Spearman's rank correlation among different alignment metrics (CKA, SVCCA, Mutual k-NN, CKNNA) over 78 vision models is high across variants, with all p-values below 2.24×10^-105

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

• Average Spearman correlation of Elo trait rankings between all three models increases to 0.87 after loving constitution character trainingfinding0.808
  Post-training inter-model convergence in trait preferences demonstrating persona convergence
• Spearman Rank Correlationmethod0.794
  Used to compare RDMs in RSA computations; noted to have sensitivity issues with differing relative extrema in embedding layers.
• Full-vector regression over 32 traits reaches leave-one-out Spearman correlations ~0.58-0.61 depending on layer windowfinding0.793
  Vector geometry features provide moderate secondary signal beyond elicitation-only screen
• Pairwise structural correlations of inter-trait cosine similarity matrices across all four architectures range from r=0.898 to r=0.986 (all p<0.001), indicating cross-architecture geometric preservation.finding0.787
  Cross-architecture geometric invariance of Big Five steering vectors
• CKA shows a very weak trend of alignment between models even within modality, compared to mutual k-NN which shows stronger trendsfinding0.786
  Explains why mutual k-NN was chosen over CKA as primary metric
• Prompt-side persona danger rankings are preserved across all four architectures with Spearman rho=0.71-0.96 (all p<10^-4).finding0.784
  Cross-architecture universality of SP rankings
• Spearman's Rank Correlationmethod0.782
  Statistical measure used to evaluate correlation between diversity metrics and diversity parameter / human judgments
• Claude achieves significantly higher Spearman correlation predicting feature activations vs neuron activationsfinding0.780
  Automated interpretability analysis of activations confirms features are more interpretable than neurons