Using cosine similarity instead of dot product for Head Contribution Score yields a consistent ranking of the same high-contribution heads

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

• In Qwen-2.5-9B, only v1 has meaningful cosine similarity to DIM direction; all additional basis vectors have cosine similarities ~1e-9finding0.785
  Appendix E replication of DIM alignment finding in Qwen model
• Cosine Similarity Measurementmethod0.768
  Used to measure alignment between DIM direction and cone basis vectors to assess overlap
• Cosine similarity between perturbed and baseline residual streams returns toward 1.0 and projection onto injection direction decays exponentially over subsequent layersfinding0.766
  Mechanistic evidence that network actively attenuates injected perturbations, explaining late-layer introspection failure
• Layer-wise Cosine Similarity Analysismethod0.765
  Geometric analysis tracking how persona vector directions evolve across transformer layers to identify the transition layer
• Generated statements achieve 85.62%-94.00% cosine similarity alignment with Perez et al. validated OCEAN and Dark Triad statementsfinding0.759
  Validates the statement synthesis pipeline as producing behavior-specific content comparable to established methods
• There is a positive association between Head Contribution Score and the resulting persona gain, suggesting higher-scoring heads tend to induce stronger steering effectsclaim0.757
  Validates the Head Contribution Score as a proxy for functional importance
• Cosine Similarity Ranking for Instruction Discoverymethod0.756
  Method to discover new reflection-inducing instructions by ranking candidate tokens by cosine similarity to steering vectors.
• SP and AS produce activation cosine similarity of only 0.11-0.20 at safety-critical layers, compared to 0.83-0.92 for SP vs FS, confirming two representationally distinct pathways.finding0.752
  Mechanistic evidence for two distinct representational pathways