Average Spearman correlation of Elo trait rankings between all three models increases to 0.87 after loving constitution character training

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

• 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^-105finding0.808
  Validates robustness of alignment metric choice
• Full-vector regression over 32 traits reaches leave-one-out Spearman correlations ~0.58-0.61 depending on layer windowfinding0.806
  Vector geometry features provide moderate secondary signal beyond elicitation-only screen
• Spearman Rank Correlationmethod0.794
  Used to compare RDMs in RSA computations; noted to have sensitivity issues with differing relative extrema in embedding layers.
• Standard deviation of Elo trait scores increases dramatically after character training, indicating more opinionated trait preferencesfinding0.792
  Distribution-level finding showing polarization of trait preferences post character training
• Spearman's Rank Correlationmethod0.773
  Statistical measure used to evaluate correlation between diversity metrics and diversity parameter / human judgments
• Baseline NLI Diversity – MNLI achieves Spearman ρ=0.59 on conTest diversity parameter correlationfinding0.772
  Comparable to top-performing automatic metric from Tevet and Berant 2021
• Claude achieves significantly higher Spearman correlation predicting feature activations vs neuron activationsfinding0.765
  Automated interpretability analysis of activations confirms features are more interpretable than neurons
• Prompt-side persona danger rankings are preserved across all four architectures with Spearman rho=0.71-0.96 (all p<10^-4).finding0.763
  Cross-architecture universality of SP rankings