Predicting Where Steering Vectors Succeed
This provides practitioners with a principled, low-cost method to predict steering vector success, replacing the current trial-and-error approach.
The paper introduces the Linear Accessibility Profile (LAP), a training-free diagnostic that predicts where steering vectors will be effective. Across 24 concept families on five models, LAP achieves ρ=+0.86 to +0.91 for predicting steering effectiveness and ρ=+0.63 to +0.92 for layer selection, and a three-regime framework explains when different steering methods work.
Steering vectors work for some concepts and layers but fail for others, and practitioners have no way to predict which setting applies before running an intervention. We introduce the Linear Accessibility Profile (LAP), a per-layer diagnostic that repurposes the logit lens as a predictor of steering vector effectiveness. The key measure, $A_{\mathrm{lin}}$, applies the model's unembedding matrix to intermediate hidden states, requiring no training. Across 24 controlled binary concept families on five models (Pythia-2.8B to Llama-8B), peak $A_{\mathrm{lin}}$ predicts steering effectiveness at $ρ= +0.86$ to $+0.91$ and layer selection at $ρ= +0.63$ to $+0.92$. A three-regime framework explains when difference-of-means steering works, when nonlinear methods are needed, and when no method can work. An entity-steering demo confirms the prediction end-to-end: steering at the LAP-recommended layer redirects completions on Gemma-2-2B and OLMo-2-1B-Instruct, while the middle layer (the standard heuristic) has no effect on either model.