How about customizing the graph dataset to iteratively fetch batch
In my customized dataset there are only feature matrices and adjacency matrices.The feature matrix is a 4000×3000 one-dimensional tensor,and edge_index is torch.Size([2, 18708])
My code is :