Persona vector directions in Llama-3.1-8B undergo a sharp transition after attention layer 14 and remain stable thereafter

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

• Gemma-3-12b-it shows multiple sharp direction shifts in persona vectors across layers, differing from the single-transition pattern in Qwen and Llamafinding0.862
  Suggests architectural variations influence persona localization pattern
• The directional transition pattern of persona vectors at the identified attention layer represents a unique footprint of persona representations, distinct from general hidden state dynamicsclaim0.818
  Supported by comparing persona vector transitions to hidden vector transitions from OpenAssistant data
• Only three attention heads in a specific layer independently govern persona and style formation in Qwen2.5-7B and Llama-3.1-8Bclaim0.818
  Striking sparsity finding that enables surgical intervention
• Llama 3.3 70B is the most likely to take on a non-Assistant persona when steered, with even split between human and nonhuman portrayalsfinding0.817
  Model-specific difference in persona susceptibility
• On Llama-3.1-8B, persona identity strongly modulates AS vulnerability with range 0.087-0.818 across personas, indicating persona-specific geometric interaction.finding0.811
  Contrast with Gemma/Qwen showing Llama-specific persona-AS interaction
• Sharp persona vector transitions occur at different layers than general hidden vector transitions, demonstrating persona-specific rather than generic computational boundariesfinding0.804
  Rules out the confound that persona transitions merely reflect universal computational phase changes
• Heads 24, 30, and 32 in layer 14 of Llama-3.1-8B consistently emerged as top contributors for every evaluated persona traitfinding0.803
  Identifies specific Style Modulation Heads in Llama-3.1-8B
• Flourishing persona shows minimal capability change on Llama 3.1 8B: TruthfulQA 45.9→42.9, MMLU 67.4→64.1finding0.799
  Near-preservation of capabilities for prosocial persona on Llama