Tag: activation steering
Browse our exclusive articles!
Comprehensive Safety Evaluation for Persona-Imbued LLMs
Discover why single-method safety tests fail for persona-based LLMs and learn the need for dual evaluation to ensure robust AI safety.
Non-Surjective Nature of Steered LLM Activations
Discover why steered LLM activations are non-surjective, limiting prompt-based control and impacting AI interpretability and safety research.
Steering Large Language Models with Persona Vectors in Education
Explore how persona vectors steer large language models in education, impacting response quality and scoring in AI-driven learning tools.
Emotion-Sensitive Decision Making in Small Language Models
Explore how emotion influences decision making in small language models using game-theoretic benchmarks and activation steering techniques.
Non-Identifiability of Steering Vectors in Large Language Models
Explore why steering vectors in large language models are non-identifiable, impacting AI interpretability and alignment strategies in NLP.