Generate Metapaths

[DuanhaoranCC]

from torch_geometric.datasets import AMiner
from torch_geometric.transforms import AddMetaPaths
data = AMiner(root='./data/aminer_pyg/')[0]
metapaths = [[("author", "paper"), ("paper", "author")],
[("author", "paper"), ("paper", "venue"), ("venue", "paper"), ("paper", "author")]]
data = AddMetaPaths(metapaths)(data)

Hi dear author, I am using the Aminer dataset to create meta paths, but I am getting the following error:

  File "/home/yhkj/anaconda3/envs/kg/lib/python3.8/site-packages/torch_geometric/transforms/add_metapaths.py", line 118, in __call__
    adj1 = adj1 @ adj2
  File "/home/yhkj/anaconda3/envs/kg/lib/python3.8/site-packages/torch_sparse/matmul.py", line 149, in <lambda>
    SparseTensor.__matmul__ = lambda self, other: matmul(self, other, 'sum')
  File "/home/yhkj/anaconda3/envs/kg/lib/python3.8/site-packages/torch_sparse/matmul.py", line 140, in matmul
    return spspmm(src, other, reduce)
  File "/home/yhkj/anaconda3/envs/kg/lib/python3.8/site-packages/torch_sparse/matmul.py", line 117, in spspmm
    return spspmm_sum(src, other)
  File "/home/yhkj/anaconda3/envs/kg/lib/python3.8/site-packages/torch_sparse/matmul.py", line 102, in spspmm_sum
    rowptrC, colC, valueC = torch.ops.torch_sparse.spspmm_sum(
MemoryError: std::bad_alloc

Did I miss something?
Thank you for your patience in answering!

[EdisonLeeeee]

std::bad_alloc means a problem with allocating a memory, I think you're most likely out of memory.

[DuanhaoranCC]

The machine memory is already 120GB. Is this caused by too many nodes?

[EdisonLeeeee]

Yes, AddMetaPaths seems to be less scalable for large graphs. I also encountered this issue in my machine with 128G memory.

[wsad1]

Can you check if the max_sample in AddMetaPaths helps.