AI-enabled systems are becoming increasingly present in our everyday lives, and as we rely more and more on their decisions, it is important to understand their reasoning. Whether it is for personal use or in a professional context, the decisions can have significant impact. To improve on creating informed decisions, we propose an interactive visualization tool designed to make dual encoder attributions explorable and understandable for users, ranging from specialized developers, trying to optimize AI systems to inexperienced, causal users seeking a more solid foundation to their understanding of AI behavior. We built a web-based application, capable of visualizing both text to text and text to image attributions, with focus on usability, consistency and support for more exploratory analysis by expert users. Our primary visualizations include interactive matrices, histograms and Part of Speech (POS)-tags for filtering, adaptive heatmaps as well as more accessible approaches, like texts enriched with basic attribution information. All these approaches are deeply interconnected and integrated in one easy-to-deploy application. The application shows that interactive visualizations can significantly enhance the interpretability of dual encoder attributions, providing insights for both expert and casual users, but also creating foundations for further integration with AI-driven systems.