Closed buaalyx closed 1 year ago
I found another bug issue with same error message DGLError Expected data to have %d rows, got %d. occurs at large batch size, so I guess I meet another bug?
The same question is being asked in https://discuss.dgl.ai/t/question-of-running-apply-edges-in-a-contrastive-model-a-potential-bug/3294 so we will discuss there. I'll reopen this issue if a bug is confirmed.
I want to implement a contrastive graph model. I have an encoder for two heterogeneous graphs
G_raw
andG_aug
, whereG_aug
is obtained by removing some edges fromG_raw
. And the specific information are as follows:The encoder works well on
G_raw
with the forward processencoder(G_raw)
but I met the following error when running the forward process on
G_aug
withencoder(G_aug)
:The corresponding code snippet is as follows:
I print relevant data before
sub_graph.apply_edges()
and gotI just don't know why the
sub_graph
with 11627 edges 'Expected data to have 13407 rows' toapply_edges()
I noticed this question: Cannot assign edge data after
g.remove_edges
, but I still meet the similar problems. My dgl version is