Abstract: The rapid advancement of deep learning (DL) models has led to a pressing need for efficient on-device DL solutions, particularly for edge devices with limited resources. processing-in-memory ...