{"slug": "the-steering-budget-examples-beat-knobs", "title": "The Steering Budget: Examples beat Knobs", "summary": "Researchers have found that generative models have a fixed 'steering budget' set by training data, limiting how much knobs like prompts can adjust properties. They show that using examples—concrete instances composed from learned data—can reach a larger, often unreachable part of this budget, enabling more expressive and complete steering. The method is verified in image and crystal-structure generation, offering a way to audit and fully exploit the budget.", "body_md": "arXiv:2607.14246v1 Announce Type: new\nAbstract: Generative models are steered with knobs -- prompts, guidance scales, property tags. Turn one as hard as you like and, past a point, it stops moving the property you care about. We find that ceiling is not a shortcoming of the model but a budget, set by the training data before the model is trained: a property's movable range splits in two -- the part a knob can reach, and a second, significant part that only examples -- concrete instances of what you want more of -- can reach. That second part is usually much larger, but not always, and the same budget says so in advance.\nReaching that second part takes a different move: instead of turning a knob, you show the model examples, composed from what it already learned rather than added to its training. A cheap audit of the training data measures the budget; we give a recipe for building the example set that reaches all of it.\nThis buys two things a knob can't. Reach: it moves a property across the whole budget, not just the part a knob reaches. Expressiveness: it steers toward targets you can only specify by example -- including ones you can't put into words. We turn these into a handful of falsifiable claims and verify them in two unrelated domains, image and crystal-structure generation -- marking where a knob is enough, and where only examples will do.", "url": "https://wpnews.pro/news/the-steering-budget-examples-beat-knobs", "canonical_source": "https://arxiv.org/abs/2607.14246", "published_at": "2026-07-17 04:00:00+00:00", "updated_at": "2026-07-17 04:27:55.366899+00:00", "lang": "en", "topics": ["generative-ai", "machine-learning", "ai-research"], "entities": [], "alternates": {"html": "https://wpnews.pro/news/the-steering-budget-examples-beat-knobs", "markdown": "https://wpnews.pro/news/the-steering-budget-examples-beat-knobs.md", "text": "https://wpnews.pro/news/the-steering-budget-examples-beat-knobs.txt", "jsonld": "https://wpnews.pro/news/the-steering-budget-examples-beat-knobs.jsonld"}}