English
Related papers

Related papers: Steered LLM Activations are Non-Surjective

200 papers

Code LLMs often default to particular programming languages and libraries under neutral prompts. We investigate whether these preferences are encoded as approximately linear directions in activation space that can be manipulated at…

Machine Learning · Computer Science 2026-03-30 Md Mahbubur Rahman , Arjun Guha , Harshitha Menon

Despite significant progress in alignment, large language models (LLMs) remain vulnerable to adversarial attacks that elicit harmful behaviors. Activation steering techniques offer a promising inference-time intervention approach, but…

Machine Learning · Computer Science 2026-01-28 Quy-Anh Dang , Chris Ngo

Controlling undesirable Large Language Model (LLM) behaviors, such as the generation of unsafe content or failing to adhere to safety guidelines, often relies on costly fine-tuning. Activation steering provides an alternative for…

Computation and Language · Computer Science 2026-03-17 Amr Hegazy , Mostafa Elhoushi , Amr Alanwar

Activation-based control steers large language models (LLMs) by intervening on their internal representations during inference, and has emerged as an effective paradigm for controlling behaviors such as persona and style. However, existing…

Computation and Language · Computer Science 2026-05-29 Yingdong Shi , Ruiming Zhang , Changming Li , Zhiyu Yang , Kaixing Zhang , Jingyi Yu , Kan Ren

Steering vectors are a lightweight method to control language model behavior by adding a learned bias to the activations at inference time. Although steering demonstrates promising performance, recent work shows that it can be unreliable or…

Machine Learning · Computer Science 2025-05-29 Joschka Braun , Carsten Eickhoff , David Krueger , Seyed Ali Bahrainian , Dmitrii Krasheninnikov

Masked diffusion language models (MDLMs) generate text via iterative masked-token denoising, enabling mask-parallel decoding and distinct controllability and efficiency tradeoffs from autoregressive LLMs. Yet, efficient representation-level…

Computation and Language · Computer Science 2026-03-31 Adi Shnaidman , Erin Feiglin , Osher Yaari , Efrat Mentel , Amit Levi , Raz Lapid

As large language models (LLMs) improve in their capacity to serve as personal AI assistants, their ability to output uniquely tailored, personalized responses that align with the soft preferences of their users is essential for enhancing…

Human-Computer Interaction · Computer Science 2025-05-15 Jessica Y. Bo , Tianyu Xu , Ishan Chatterjee , Katrina Passarella-Ward , Achin Kulshrestha , D Shin

Intrinsic self-correction refers to the phenomenon where a language model refines its own outputs purely through prompting, without external feedback or parameter updates. While this approach improves performance across diverse tasks, its…

Computation and Language · Computer Science 2026-02-12 Yu-Ting Lee , Fu-Chieh Chang , Yu-En Shu , Hui-Ying Shih , Pei-Yuan Wu

Large Language Models (LLMs), despite advances in instruction tuning, often fail to follow complex user instructions. Activation steering techniques aim to mitigate this by manipulating model internals, but have a potential risk of…

Machine Learning · Computer Science 2026-03-10 Minjae Kang , Jaehyung Kim

The field of mechanistic interpretability in pre-trained transformer models has demonstrated substantial evidence supporting the ''linear representation hypothesis'', which is the idea that high level concepts are encoded as vectors in the…

Machine Learning · Computer Science 2025-10-08 Damjan Kalajdzievski

Jailbreak prompts can trigger harmful completions on aligned LLMs, In accordance, safety steering has been proposed: test-time activation interventions that steer jailbreak activations to trigger refusal while preserving benign utility.…

Cryptography and Security · Computer Science 2026-05-26 Luoyu Chen , Weiqi Wang , Zhiyi Tian , Chenhan Zhang , Feng Wu , Jianhuan Huang , Ahmed Asiri , Shui Yu

Deploying LLMs in real-world applications requires controllable output that satisfies multiple desiderata at the same time. While existing work extensively addresses LLM steering for a single behavior, \textit{compositional steering} --…

Computation and Language · Computer Science 2026-04-21 Gorjan Radevski , Kiril Gashteovski , Giwon Hong , Carolin Lawrence , Goran Glavaš

Precise control over language model generation is vital for ensuring both safety and reliability. Although prompt engineering and steering are commonly used to intervene in model behaviors, the vast number of parameters in models often…

Computation and Language · Computer Science 2025-06-04 Mengru Wang , Ziwen Xu , Shengyu Mao , Shumin Deng , Zhaopeng Tu , Huajun Chen , Ningyu Zhang

Large language models (LLMs) often encode cognitive behaviors unpredictably across prompts, layers, and contexts, making them difficult to diagnose and control. We present CBMAS, a diagnostic framework for continuous activation steering,…

Artificial Intelligence · Computer Science 2026-01-13 Ahmed H. Ismail , Anthony Kuang , Ayo Akinkugbe , Kevin Zhu , Sean O'Brien

Large Language Models (LLMs) often generate inconsistent responses when prompted with semantically equivalent paraphrased inputs. Recently, activation steering, a technique that modulates LLMs' behaviours by adjusting their latent…

Computation and Language · Computer Science 2025-01-23 Jingyuan Yang , Rongjun Li , Weixuan Wang , Ziyu Zhou , Zhiyong Feng , Wei Peng

Controlling multiple behavioral attributes in large language models (LLMs) at inference time is a challenging problem due to interference between attributes and the limitations of linear steering methods, which assume additive behavior in…

Machine Learning · Computer Science 2026-04-07 Narmeen Oozeer , Luke Marks , Shreyans Jain , Fazl Barez , Amirali Abdullah

Steering has emerged as a practical approach to enable post-hoc guidance of LLMs towards enforcing a specific behavior. However, it remains largely underexplored for multimodal LLMs (MLLMs); furthermore, existing steering techniques, such…

Machine Learning · Computer Science 2025-11-04 Jayneel Parekh , Pegah Khayatan , Mustafa Shukor , Arnaud Dapogny , Alasdair Newson , Matthieu Cord

The memorization of training data by Large Language Models (LLMs) poses significant risks, including privacy leaks and the regurgitation of copyrighted content. Activation steering, a technique that directly intervenes in model activations,…

Computation and Language · Computer Science 2025-03-11 Manan Suri , Nishit Anand , Amisha Bhaskar

Recent work has shown that LLMs can sometimes detect when steering vectors are injected into their residual stream and identify the injected concept -- a phenomenon termed "introspective awareness." We investigate the mechanisms underlying…

Machine Learning · Computer Science 2026-05-18 Uzay Macar , Li Yang , Atticus Wang , Peter Wallich , Emmanuel Ameisen , Jack Lindsey

Large language models (LLMs) are prone to capturing biases from training corpus, leading to potential negative social impacts. Existing prompt-based debiasing methods exhibit instability due to their sensitivity to prompt changes, while…

Computation and Language · Computer Science 2025-07-08 Yichen Li , Zhiting Fan , Ruizhe Chen , Xiaotang Gai , Luqi Gong , Yan Zhang , Zuozhu Liu