CASA: Cross-Attention via Self-Attention for Efficient Vision-Language Fusion Paper • 2512.19535 • Published 4 days ago • 8
