Metraipolis › archive › dc_bc2bb26767a2

Signatures of Steerability in Activation Space of Language Models — arXiv:2609.14151

Source: arXiv (en)
Original: https://arxiv.org/abs/2609.14151
License (as archived): CC0-1.0
Archived: 2026-10-15T22:50:55.721Z
SHA-256 of the archived text: 325684425a9933af279aedb9cfbf86fa375f64655795bcf46aa09b2eb1fba0ad
Stable URL of this page: https://metraipolis.com/documento/dc_bc2bb26767a2

How to cite

Signatures of Steerability in Activation Space of Language Models — arXiv:2609.14151 — arXiv (en), https://arxiv.org/abs/2609.14151 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_bc2bb26767a2 (Metraipolis archive dc_bc2bb26767a2, sha256 325684425a9933af279aedb9cfbf86fa375f64655795bcf46aa09b2eb1fba0ad)

BibTeX

@misc{dcbc2bb26767a2,
  title = {Signatures of Steerability in Activation Space of Language Models — arXiv:2609.14151},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.14151},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_bc2bb26767a2; sha256 325684425a9933af279aedb9cfbf86fa375f64655795bcf46aa09b2eb1fba0ad},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.14151. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Steering language models using a set of contrastive representations has been a canonical and computationally efficient method for controlling model behavior. Despite this success in controlling certain model behaviors, the effectiveness of activation steering varies markedly across concepts; the generalization properties of steering vectors are often considered a function of the dataset used to construct them. We make this dataset-dependence claim more rigorous and show that simple separation metrics strongly correlate with the downstream steerability of language models across diverse settings, even after controlling for layers and dataset effects. Beyond prediction, we provide evidence from a synthetic superposition experiment that separation metrics are strongly correlated with alignment between the empirical and true feature direction. Our results suggest that simple separability statistics can serve as practical diagnostics for when steering vectors are likely to work.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-15 in cs.AI).
Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.14151 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-15T22:29:03.630Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-15?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).