TGB is a collection of challenging and diverse benchmark datasets
for realistic, reproducible, and robust evaluation for machine learning on temporal graphs.
TGB includes both dynamic link and node property prediction tasks and an automated pipeline
from dataset downloading, dataloading, evaluation and submission to the TGB leaderboard.
TGB is driven by community feedback and suggestions, if you would like to contribute a novel dataset or report any issues,
please email us or visit our github.
TGB provides diverse and realistic datasets, containing millions of nodes, edges and timestamps
Multiple Data Formats
TGB datasets are supported as numpy arrays, PyTorch tensors and PyG compatible TemporalData objects
TGB provides dataset splits and evaluators for reproducible and standardized evaluation for temporal graphs