Persona vectors form within 0.22% of OLMo-3 pretraining (≈12.6B tokens)

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

• Persona vectors are present from the earliest available Apertus-8B checkpoint at 1.4% of pretraining (210B tokens)finding0.868
  Replication finding on Apertus confirming early persona formation generalizes across model families
• Persona vectors extracted from base pretraining checkpoints steer fully post-trained OLMo-3-7B-Instructfinding0.861
  Shows persona directions persist through all alignment stages, answering key part of RQ2
• Evil persona vector saturates by approximately 12B tokens in OLMo-3finding0.840
  Quantitative result showing Evil emerges earliest due to ubiquity and simplicity in pretraining data
• Apertus-8B persona vectors show reduced geometric drift compared to OLMo-3, consistent with earliest checkpoint being later in relative pretrainingfinding0.837
  Explains geometric differences in replication through earlier consolidation of Apertus persona space
• Evil persona steering delta of +9.69 (p=0.011) at stage1-step3000 (12.6B tokens) in OLMo-3finding0.810
  First significant evil persona steering result; establishes emergence at 0.22% pretraining
• Cosine similarity between earliest extractable checkpoint persona vector and final-pretraining direction starts at approximately 0.3finding0.800
  Quantifies geometric distance of early persona directions from final direction
• Projection of last prompt token onto persona vector correlates r=0.75–0.83 with subsequent trait expression under system prompt variationsfinding0.790
  Main monitoring result showing persona vectors can predict behavioral shifts before text generation begins
• Persona vectors act as gateway features that gate access to broad repertoires of inferential paths already encoded in the weightsclaim0.785
  Interpretive claim about the mechanistic role of persona vectors explaining emergent misalignment