Baseline NLI Diversity – MNLI achieves Spearman ρ=0.59 on conTest diversity parameter correlation

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

• Confidence NLI Diversity achieves ρ=0.64 correlation with human diversity judgments on conTestfinding0.852
  Highest human correlation for semantic diversity metric
• # Neutral shows weak correlation with diversity (ρ=0.05 decTest, ρ=-0.08 conTest)finding0.792
  Ablation result showing neutrals are not strong indicators of diversity
• 95% confidence intervals overlap between Baseline NLI Diversity, Confidence NLI Diversity, SentBERT, and human judgments on conTestfinding0.776
  Indicates lack of statistically significant differences between top methods
• Average Spearman correlation of Elo trait rankings between all three models increases to 0.87 after loving constitution character trainingfinding0.772
  Post-training inter-model convergence in trait preferences demonstrating persona convergence
• 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^-105finding0.770
  Validates robustness of alignment metric choice
• Confidence NLI Diversity achieves state-of-the-art performance on measuring semantic diversityclaim0.769
  Main performance claim of the paper
• Diversity Threshold Generation results in average 137% increase in NLI Diversity compared to standard generationfinding0.767
  Key headline result of the DTG procedure across all conditions
• Interest concept: Spearman ρ=0.76, isotonic R²=0.54 between logit self-report and probe score in LLaMA-3.2-3B (n=400)finding0.765
  Strongest pooled introspective coupling across the four emotive concepts in the primary model