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 humor

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

• Hypothesis 1 (Gateway Features): Persona vectors act as gateway features—single directions in activation space that shape LLM behavior across most, if not all, contextshypothesis0.811
  First of three hypotheses about persona implementation in LLMs, motivating the persona views
• Heads 24, 30, and 32 in layer 14 of Llama-3.1-8B consistently emerged as top contributors for every evaluated persona traitfinding0.810
  Identifies specific Style Modulation Heads in Llama-3.1-8B
• When steered to the extreme away from the Assistant, Llama and Gemma shift to a theatrical persona characterized by mystical, poetic prose; Qwen more often hallucinates a human persona at extremesfinding0.810
  Characterizes what is on the far end of the Assistant Axis away from the Assistant
• For amplifying safety-related personas (evil, sycophancy, hallucination) in Llama-3.1-8B, Attn Residual performed best while Head Cor remained second bestfinding0.802
  Exception to the general Head Cor superiority, suggesting safety personas involve circuits beyond style modulation
• Qwen 3 32B is most likely to hallucinate human personas (names, birthplaces, years of experience) when steered away from the Assistantfinding0.798
  Model-specific difference in how steered personas manifest
• Evil and Impolite persona vectors evolve in parallel in MDS space, suggesting intertwined representationsfinding0.791
  Geometric finding from MDS analysis suggesting shared representational structure
• Persona vectors extracted from base pretraining checkpoints steer fully post-trained OLMo-3-7B-Instructfinding0.791
  Shows persona directions persist through all alignment stages, answering key part of RQ2
• Only three attention heads in a specific layer independently govern persona and style formation in Qwen2.5-7B and Llama-3.1-8Bclaim0.790
  Striking sparsity finding that enables surgical intervention