Search results for: 'From One-to-One to Many-to-Many: Dynamic Cross-Layer Injection for Deep Vision-Language Fusion'