Data attribution in text-to-image generative models is a crucial yet underexplored problem, particularly at the regional level, where identifying the most influential training regions for generated ...