Evil and Impolite persona vectors evolve in parallel in MDS space, suggesting intertwined representations

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

• Hypothesis 2 (Persona Space): Persona vectors jointly compose a persona space; an instance's general dispositional profile is specified by a combination of activations along multiple persona vectorshypothesis0.802
  Second of three hypotheses about persona implementation, supported by PCA evidence from Lu et al.
• Chen et al. 2025 identified seven persona vectors in Qwen2.5-7B-Instruct and Llama-3.1-8B-Instruct: evil, sycophancy, hallucination, optimism, impoliteness, apathy, and humorfinding0.791
  Systematic identification of multiple coexisting persona vectors in two open-source models
• There may exist a natural 'persona basis' characterizing the space of all model personas, with correlations between persona vectors predicting co-expression of traitshypothesis0.789
  Open question proposed by authors for future work on the dimensionality and structure of persona space
• The evil persona region is privileged: models fine-tuned on different narrow datasets converge on the same evil misalignment directionclaim0.780
  Support for Hypothesis 3 that the evil persona corresponds to a genuine basin of attraction
• The directional transition pattern of persona vectors at the identified attention layer represents a unique footprint of persona representations, distinct from general hidden state dynamicsclaim0.780
  Supported by comparing persona vector transitions to hidden vector transitions from OpenAssistant data
• Ududec et al. 2026 find that once a model has entered the evil persona region over the course of a conversation, it is difficult to steer it backfinding0.777
  Evidence that the evil persona region exhibits the stickiness hallmark of a genuine attractor basin
• A linear preference vector encoding how much a model likes a given task is persona-relative: it activates strongly for creative writing in the assistant but for phishing in the evil personafinding0.775
  Evidence that core representations like preferences are persona-relative, supporting claim that personas gate content of representations
• Gemma-3-12b-it shows multiple sharp direction shifts in persona vectors across layers, differing from the single-transition pattern in Qwen and Llamafinding0.774
  Suggests architectural variations influence persona localization pattern