'How to change the node tensor type on an entire PyTorch Geometric Dataset?
Is there a function to change the tensor type (from long to float in my case) on the node features of every graph in a pytorch geometric dataset? Iterating over the dataset with
for graph in dataset:
graph.x = graph.x.float()
does not work. It is possible to transform the dataset to a list and then use the approach above, but all dataset methods (e.g. shuffle) are lost in the process.
Is there a method to quickly transform the tensor types of every graph?
Sources
This article follows the attribution requirements of Stack Overflow and is licensed under CC BY-SA 3.0.
Source: Stack Overflow
| Solution | Source |
|---|