Metraipolis › archive › dc_f8f20591d52e
Deployable Per-Instance Multi-Layer Activation Steering for Large Language Models — arXiv:2608.08829
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.08829
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-19T08:50:07.421Z
- SHA-256 of the archived text
be531c50e749d135182b57279797683282c34cdfa1749e946bd12a3ec6169801- Stable URL of this page
https://metraipolis.com/documento/dc_f8f20591d52e
How to cite
Deployable Per-Instance Multi-Layer Activation Steering for Large Language Models — arXiv:2608.08829 — arXiv (en), https://arxiv.org/abs/2608.08829 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_f8f20591d52e (Metraipolis archive dc_f8f20591d52e, sha256 be531c50e749d135182b57279797683282c34cdfa1749e946bd12a3ec6169801)
BibTeX
@misc{dcf8f20591d52e,
title = {Deployable Per-Instance Multi-Layer Activation Steering for Large Language Models — arXiv:2608.08829},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.08829},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_f8f20591d52e; sha256 be531c50e749d135182b57279797683282c34cdfa1749e946bd12a3ec6169801},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.08829. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Activation steering edits the behaviour of a frozen language model by adding a learned vector to its residual stream, and current practice fixes the injection layers globally per task. We argue that the best layers are an instance-level decision, and we make per-instance, multi-layer selection both well understood and deployable. On two open-weight 8B models and six binary persona traits, a per-instance oracle over layer subsets shows that the best layers vary from one input to the next: on most trait-model pairs, no fixed global layer set recovers the per-instance benefit. A greedy rule that ranks layers by single-layer marginal effect recovers nearly all of the oracle's benefit, but both must score candidate layers against the gold answer, so neither can run at deployment; the rule instead becomes the target a prompt-only predictor is trained to reproduce. Our deployable recipe needs no label at inference: a per-instance layer ranker read off the prompt embedding, a classifier that infers the steering direction, and an adaptive gate that scores short steered passes against that inferred direction and steers no more layers than necessary. The recipe recovers most of the oracle's lift (the bulk on the stronger model, a clear majority on the harder one), never drives any trait-model pair below its unsteered alignment baseline on average, and largely avoids the fluency collapse that strong global selection incurs at higher layer counts. A mechanistic account, "direction over magnitude", explains the behavioural flip under a mis-directed global set, the output collapse from steering too many layers, and the ceiling of unsteerable inputs. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-11 in cs.CL). Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2608.08829 — 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 2026-10-19T08:49:41.876Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-08-11?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).