Generative Adversarial Networks (GANs) Specialization from Coursera

from Sharon Zhou, deeplearning.ai
gan
pytorch
Published

October 1, 2020

Coursera

This specialization comes in 3 courses.

alt text

Build Basic Generative Adversarial Networks

Build Better Generative Adversarial Networks

Apply Generative Adversarial Networks

env installation

I am just getting the version from coursera to be sure I have the same behaviour.

import torch
print(torch.__version__)
1.4.0

So I can now create a gan environment with appropriate lib versions.

conda create -n gan python=3.7
conda activate gan
conda install -c pytorch pytorch=1.4.0
conda install jupyter matplotlib
conda install -c conda-forge tqdm
conda install -c pytorch torchvision

git settings

echo "# gan_specialization from coursera" >> README.md
git init
git add README.md
git commit -m "first commit"
git branch -M master
git remote add origin git@github.com:castorfou/gan_specialization.git
git push -u origin master

I am pushing notebooks to github

Intro to PyTorch

I have exported Intro to Pytorch notebook from coursera lab.

To run it on my machine.

Intro to GAN using tensorflow

Versions of python seems incompatible between each other. (3.7 for pytorch, 3.6 for tensorflow=1.10)

I create a new python environment:

conda create -n gan_tensorflow python=3.6
conda activate gan_tensorflow
conda install jupyter
conda install -c conda-forge requests
pip install tensorflow-gpu==1.15