RuntimeError while run full_batch.py

[EvanYu800112]

Hi,
I run python full_batch.py on ogbn_arxiv dataset, while a runtime error occurred. I have tried searching for a solution, but I still don't know why it happened....
Has anyone run into the same problem can give some help? Thank you in advance.

RuntimeError:

init(torch.torch_sparse.storage.SparseStorage self, Tensor? row, Tensor? rowptr, Tensor? col, Tensor? value, (int, int)? sparse_sizes, Tensor? rowcount, Tensor? colptr, Tensor? colcount, Tensor? csr2csc, Tensor? csc2csr, bool is_sorted) -> (None):
Expected a value of type 'Optional[Tensor]' for argument 'row' but instead found type 'int'.
:
File "/home/Evan/PyEnv_3.7/lib/python3.7/site-packages/torch_sparse/storage.py", line 283
col = idx % num_cols

    return SparseStorage(row=row, rowptr=None, col=col, value=self._value,
           ~~~~~~~~~~~~~ <--- HERE
                         sparse_sizes=(num_rows, num_cols), rowcount=None,
                         colptr=None, colcount=None, csr2csc=None,

[rusty1s]

Related to pyg-team/pytorch_geometric#1125 (comment) caused by the division op changes introduced in PyTorch 1.5. Please modify storage.py accordingly or update to PyTorch 1.5. Sorry for the inconveniences!

[EvanYu800112]

I modify storage.py and it works now!!!
Thank you for your help :)