程系);Machine Learning Department, Mohamed bin Zayed University of Artificial Intelligence, Abu Dhabi, United Arab Emirates(Mohamed bin Zayed人工智能大学机器学习系);Department of Computing, Hong Kong Polytechnic University, Hong Kong, China(香港理工大学计算机系)