Activation-Space Personality Steering: Hybrid Layer Selection for Stable Trait Control in LLMs

智能体社会模拟 分类

2025

11-22

Activation-Space Personality Steering: Hybrid Layer Selection for Stable Trait Control in LLMs

11-21

BELIEF DYNAMICS REVEAL THE DUAL NATURE OF IN-CONTEXT LEARNING AND ACTIVATION STEERING

11-21

Can Role Vectors Affect LLM Behaviour?

11-21

Personality as a Probe for LLM Evaluation: Method Trade-offs and Downstream Effects

11-21

Psychological Steering in LLMs: An Evaluation of Effectiveness and Trustworthiness

11-21

PROGRAMMABLE COGNITIVE BIAS IN SOCIAL AGENTS

11-20

PERSONA VECTORS: MONITORING AND CONTROLLING CHARACTER TRAITS IN SLANGUAGE MODELS

11-20

BILLY : Steering Large Language Models via Merging Persona Vectors for Creative Generation

11-19

Distribution Shift Alignment Helps LLMs Simulate Survey Response Distributions

11-19

Language Model Probabilities are Not Calibrated in Numeric Contexts