Installation

System requrements

GammaGL works with the following operating systems:

  • Linux

GammaGL requires Python version 3.9, 3.10, 3.11, 3.12.

Backend

  • tensorflow : We recommend tensorflow version under 2.12.0

  • pytorch : Support version from 2.1 to 2.3, the defalut backend

  • paddlepaddle : We recommend paddlepaddle version under 2.3.2

  • mindspore : Support version to 2.2.10

Quick Start with PyTorch

PyTorch
Your OS
CUDA
Run:

      

   

   
   

If you choose the other backend, you can directly install gammagl with pip install gammagl.






Install from pip


1. Python environment (Optional): We recommend using conda package manager


conda create -n gammagl python=3.10
source activate gammagl





2. Backend: Select and Install your favorite deep learning backend. For example:


# tensorflow
pip install tensorflow-gpu # GPU version
pip install tensorflow # CPU version
# pytorch
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
# paddlepaddle
python -m pip install paddlepaddle-gpu
# mindspore
pip install https://ms-release.obs.cn-north-4.myhuaweicloud.com/2.2.0/MindSpore/unified/x86_64/mindspore-2.2.0-cp39-cp39-linux_x86_64.whl --trusted-host ms-release.obs.cn-north-4.myhuaweicloud.com -i https://pypi.tuna.tsinghua.edu.cn/simple







Note


    

  • For tensorflow, we recommend you to use version under 2.11.0 (For Windows users, please install version 2.10.1 as 2.11 is not supported on Windows).
    • 

  • For pytorch, we support the latest version, e.g. pytorch 2.1.0+cu118.
    • 

  • For paddlepaddle, we recommend you to use version under 2.3.2.
    • 

  • For mindspore, we support the latest version, e.g. mindspore 2.2.0+cu116.
    • 





3. GammaGL: Install GammaGL and its dependencies.


pip install gammagl-pt23==0.5.0









Install from source


1. Python environment (Optional): We recommend using conda package manager


conda create -n gammagl python=3.10
source activate gammagl





2. Backend: Select and Install your favorite deep learning backend. For example:


# tensorflow
pip install tensorflow-gpu # GPU version
pip install tensorflow # CPU version
# pytorch
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
# paddlepaddle
python -m pip install paddlepaddle-gpu
# mindspore
pip install https://ms-release.obs.cn-north-4.myhuaweicloud.com/2.2.0/MindSpore/unified/x86_64/mindspore-2.2.0-cp39-cp39-linux_x86_64.whl --trusted-host ms-release.obs.cn-north-4.myhuaweicloud.com -i https://pypi.tuna.tsinghua.edu.cn/simple







Note


    

  • For tensorflow, we recommend you to use version under 2.11.0 (For Windows users, please install version 2.10.1 as 2.11 is not supported on Windows).
    • 

  • For pytorch, we support the latest version, e.g. pytorch 2.3.0+cu118.
    • 

  • For paddlepaddle, we recommend you to use version under 2.3.2.
    • 

  • For mindspore, we support the latest version, e.g. mindspore 2.2.0+cu116.
    • 





3. TensorLayerX: Install TensorLayerX. For example:


pip install git+https://github.com/dddg617/tensorlayerx.git@nightly





4. GammaGL: Install GammaGL and its dependencies.


pip install pybind11 pyparsing
git clone --recursive https://github.com/BUPT-GAMMA/GammaGL.git
cd GammaGL
python setup.py install build_ext --inplace







Note


    

  • pybind11 and pyparsing is required, otherwise, you cannot install GammaGL properly.
    • 

  • If you want to setup with cuda, please set WITH_CUDA to True in setup.py.
    • 

  • If you want to develop GammaGL locally, you may use the following command to build package:
    • 





python setup.py build_ext --inplace









How to Run


Take GCN as an example:


# cd ./examples/gcn
# set parameters if necessary
python gcn_trainer.py --dataset cora --lr 0.01





If you want to use specific backend or GPU, just set environment variable like:


CUDA_VISIBLE_DEVICES="1" TL_BACKEND="paddle" python gcn_trainer.py







Note


The DEFAULT backend is tensorflow and GPU is 0. The backend TensorFlow will take up all GPU left memory by default.
The CANDIDATE backends are tensorflow, paddle, torch and mindspore.
Set CUDA_VISIBLE_DEVICES=" " if you want to run it in CPU.