On Nueral biaprtite matching about neural-bipartite-matching HOT 3 CLOSED

Dear balexx2,

This project is about 4 years old. If you're using the latest versions of NN libraries, such as PyG/PyTorch, quite likely there's an API mismatch. (Which I believe is the case here).

For example, if you take a look at flow_datasets.py, there are __len__ functions defined. You can use these to define len, function in the class. E.g.

    def len(self):
        return self.__len__()

Furthermore, you're not supposed to run flow_datasets.py. It's a file that's imported by other scripts. Did you take a look at the README provided and follow the steps, before opening the issue?

Lastly, this is not a webpage/charitable/profitable project so there is no need to donate and support it. Also, as this is a research, not an industry project, I have no plans to update it to use the latest libraries.

from neural-bipartite-matching.

from neural-bipartite-matching.

Hello Prof, I embarked on understanding Nueral bipartite matching as clearly explained in your paper. i have tested 98% of your successfully. i have only met the error: IndexError: index 6 is out of bounds for dimension 0 with size 6 and it points to utils.py script especially at this function:
def create_inv_edge_index(batch_size, size, edge_index):
iei = torch.full((batch_sizesize, batch_sizesize), -100, dtype=torch.long)
for i in range(len(edge_index[0])):
iei[edge_index[0][i]][edge_index[1][i]] = i
return iei
tedge_index = torch.tensor(edge_index).clone().detach()
tensor([[ 5., 1.],
[ 1., 1.],
[14., 1.],
[ 7., 1.],
[ 1., 1.],
[ 6., 1.],
[12., 1.],
[14., 1.],
[15., 1.],
[ 5., 1.],
[10., 1.],
[ 2., 0.],
[ 9., 0.],
[14., 0.],
[15., 0.],
[13., 0.],
[ 5., 0.],
[ 3., 0.],
[10., 0.],
[ 9., 0.],
[15., 0.],
[14., 0.]], grad_fn=)
ROOT ./bfs_less_wired
PROCESSED bfs_less_wired/processed
ROOT ./bfs_less_wired
PROCESSED bfs_less_wired/processed
ROOT ./all_iter_less_wired
PRP []
PROCESSED all_iter_less_wired/processed
ROOT ./all_iter_less_wired
PRP []
PROCESSED all_iter_less_wired/processed
Processing...
Done!
Processing...
Done!
Processing...
Done!
Processing...
Done!
Traceback (most recent call last):

File "/tmp/ipykernel_5044/3463109613.py", line 73, in
inv_edge_index = utils.create_inv_edge_index(len(GRAPH_SIZES), GRAPH_SIZES.max(), batch.edge_index)

File "/home/nahondo/Desktop/nsm/utils.py", line 221, in create_inv_edge_index
iei[edge_index[0][i]][edge_index[1][i]] = i

IndexError: index 6 is out of bounds for dimension 0 with size 6

from neural-bipartite-matching.