Contrastive Covariance Subspaces for Activation Steering
Activation steering controls large language model behavior by adding a direction to hidden activations during inference. For steering to work well, that direction must be a pure signal for the target concept: any confound entangled with it, such as an incidental topic or writing style, is injected into the residual str...
Alexandros Xenos, Manos Plitsis, N. Foteinopoulou et al.
· 0 citations