This code mostly relies on pytorch implementation of Style Space Analysis https://github.com/xrenaa/StyleSpace-pytorch.git and styleCLIP official implementation https://github.com/orpatashnik/StyleCLIP.git
bash docker.sh
W plus space representation of an image of size 1, 18, 512 Each convolutional layer in block with resolution 4, 8, 16, ..., 1024 consist of latent representation