Cited publications in International Conference on Learning Representations:Graph Neural Networks Can (Often) Count Substructures (2025)