Models
A searchable catalog of models covered.
-
Alexnet
Primary early deep convolutional neural net for use in images
-
FCnet
Fully convolutional network which introduced tconvs for upsampling a la unets
-
Gated PixelCNN
Variation of pixelcnn which has two stacks of convolutions with gated activations for maximal receptive field size
-
Pixel Recurrent Neural Networks
Autoregressive model which is an important immediate ancestor to many important convnets
-
Pixelcnn++
Very strong early convnet that was the backbone for the first diffusion nets
-
Resnet
Primary progenitor in the use of residual layers
-
Resnet VAE
Variation of pixelcnn which has two stacks of convolutions with gated activations for maximal receptive field size
-
Unet
Core image2image architecture and inspiration for backbone net in early diffusion models
-
VGGnet
Important architectural predecessor to modern deep convnets