Introduction to PyTorch

StatML 2020/2021 CDT Introduction to PyTorch tutorial

This repository contains 3 notebooks

01-introduction.ipynb: Introduction to the fundamentals of PyTorch and automatic differentation
02-deep-kernel-learning-challenge-student.ipynb: Hands-on implementation of deep kernel learning with PyTorch
03-dataloading-with-pytorch.ipynb: Basics of dataloading pipelines with PyTorch

Installation

For this tutorial, you are roughly going to need a Jupyter kernel with torch, gpytorch, torchvision, and usual scientific libraries installed. If your usual python environment already has all of this, you can go ahead and use it. Otherwise, follow the below instruction to set up an environment.

Code implemented with Python 3.8. Instructions written for 3.8.0 but can be adapted.

Clone and go to repository

$ git clone https://github.com/shahineb/statml-cdt-pytorch-tutorial.git
$ cd statml-cdt-pytorch-tutorial

Setting up environment

Create and activate a dedicated environment with your favorite virtual environment management tool.

With virtualenv :

$ virtualenv --python=python3.8 venv-pytorch-tutorial
$ source venv-pytorch-tutorial/bin/activate
$ (venv-pytorch-tutorial)

With pyenv :

$ pyenv virtualenv 3.8.0 venv-pytorch-tutorial
$ pyenv activate venv-pytorch-tutorial
$ (venv-pytorch-tutorial)

With conda :

$ conda create --name venv-pytorch-tutorial python=3.8
$ source activate venv-pytorch-tutorial
$ (venv-pytorch-tutorial)

Install dependencies

$ (venv-pytorch-tutorial) pip install -r requirements.txt

Create Jupyter Kernel

$ (venv-pytorch-tutorial) python -m ipykernel install --user --name pytorch-tutorial  --display-name "pytorch-tutorial"

You can now choose kernel named pytorch-tutorial from Jupyter.