Building our own dataset on TF backend


I have my own dataset that is in the same format with PPI dataset, where there are 3 .npy files and 1 dictionary. I want to wrap this data using your dataset object so I checked the tutorial here and also digging into the github code here. I want to later train GNN using TF2 backend. I saw in the data creation examples, you use pytorch backend. How can I build the dataset object using TF2 backend? Thanks


Our datasets are backend-agnostic by numpy. Did you meet any problem following the code you posted?

I guess I have to build Tensorflow tensor instead of PyTorch tensor (torch.from_numpy) in that case? I’ll give it a try. Thanks

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.