Back to papers

GitHub - wassname/w2schar-mini: Weak to strong character steering. Can a small model use weight steering to align a stronger model?

Abstract

Weak to strong character steering. Can a small model use weight steering to align a stronger model? - wassname/w2schar-mini

Research area

character steeringrepresentation engineering
Published
Source
other
View paper
Sign in to read and join the discussion.