Output shape of mini batch for graph neural network

I have a dataset containing 100 graph data. Each has different number of nodes, but 1024 features. And each graph has a target 0 or 1.

Now while using DataLoader to load the data using a batch size of 32, I am getting x shape of (something, 1024). I understand the value of "something " depends on which 32 graphs are being concatenated in the batch matrix. But the output shape (y) remains (32,2).

Now while training I am getting error that, expected input batch size (7557) to batch target batch size (32). How can I solve this issue?

Ik this is a very basic qus. But I am having some issues to deal with it.