{
  "abstract": "Difference-in-means steering requires activations recorded while a model shows the desired behavior, which a sandbagging model withholds by deliberately underperforming. We introduce Activation Flow (ActFlow), which manufactures these activations from $k$ correct labels without fine-tuning. ActFlow sets target logits that rank each labeled item's correct answer first, and moves the logits toward them by adding one vector $x$ to all $k$ residual streams at one layer. ActFlow is a family of ordinary differential equations for $x$, one for each rule that maps the required logit change to the velocity of $x$. The smallest-norm rule lands exactly on the targets, while the others keep only the top singular directions of the Jacobian. We test ActFlow on three instruction-tuned models, each locked by a sandbagging prompt and by a password-locked LoRA. At $k=40$, ActFlow keeping five singular directions raises the mean held-out ARC-Easy accuracy over the six locked models from $0.05$ to $0.85$, against $0.88$ for rank-$16$ LoRA fine-tuning, which trains about $3{,}000$ times as many parameters, and $0.92$ for the honest models. Furthermore, it scores higher than the smallest-norm rule in 16 of the 18 combinations of locked model and $k$, and its steering direction is nearly orthogonal to the honest difference-in-means direction. It also unlocks two LoRA locks where the honest direction fails. Surprisingly, with one labeled item, the single-step smallest-norm rule raises the Qwen2.5-7B prompt lock from $0.04$ to $0.59$.",
  "abstract_shown": true,
  "affiliations": [],
  "article_type": null,
  "arxiv": "2609.32530",
  "author_total": 3,
  "authors": [
    {
      "name": "Hong Kiat Tan"
    },
    {
      "name": "Linh Le"
    },
    {
      "name": "David Williams-King"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/fgkt78xkmv5a9.bib",
  "cited_by": 0,
  "doi": "10.48550/arxiv.2609.32530",
  "fulltext_url": null,
  "id": "fgkt78xkmv5a9",
  "issns": [],
  "landing_url": "https://arxiv.org/abs/2609.32530",
  "language": null,
  "licence": "LicenseRef-arXiv-nonexclusive-1.0",
  "licence_label": "arXiv licence",
  "licence_source": "article",
  "licence_url": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
  "metadata_licence": "CC0-1.0",
  "native_subjects": [
    [
      "arxiv",
      "cs.AI"
    ],
    [
      "arxiv",
      "cs.CL"
    ],
    [
      "arxiv",
      "cs.LG"
    ],
    [
      "arxiv",
      "math.DG"
    ],
    [
      "openalex",
      "1702|Adversarial Robustness in Machine Learning"
    ]
  ],
  "openalex": "W7214722432",
  "pdf_url": "https://arxiv.org/pdf/2609.32530",
  "pmcid": null,
  "pmid": null,
  "published": "2026-10-02",
  "publisher": null,
  "ris": "https://research.draconitic.com/paper/fgkt78xkmv5a9.ris",
  "sources": [
    {
      "first_seen": "2026-10-30T12:27:10Z",
      "native_id": "2609.32530",
      "source": "arxiv",
      "url": "https://arxiv.org/abs/2609.32530"
    }
  ],
  "subjects": [
    {
      "field": "mathematics",
      "origin": "arxiv:math.DG",
      "subfield": "geometry-topology"
    }
  ],
  "title": "Activation Flow: Manufacturing Activations for Steering",
  "title_text": "Activation Flow: Manufacturing Activations for Steering",
  "type": "preprint",
  "updated_at": "2026-10-30T12:27:10Z",
  "url": "https://research.draconitic.com/paper/fgkt78xkmv5a9",
  "venue": "arXiv",
  "versions": []
}