本文へ移動
AI NEWS JOURNAL
arXiv cs.CLAI NEWS / JOURNAL

Creating Steering Vectors via Input Optimisation

arXiv:2611.17481v1 Announce Type: new 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.

翻訳サービスが停止中のため原文を表示しています。

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. Submission history Access Paper: View PDF HTML (experimental) TeX Source Additional Features Audio Summary Current browse context: References & Citations Google Scholar Semantic Scholar Bookmark Institution 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.

出典

原文を読む ↗

関連テーマ

← すべてのニュースへ