Metraipolis › archive › dc_749643e9524e

Sweep Less, Steer More: Using Activation Steering Geometry for Reduced Hyperparameter Search — arXiv:2611.01496

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.01496
License (as archived)
CC0-1.0
Archived
2026-11-03T18:17:55.999Z
SHA-256 of the archived text
d40830d085fac2b624b606c8b349d56f618bb3a5191dc99fb91067a1d5388faa
Stable URL of this page
https://metraipolis.com/documento/dc_749643e9524e

How to cite

Sweep Less, Steer More: Using Activation Steering Geometry for Reduced Hyperparameter Search — arXiv:2611.01496 — arXiv (en), https://arxiv.org/abs/2611.01496 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_749643e9524e (Metraipolis archive dc_749643e9524e, sha256 d40830d085fac2b624b606c8b349d56f618bb3a5191dc99fb91067a1d5388faa)
BibTeX
@misc{dc749643e9524e,
  title = {Sweep Less, Steer More: Using Activation Steering Geometry for Reduced Hyperparameter Search — arXiv:2611.01496},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.01496},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_749643e9524e; sha256 d40830d085fac2b624b606c8b349d56f618bb3a5191dc99fb91067a1d5388faa},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.01496. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Representation steering methods like difference-in-means (DIM) offer a lightweight approach to controlling LLM behavior, but their effectiveness varies widely across concepts, datasets, and layers, making principled deployment costly without exhaustive hyperparameter sweeps. We build on the discriminability index $d'$, a compute-cheap geometric measure of how well contrastive training activations separate along the steering direction. Across five behaviors on 3 models, we demonstrate uses beyond its established role as a dataset-level steerability predictor. First, in the multiple-choice question answering (MCQA) setting, we show that $d'$ reliably identifies performant layers at which to steer, eliminating the need for a layer-wise sweep. Second, we extend this analysis to open-ended multi-token generation, and find that $d'$ remains successful for layer selection across different settings. Lastly, we test whether the position of a generated token's representation on the DIM line predicts steering success on that individual example. In the MCQA setting, this works for high-$d'$ datasets but is inconsistent across behaviors, models, and training-token choices in open-ended generation. This highlights that methods developed and conclusions drawn in single-token MCQA settings do not necessarily transfer to open-ended generation.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI); Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2611.01496 — 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 2026-11-03T18:09:53.592Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).