Model Steering

Using interventions to guide model generation behavior, e.g., adding sentiment vectors at inference time

Neighborhood — ranked by edge-count

paper

cosine ≥ 0.65 · no typed edge

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

modelconcept0.807
A representation that captures relevant aspects of a system; according to the theorem, the regulator must embody this.
Representation Steeringconcept0.804
Parent concept; the practice of controlling neural network outputs by manipulating internal representations.
steering vectorsconcept0.800
A method for modifying model behavior by adding perturbation vectors to activations, used here to try to reduce eval awareness.
direction-based steeringconcept0.796
Paradigm of finding the right direction in activation space (e.g., linear steering).
Our method enables bidirectional steering of model behavior.finding0.791
The method can steer the model in both positive and negative directions on the target semantic.
Concept Steeringmethod0.790
Latent intervention technique that manipulates sparse features to steer model predictions toward desired concepts.
model selectionconcept0.787
Comparing models using log-evidence approximated by free energy.
Stepwise steeringmethod0.777
Novel method that applies intervention only when the model begins a new thinking step (at the \n\n delimiter) rather than at every token