All the datasets work with DGL 0.5 or later. Please update the environment using the yml files in the root directory if the use of these datasets throw error(s).
Nothing to do. The TU datasets are automatically downloaded.
MNIST size is 1.39GB and CIFAR10 size is 2.51GB.
# At the root of the project
cd data/
bash script_download_superpixels.sh
Script script_download_superpixels.sh is located here. Codes to reproduce the datasets for MNIST and for CIFAR10.
ZINC size is 58.9MB.
ZINC-full size is 1.17GB.
# At the root of the project
cd data/
bash script_download_molecules.sh
Script script_download_molecules.sh is located here. Code to reproduce the ZINC dataset is here and the ZINC-full dataset is here.
PATTERN size is 1.98GB and CLUSTER size is 1.26GB.
# At the root of the project
cd data/
bash script_download_SBMs.sh
Script script_download_SBMs.sh is located here. Codes to reproduce the datasets for PATTERN and for CLUSTER.
TSP size is 1.87GB.
# At the root of the project
cd data/
bash script_download_TSP.sh
Script script_download_TSP.sh is located here. Codes to reproduce the TSP dataset is here.
CSL size is 27KB.
# At the root of the project
cd data/
bash script_download_CSL.sh
Script script_download_CSL.sh is located here.
COLLAB size is 360MB.
No script to run. The COLLAB dataset files will be automatically downloaded from OGB when running the experiment files for COLLAB.
GraphTheoryProp size is 1.2GB.
# At the root of the project
cd data/
bash script_download_graphtheoryprop.sh
Script script_download_graphtheoryprop.sh is located here.
CYCLES size is 81MB.
# At the root of the project
cd data/
bash script_download_cycles.sh
Script script_download_cycles.sh is located here.
# At the root of the project
cd data/
bash script_download_all_datasets.sh
Script script_download_all_datasets.sh is located here.