Persona region

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

• Personaconcept0.818
  Stable, reidentifiable dispositional profiles that can be interpreted as characters with broadly coherent beliefs, values and traits
• Persona Spaceconcept0.807
  Low-dimensional space of activation directions corresponding to diverse character archetypes in LLMs
• Persona Featureconcept0.791
  An SAE latent representing a simulated character with consistent traits, as opposed to a specific trait
• Persona Jailbreakconcept0.767
  Jailbreak technique that attempts to change the model's understanding of the assistant character's personality and constraints
• Persona Vectorconcept0.762
  A linear direction in a language model's activation space that encodes a specific personality trait, enabling monitoring and control of that trait
• Persona Constructionconcept0.760
  The process of building a coherent model persona from character archetypes and traits during training
• Persona Fidelityconcept0.758
  The degree to which an LLM's generated responses consistently reflect an assigned persona
• The persona region is not continuously maintained during user turns but is active only while the model is producing its own tokensclaim0.756
  Finding from mini experiment 1 that the assistant axis tracks the user independently of the current generation persona during user tokens