GNN

graph neural network

A colaborative work of :

1. Aalok, aalok@iitgn.ac.in
2. Anshul, av.vermaans@gmail.com
3. Ashish, ashish.tiwari@iitgn.ac.in
4. Prajwal, singh_prajwal@iitgn.ac.in

Revision 1.02, 16 October 2020

Setup

Set up conda

1. Create conda environment

conda env create -f env.yml

or

conda create -n shanmuga_iitg -f env.yml

1. Setup pytorch and pytorch-geometric for the environment

Install pytorch 1.5.0 make sure to install the CPU only version in a PC without CUDA-device (GPUs)

conda install -c pytorch pytorch=1.5.0 torchvision=0.6.0 cpuonly

build wheels for this version and install essentials for CPU or GPU versions ref

pip install torch-scatter==latest+cpu -f https://pytorch-geometric.com/whl/torch-1.5.0.html
pip install torch-sparse==latest+cpu -f https://pytorch-geometric.com/whl/torch-1.5.0.html
pip install torch-cluster==latest+cpu -f https://pytorch-geometric.com/whl/torch-1.5.0.html
pip install torch-spline-conv==latest+cpu -f https://pytorch-geometric.com/whl/torch-1.5.0.html
pip install torch-geometric==1.5.0

Replace cpu with cu92, cu101 or cu102 for other CUDA-devices GPUs.

If you are using windows you might need additional .dll files to be able to work with pytorch ref

To resolve the issue download additional packages and copy them to C:\Windows\System32, this should resolve all the issues and allow you to use PyTorch.

Environment ready.

1. Activate the environment

conda activate shanmuga_iitg

Development environment

Pycharm

Jupyter Notebook

Spyder

Setup to use the repo in google-colab

to be added

Setup to use the repo in an EC2 instance

to be added