The output size are different between prediction and true

[Byun-jinyoung]

Dear PyG users,
Because I recently have known pytorch_geometric, I tried to apply it.

I have inputs like this:

Data(x=[188, 15], y=[188], pos=[188, 3], edge_index=[2, 4518], edge_attr=[4518, 7])
Data(x=[188, 15], y=[188], pos=[188, 3], edge_index=[2, 4518], edge_attr=[4518, 7])
Data(x=[169, 15], y=[169], pos=[169, 3], edge_index=[2, 1744], edge_attr=[1744, 7])
Data(x=[169, 15], y=[169], pos=[169, 3], edge_index=[2, 1744], edge_attr=[1744, 7])
Data(x=[95, 15], y=[95], pos=[95, 3], edge_index=[2, 1114], edge_attr=[1114, 7])

When I performed the node classification task, I encountered the error as below:

  File "/home/byun/PROject/proteinGAE/utils/pyg_utils.py", line 34, in train
    loss = criterion(pred_y, true_y)
  File "/home/byun/anaconda3/envs/pyg_torch/lib/python3.7/site-packages/torch/nn/modules/module.py", line 1102, in _call_impl
    return forward_call(*input, **kwargs)
  File "/home/byun/anaconda3/envs/pyg_torch/lib/python3.7/site-packages/torch/nn/modules/loss.py", line 1152, in forward
    label_smoothing=self.label_smoothing)
  File "/home/byun/anaconda3/envs/pyg_torch/lib/python3.7/site-packages/torch/nn/functional.py", line 2846, in cross_entropy
    return torch._C._nn.cross_entropy_loss(input, target, weight, _Reduction.get_enum(reduction), ignore_index, label_smoothing)
ValueError: Expected input batch_size (32076) to match target batch_size (32078)

Thus, I printed the output shape of both prediction and true value for each batch (left: prediction, right: true)

torch.Size([33838, 21]), torch.Size([33838]) 
BATCH IDX: 0    Train Loss: 25.1011     Train Acc: 4.5895
torch.Size([32028, 21]), torch.Size([32028])
torch.Size([32076, 21]), torch.Size([32078])

As you can see, the last line showed the different shapes of each tensor.

Could you tell me what I'm missing?

Thank you :)

[rusty1s]

There might exist some example in your graph where data.x.size(0) != data.y.size(0). Can you check?