Projection of last prompt token onto persona vector correlates r=0.75–0.83 with subsequent trait expression under system prompt variations

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

• Finetuning shift along persona vectors correlates r=0.76–0.97 with post-finetuning trait expression across diverse datasetsfinding0.833
  Core empirical result showing persona vectors capture trait-specific signal mediating finetuning-induced persona shifts
• Do correlations between persona vectors predict co-expression of the corresponding traits?question0.816
  Specific open question about the predictive value of persona vector geometry for behavioral co-expression
• Long Persona system prompt increases non-fixed-point token percentage to 0.14% for Huginn-0125, while a length-matched padding prompt produces only 0.05%finding0.805
  Shows that semantic content (not just length) of system prompt influences occurrence of orbit/slider behavior
• Cosine similarity between earliest extractable checkpoint persona vector and final-pretraining direction starts at approximately 0.3finding0.798
  Quantifies geometric distance of early persona directions from final direction
• There may exist a natural 'persona basis' characterizing the space of all model personas, with correlations between persona vectors predicting co-expression of traitshypothesis0.793
  Open question proposed by authors for future work on the dimensionality and structure of persona space
• Persona vectors capture signal specific to their assigned trait, as evidenced by within-trait correlations exceeding cross-trait baselinesclaim0.793
  Author's interpretation establishing that persona vectors are not merely general misalignment indicators
• Persona vectors form within 0.22% of OLMo-3 pretraining (≈12.6B tokens)finding0.790
  Core quantitative result answering RQ1 for OLMo-3
• Personas are latent factors that persist for many tokens, so recent expression of a persona predicts near-future expressionclaim0.789
  Author's hypothesis explaining why persona vectors extracted from exhibited-trait activations generalize to causal influence