Dark mode
Search
6 items with this tag.
Steering Llama-2 with Contrastive Activation Additions
People Care About Each Other Even Though They Have Imperfect Motivational Pointers?
Formalizing Policy-Modification Corrigibility
A Certain Formalization of Corrigibility Is VNM-Incoherent
Non-Obstruction: A Simple Concept Motivating Corrigibility
Corrigibility as Outside View