method
active
method:linear-steering

linear steering

Typical approach that adds a scaled steering vector to representations; the paper argues this is mismatched with actual representation geometry.

Neighborhood — ranked by edge-count

Findings (2)

finding

Concepts (7)

concept
  • Manifold Steering
    contradictsextends
    Central framework: steering neural networks by intervening along the curved manifold where a concept lives, rather than in straight lines through activation space.
  • A method for modifying model behavior by adding perturbation vectors to activations, used here to try to reduce eval awareness.
  • The implicit assumption of linear steering methods, which the paper argues is inappropriate for neural activation spaces
  • Linear steering implicitly assumes a flat, Euclidean activation space, leading to off-manifold excursions.
  • Parent concept; the practice of controlling neural network outputs by manipulating internal representations.
  • Paradigm of finding the right direction in activation space (e.g., linear steering).
  • General technique of modifying activations to control model behavior.

Related by similarity (8)

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.