Orthogonal Rotation Matrix

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

• Orthogonality thesisframework0.740
  Classical thesis that any level of intelligence can be combined with any goal.
• Rank-one matrixconcept0.732
  The simple matrix form into which VPD constrains subcomponents to enforce mechanistic simplicity.
• Copying Matrixconcept0.732
  An OV circuit matrix that maps tokens to increasing the logit of those same tokens; detectable via positive eigenvalues
• Orthogonal Decomposition of Representation Spaceconcept0.731
  Mathematical structure central to distributed interchange interventions; representation space decomposed into orthogonal subspaces each aligned with a high-level variable.
• rotation symmetryconcept0.730
  Rotation about a point by an angle 360°/n leaving the pattern invariant
• Learned rotation matrices are non-trivial: majority of basis vectors are rotated, indicating highly distributed representationsfinding0.730
  Learned rotations reveal that direct probes over standard activation bases would miss the actual causal role of representations.
• Rotational Symmetryconcept0.704
  One of four key isometries; rotation about a point through a given angle, with n-fold variants based on 360°/n invariance.
• Weight matrix decompositionconcept0.697
  The core idea of decomposing weight matrices into components for interpretability.