ExpertLens: Activation Steering Features Are Highly Interpretable
"概念表示" 相关文章
Do Sparse Subnetworks Exhibit Cognitively Aligned Attention? Effects of Pruning on Saliency Map Fidelity, Sparsity, and Concept Coherence
SPARC: Concept-Aligned Sparse Autoencoders for Cross-Model and Cross-Modal Interpretability
