Improved Representation Steering for Language Models

TL;DR

Steering methods for language models (LMs) seek to provide fine-grained and interpretable control over model generations by variously changing model inputs, weights, or representations to adjust behav…

Opening excerpt from the authors’ abstract. source