EfficientViT: Memory Efficient Vision Transformer with Cascaded Group Attention. (arXiv:2305.07027v1 [cs.CV])

Vision transformers have shown great success due to their high model
capabilities. However, their remarkable performance is accompanied by heavy
computation costs, which makes them unsuitable for real-time applications. In
this paper, we propose a family of high-speed vision transformers named
EfficientViT. We find that the speed of existing transformer models is commonly
bounded by memory inefficient operations, especially the tensor reshaping and
element-wise functions in MHSA. Therefore, we design a new building block with
a sandwich layout, i.e., using …

applications arxiv attention computation costs family memory paper performance real-time real-time applications speed success transformer transformers vision