[2401.00463] Analyzing Local Representations of Self-supervised Vision Transformers