One Vector or Many? Clustered Optimal-Transport Steering under Quantization
arXiv:2610.40055v1 Announce Type: cross Abstract: Activation steering changes what a language model generates by editing its hidden states at inference time. Its most common form adds one vector, the difference of class means (DiM), to every hidden state. CHaRS (concept heterogeneity-aware representation steering) uses optimal transport instead: an entropic plan pairs the clusters of the two classes, and each hidden state moves along a mixture of the cluster displacements, weighted toward its n…