Xiaoyong Xue

Sparse Relaxed-Lasso Steering: Automatic Sparse-Autoencoder Feature Selection for Precise Image Editing

Precise, training-free editing of text-to-image diffusion models requires balancing alignment (faithful attribute manifestation), consistency (preserving non-target content), and quality (artifact-free textures). Sparse autoencoder (SAE) steering offers interpretable, smooth ``slider-like'' control …