Probing and Steering for Safety-Relevant Behaviours in Large Language Models

Nimun Bajwa, Massimo Poesio

Activation probes are computed directions in a language model's residual stream, used both to detect safety-relevant behaviours and, by adding their direction back into the stream,…