Bug on line 132 in the 'tudataset.py' file

[StevanStanovic]

🐛 Describe the bug

Hello,

There is a bug on line 132 in the 'tudataset.py' file for the 2.1.0 version of Pytorch Geometric.
The message error is:
ValueError: not enough values to unpack (expected 3, got 2)

I fixed it by removing the 'self.sizes' part and replacing the 'num node_labels', 'num_node attributes', 'num_edge_labels' and 'num_edge attributes' methods with those from version 2.0.4.

I have attached a screenshot (PNG image) of my solution. I hope this will be useful.

Stevan Stanovic

Environment

• PyG version: 2.1.0
• PyTorch version: 1.12.1
• OS: Linux
• Python version: 3.9.12
• CUDA/cuDNN version: 11.6
• How you installed PyTorch and PyG (conda, pip, source):
  conda create -n test python=3.9.12
  conda activate test
  pip3 install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu116
  pip install torch-scatter torch-sparse torch-cluster torch-spline-conv torch-geometric -f https://data.pyg.org/whl/torch-1.12.0+cu116.html
• Any other relevant information (e.g., version of torch-scatter):
  torch 1.12.1+cu116
  torch-cluster 1.6.0
  torch-geometric 2.1.0.post1
  torch-scatter 2.0.9
  torch-sparse 0.6.15
  torch-spline-conv 1.2.1
  torchaudio 0.12.1+cu116
  torchvision 0.13.1+cu116

[rusty1s]

Mh, if len(out) != 3 then it should raise a RuntimeError in L127. Can you verify why this error is not thrown for you? Furthermore, try to remove any old processed data from PyG 2.0.4 and try again. I am pretty sure TUDataset works just fine with PyG 2.1.0 (at least it works flawlessly for me).

[StevanStanovic]

The 'If' condition is not correct. I think the 'if' condition below is the solution for the 2.1.0 version. 

if not(isinstance(out, tuple) and len(out) == 3)

If the two conditions are True, the RumtimeError appears, and, inversely, if one of the two conditions is False.

Thank you for the quick response.

[rusty1s]

Oh, you are right. Is has to be if not(isinstance(out, tuple) or len(out) != 3).