Skip to content
AI News HubLIVE
Original source2 min read

Creating Steering Vectors via Input Optimisation

Summary

This paper introduces ActGCG, a method that optimises an input suffix to match target activations produced by activation steering, enabling black-box reproduction of steering vector behaviour. Experiments show that for steering a model to answer in German, a discrete suffix almost perfectly replicates the steering vector's effect and transfers across prompts and models, while for anti-refusal steering the equivalence is weaker, succeeding under a phrase-avoidance metric but only partially under the stricter HarmBench classifier.

SourcearXiv Computational LinguisticsAuthor: Batuhan Akar
Creating Steering Vectors via Input Optimisation
Report an error

The correction channel is not available yet. You can copy the article reference below for later.

Correction instructions
Read article

[Submitted on 12 Nov 2026]

Title:Creating Steering Vectors via Input Optimisation

View a PDF of the paper titled Creating Steering Vectors via Input Optimisation, by Batuhan Akar

View PDF HTML (experimental)

Abstract:Activation steering influences a language model's behaviour by adding a steering vector to its residual stream during generation, but it requires white-box access to modify the hidden states at every generation step. We introduce ActGCG, a modification of Greedy Coordinate Gradient which optimises a suffix to match the target activations obtained by adding a steering vector to the residual stream. We further introduce universal suffixes trained across many prompts at once, a model-translation step that rewrites the suffix into human-readable text midway through optimisation, and a soft-suffix variant to remove the discrete token space constraint. We study two behaviours: steering a model to answer in German, and anti-refusal steering for jailbreaking. For German, a discrete suffix reproduces the steering vector's behaviour almost perfectly and transfers well across prompts and models. For anti-refusal, the equivalence is weaker, succeeding under a phrase-avoidance metric but only partially under the stricter HarmBench classifier. Our results show that a steering vector's behaviour is reachable from the input, with no access to model internals at generation time, up to a loss that depends on the concept being steered.

Comments: Bachelor's Thesis, ETH Zurich

Subjects:

Computation and Language (cs.CL); Machine Learning (cs.LG)

Cite as: arXiv:2611.17481 [cs.CL]

(or arXiv:2611.17481v1 [cs.CL] for this version)

https://doi.org/10.48550/arXiv.2611.17481

arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Batuhan Akar [view email] [v1] Thu, 12 Nov 2026 20:06:14 UTC (738 KB)

Full-text links:

Access Paper:

View a PDF of the paper titled Creating Steering Vectors via Input Optimisation, by Batuhan Akar

View PDF

HTML (experimental)

TeX Source

view license

Additional Features

Audio Summary

Current browse context:

cs.CL

new | recent | 2026-11

Change to browse by:

cs cs.LG

References & Citations

SciX

Google Scholar

Semantic Scholar

Loading...

Data provided by:

Bibliographic Tools

Bibliographic and Citation Tools

Bibliographic Explorer Toggle

Bibliographic Explorer (What is the Explorer?)

Connected Papers Toggle

Connected Papers (What is Connected Papers?)

Litmaps Toggle

Litmaps (What is Litmaps?)

scite.ai Toggle

scite Smart Citations (What are Smart Citations?)

Code, Data, Media

Code, Data and Media Associated with this Article

alphaXiv Toggle

alphaXiv (What is alphaXiv?)

Links to Code Toggle

CatalyzeX Code Finder for Papers (What is CatalyzeX?)

DagsHub Toggle

DagsHub (What is DagsHub?)

GotitPub Toggle

Gotit.pub (What is GotitPub?)

Huggingface Toggle

Hugging Face (What is Huggingface?)

ScienceCast Toggle

ScienceCast (What is ScienceCast?)

Demos

Demos

Replicate Toggle

Replicate (What is Replicate?)

Spaces Toggle

Hugging Face Spaces (What is Spaces?)

Spaces Toggle

TXYZ.AI (What is TXYZ.AI?)

Related Papers

Recommenders and Search Tools

Link to Influence Flower

Influence Flower (What are Influence Flowers?)

Core recommender toggle

CORE Recommender (What is CORE?)

Author

Venue

Institution

Topic

About arXivLabs

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)

Key points and analysis

Article intelligence

EngineersAdvanced

Key points

  • ActGCG optimises input suffixes to match activations from steering vectors.
  • Universal suffixes, model-translation, and soft-suffix variants enhance practicality.
  • German steering is almost perfectly reproduced and transfers across models.
  • Anti-refusal steering shows weaker equivalence, especially under strict classifiers.

Highlights and analysis are generated automatically and may contain errors. Check the original source.