Arxiv

FasterViT: Fast Vision Transformers with Hierarchical Attention
FasterViT: Fast Vision Transformers with Hierarchical Attention
Vision transformer architecture with new hierarchical attention optimezed for throughput and high resolution images.
A. Hatamizadeh, G. Heinrich, H. Yin, A. Tao, J. Alvarez, J. Kautz, P. Molchanov
PDF Cite Code