Gradient Concentration, Not Weight Saliency, Explains Representation-Level Class Unlearning | Digital Library | PAMCET | PAMCET