Abstract: Vision Transformers (ViTs) are new foundation models for vision applications. Edge-deploying ViTs to realize energy-saving, low-latency, and high-performance dense predictions have wide ...