Global context ViT

tags

Transformers, Computer vision, ViT

paper

(Hatamizadeh et al. 2022)

Architecture

This is a hierarchical version of ViT with both local and global attention.

Parameter count

90M

Bibliography

1. Ali Hatamizadeh, Hongxu Yin, Jan Kautz, Pavlo Molchanov. June 20, 2022. June 20, 2022DOI.