An Error happened when I using the pyg=2.3.0

[StefanIsSmart]

I will update later

[rusty1s]

How can I run your example?

[StefanIsSmart]

Thank you again!
I will rewrite the debug report again, and give you a mini processed data sample. Do you think it's okay?

[StefanIsSmart]

Hello, when I want to reproduce the error and give you a mini-sample,
I find the result is the processed data is from the PyG2.2.0,
If I regenerate the processed data, the error is gone!

[StefanIsSmart]

Hi @rusty1s

I find that:

The processed data from PyG2.5.0 and run with PyG2.2.0 will happened:

  File "/export/disk3/why/software/Miniforge3/envs/PyG250/lib/python3.8/site-packages/torch_geometric/data/in_memory_dataset.py", line 319, in __getattr__
    raise AttributeError(f"'{self.__class__.__name__}' object has no "
AttributeError: 'MyDataset_batch_ShortQM9' object has no attribute '_data'

The processed data from PyG2.2.0 and run with PyG2.5.0 / PyG2.3.0 will happened:

  File "/export/disk3/why/software/Miniforge3/envs/PyG220/lib/python3.7/site-packages/torch_geometric/data/in_memory_dataset.py", line 83, in get
    cls=self.data.__class__,
AttributeError: 'MyDataset_batch_ShortQM9' object has no attribute 'data'

[rusty1s]

I cannot reproduce this on any dataset in PyG (e.g, tested with TUDataset), so this seems to be a special issue. Still curious how I can run your example to help you :(

[StefanIsSmart]

How can I run your example?

How can I run your example?

How can I run your example?