ImbalancedSampler Error

[1AngelUS]

Hi,

I am using ImbalancedSampler for hetero data:
sampler = ImbalancedSampler(data['p'].y.long())
train_loader = NeighborLoader(data, batch_size = 500, num_neighbors=[20,10]*2,input_nodes=train_input_nodes,
sampler=sampler)

This is throwing the following error:
RuntimeError: number of categories cannot exceed 2^24

[rusty1s]

How big is your y tensor? PyTorch currently has a limitation of max 2^24 samples :(