Closed Xelawk closed 1 year ago
The size of adjacency matrices are varient in traning phase, how did you address this problem, or any strategy?
GNN model natively can handle graphs with different size, if you are willing to batch small graphs to a large one, please refer to https://docs.dgl.ai/generated/dgl.batch.html?highlight=batch#dgl.batch
The size of adjacency matrices are varient in traning phase, how did you address this problem, or any strategy?