torch geometric library issue

[nyongja]

When running ml-100k dataset code, the torch_geometric library has an error.
It happens because the function is not implemented in torch_geometric library data/dataset.py.
How should I implement this part?

[muhanzhang]

Hi! Can you paste more details about the error message? It could be related to a pytorch_geometric version issue.

[nyongja]

The error message is as follows.
Additionally, my torch_geometric version is 2.0.1.
However, in the torch_geometric 1.4.2 library document, the function is not implemented within data/dataset.py.
(https://pytorch-geometric.readthedocs.io/en/1.4.2/_modules/torch_geometric/data/dataset.html#Dataset)
Thank you.

Traceback (most recent call last):
File "Main.py", line 389, in
model = IGMC(
File "/home/nyongja/study/IGMC/models.py", line 177, in init
super(IGMC, self).init(
File "/home/nyongja/study/IGMC/models.py", line 22, in init
self.convs.append(gconv(dataset.num_features, latent_dim[0]))
File "/home/nyongja/anaconda3/envs/igmc/lib/python3.8/site-packages/torch_geometric/data/dataset.py", line 115, in num_features
return self.num_node_features
File "/home/nyongja/anaconda3/envs/igmc/lib/python3.8/site-packages/torch_geometric/data/dataset.py", line 104, in num_node_features
data = self[0]
File "/home/nyongja/anaconda3/envs/igmc/lib/python3.8/site-packages/torch_geometric/data/dataset.py", line 198, in getitem
data = self.get(self.indices()[idx])
File "/home/nyongja/anaconda3/envs/igmc/lib/python3.8/site-packages/torch_geometric/data/dataset.py", line 91, in indices
return range(self.len()) if self._indices is None else self._indices
File "/home/nyongja/anaconda3/envs/igmc/lib/python3.8/site-packages/torch_geometric/data/dataset.py", line 63, in len
raise NotImplementedError
NotImplementedError

[muhanzhang]

Hi, since the newer versions of PyG have so many incompatibilities with the original IGMC code, I created a latest branch to maintain a runnable IGMC with the latest PyG environment. Please have a try. However, the latest version still cannot catch up with the old version's performance.

[nyongja]

The latest branch's code is executed! Thank you. I enjoyed reading your paper:)