Open Gingersna opened 5 years ago
In these datasets, node in the train set may be related to the node in test/val set, so I want to know how to sample to avoid data leakage
As far as I know these train/val/test splits should be disjunct. Why do you think that there is an overlap?
In these datasets, node in the train set may be related to the node in test/val set, so I want to know how to sample to avoid data leakage