Content-Aware Token Sharing for Efficient Semantic Segmentation With Vision Transformers | Read Paper on Bytez