Masked Cosine Similarity

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

• Cosine Similarity-Based Deception Detectionconcept0.819
  Detection mechanism computing cosine similarity between activation vectors and steering vectors to classify deception
• Cosine Similarity Measurementmethod0.814
  Used to measure alignment between DIM direction and cone basis vectors to assess overlap
• Cosine similarity between truth probesmethod0.797
  Geometric evaluation of truth direction alignment across layers and prompt templates.
• Cosine Similarity Binary Classifiermethod0.786
  Classifier using cosine similarity between activation vectors and steering vectors to detect deception with 89% accuracy
• Pairwise Cosine Similarity Analysismethod0.780
  Used to quantify the semantic clustering of adjective-set embeddings across model families and conditions
• Cosine Similarity Ranking for Instruction Discoverymethod0.752
  Method to discover new reflection-inducing instructions by ranking candidate tokens by cosine similarity to steering vectors.
• In Qwen-2.5-9B, only v1 has meaningful cosine similarity to DIM direction; all additional basis vectors have cosine similarities ~1e-9finding0.747
  Appendix E replication of DIM alignment finding in Qwen model
• Attribution Similaritymethod0.744
  Correlating attribution vectors (feature activation × logit weight of next token) across model pairs to measure functional universality