Runjin Chen, Andy Arditi, Henry Sleight, Owain Evans, Jack Lindsey · 2025 · arXiv
Automatically extracts directions for character traits such as sycophancy and ‘evil’, then uses them to monitor, steer and predict trait shifts caused by fine-tuning.
Atlas / Digital minds / Paper
Persona vectors: monitoring and controlling character traits in language models
Personas & Functional Emotions
Help improve this map
Know a paper that belongs here, or think something needs correcting?
Suggest a correction ↗Contributions are reviewed before they appear.