PyTorch Tutorial

This repository contains the material for the "PyTorch Tutorial" I gave at the PyData Berlin 2018 conference.

Post for the tutorial https://www.youtube.com/watch?v=_H3aw6wkCv0

Video https://nodata.science/pydata-pytorch-tutorial.html

Content

• 00-index

Setup

Please make sure conda is installed.

Manual install

Due to some problems with the environment.yml here are instructions for a manual installation:

# create conda environment named ppt
conda new --name ppt
source activate ppt

# Install dependencies
conda install -y matplotlib numpy scipy tensorflow
# For linux
conda install -y pytorch-cpu torchvision-cpu ignite -c pytorch
# For Mac
conda install -y pytorch torchvision ignite -c pytorch
# general
conda install -y jupyterlab -c conda-forge
pip install tensorboardX tensorboard scikit-learn
pip install -e .

Install with environment.yml

Then:

# create a conda environment
conda env create -f environment.yml

activate the conda environment

source activate pydata_pytorch_tutorial

and install the ppt package (this project basically)

pip install -e .

Mac

If you have problems with the dependencies under mac check out this issue: sotte#2

# You might have to use the `environment_mac.yml`
conda env create -f environment_mac.yml
# and manually update freetype and matplotlib
conda update freetype matplotlib

Download data and models

Download data and models for the tutorial:

python download_data.py

Then you should be ready to go. Start jupyter lab

jupyter lab

Misc

To get the Table of Contents displayed within jupyter lab do the following:

# install node
conda install -c conda-forge nodejs
# install the toc extension
jupyter labextension install jupyterlab-toc