AttributeError when trying to call torch_geometric.datasets.QM9

[JEAM-professional]

I am attempting to use torch_geometric.datasets.QM9 to subsequently fine-tune a pre-trained model. The simple code:

from torch_geometric.datasets import QM9

path = './dsgdb9nsd.xyz'
dataset = QM9(path)

produces the following error for me:

Processing...
  4%|████▋                                                                                                                              | 4804/133885 [00:01<00:29, 4370.57it/s]
Traceback (most recent call last):
  File "/Test.py", line 6, in <module>
    dataset = QM9(path)
  File "/.../.venv/lib/python3.14/site-packages/torch_geometric/datasets/qm9.py", line 154, in __init__
    super().__init__(root, transform, pre_transform, pre_filter,
    ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                     force_reload=force_reload)
                     ^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/.../.venv/lib/python3.14/site-packages/torch_geometric/data/in_memory_dataset.py", line 81, in __init__
    super().__init__(root, transform, pre_transform, pre_filter, log,
    ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                     force_reload)
                     ^^^^^^^^^^^^^
  File "/.../.venv/lib/python3.14/site-packages/torch_geometric/data/dataset.py", line 116, in __init__
    self._process()
    ~~~~~~~~~~~~~^^
  File "/.../.venv/lib/python3.14/site-packages/torch_geometric/data/dataset.py", line 266, in _process
    self.process()
    ~~~~~~~~~~~~^^
  File "/.../.venv/lib/python3.14/site-packages/torch_geometric/datasets/qm9.py", line 249, in process
    N = mol.GetNumAtoms()
        ^^^^^^^^^^^^^^^
AttributeError: 'NoneType' object has no attribute 'GetNumAtoms'

I have tried both manually downloading the dataset and letting the QM9 function downloading them with the exact same error. Any help here would be appreciated.

~ Juanes

[JEAM-professional]

Oh, just solved my own problem but still good for you to know. It worked with python3.12 (instead of the pyhon3.14 I was using in the previous attempt.)