Nick1899
Reputation: 13
How to make PyTorch Geometric DataLoader to create multiple graphs in batch instead of one giant graph?
PyTorch Geometric DataLoader, while using batches with sizes greater than 1, creates a large graph with isolated subgraphs, instead of a list of graphs with padding to equal size. Unfortunately, this way is inappropriate for my task. So is there a way to make PyTorch Geometric DataLoader create a list of graphs instead of one large graph?
Upvotes: 0
Views: 91
Answers (0)
Related Questions
- Pytorch Geometric graph batching not using DataLoader for Reinforcement learning
- Why batch training with maximum batchsize using PyTorch dataloader exhibits worse performance than inputing entire datasets to the networks?
- Pytorch geometric: Having issues with tensor sizes
- How to catch multiple exceptions in one line? (in the "except" block)
- Pytorch Device-Side Assert on Backward()
- PyTorch Geometric - How to sample huge graph to train GNN with mini-batching
- How can one use the RemoveIsolatedNodes transform in Pytorch Geometric?
- How to batch a nested list of graphs in pytorch geometric