KWW94/Kind-PyTorch-Tutorial

Kind PyTorch Tutorial for beginners

Requirements

• python 3.5.2
• Pytorch (http://pytorch.org/)
• Numpy
• matplotlib

Optional

• visdom (https://github.com/facebookresearch/visdom)

Table of contents

• Pytorch Tensor Basics
• Variable and Automatic Gradient Calculation
• Linear Regression Example (Loss & Optimizer)
• MNIST CNN (cpu & gpu)
• Input Pipeline & Pretrained Model
• Autoencoder & Model Save
• Denoising Autoencoder
• Simple Character RNN
• Vanilla GAN (Layer naming & Multi gpu)
• InfoGAN using Least Squares loss
• Image Style Transfer using ResNet
• Semantic Segmentation (U-Net & Fusion-Net)