KisuYang / pytorch_deep_learning_models

This repository includes deep learning implementation with pytorch.

Repository related rules

• All codes are written by Python + Pytorch.
• All codes include only core parts (e.g. model only) for simplicity.
• All models MUST have input parameter 'n_classes' to specify the number of output classes and reject others. Others should be defined with default values.
• Accordingly, all models should be able to declared in this form. e.g. "model(n_classes=10)"

Table

Base Network Architecture

• depthwise seperable
• mobilenet_v1
• mobilenet_v2
• squeezenet
• squeezenext
• shufflenet
• efficientnet
• resnet
  • resnet with bottleneck
  • dilated residual network
• inception
• densenet

Semantic Segmentaion

• aspp
• pspnet
• deeplab v3+

Generative Adversarial Networks

• vanilla_gan
• dcgan

Variational Autoencoder