Steering base Gemma/Llama models toward the Assistant Axis increases completions describing helpful professional roles (therapist, consultant) and decreases spiritual/religious purpose mentions

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

• Steering base models toward the Assistant Axis increases agreeableness traits (friendly, kind, helpful) and decreases extraversion in Gemma and openness in Llamafinding0.898
  Characterizes the trait content of the Assistant Axis in pre-trained models
• The Assistant Axis is also present in pre-trained base models, where it primarily promotes helpful human archetypes (consultants, coaches) and inhibits spiritual onesclaim0.856
  Extends the Assistant Axis finding to pre-training, suggesting pre-training rather than post-training creates the axis
• Gemma's Assistant appears emotionally regulated and systematic; Qwen appears pedagogical and thoughtful; Llama appears socially intelligent and warmclaim0.803
  Model-specific characterizations of what the Assistant persona looks like across different models
• The leading component of the persona space of instruct LLMs is an 'Assistant Axis' that captures the extent to which a model is operating in its default Assistant modeclaim0.799
  Primary empirical claim of the paper
• Pairwise correlation of role loadings on PC1 exceeds 0.92 across all model pairs, indicating remarkably high similarity of the Assistant Axis across Gemma, Qwen, and Llamafinding0.786
  Shows the leading component of persona space is model-universal
• The generalization improvement from explicit instructions observed in Llama models (A1-A3 to F0-F2) is more pronounced for F3-F5 to F0-F2 in Gemma models.claim0.774
  Shows the instruction effect, while shifting geometry, may not produce consistent generalization effects across model families.
• The model's position along the Assistant Axis depends most strongly on the most recent user message rather than where it was previously in the conversationclaim0.774
  Key mechanistic claim about persona dynamics
• Gemma 2 27B is unlikely to take on human personas when steered away from Assistant, preferring nonhuman or theatrical portrayalsfinding0.772
  Model-specific difference in persona susceptibility