method
active
method:activation-addition-steeringActivation Addition Steering
Add scaled steering vector to layer activations.
Neighborhood — ranked by edge-count
Papers (1)
paper
Methods (3)
method
- Activation Steeringrelated_toCausal intervention technique: edit NLA explanation, reconstruct via AR, use difference as steering vector to manipulate model behavior.
- Difference-in-Means SteeringextendsSteering vector from mean activation difference between contrasting datasets.
- Linear Probe Weight SteeringextendsUses normalized probe weights as steering vector.
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- Technique of injecting steering vectors into model activations to test introspection and causal control of emotion vectors.
- Core technique: takes mean difference of model activations on contrastive prompts and adds the resulting vector to the residual stream at inference time.
- Intervention method that adds a learned direction vector to residual stream activations to steer model behavior
- Modifying model behavior by clamping SAE feature activations to specific values during forward pass.
- Steering method deriving vectors from contrastive prompt pairs and adding to first-token activations.
- Foundational paper introducing activation steering methodology used in this work
- Key distinction showing steering offers value beyond prompting; supported by Figure 5 and random vector experiments.
- Adding steering vector in forward direction to push model activations toward stronger reflective behavior.