RuntimeError: mat1 and mat2 shapes cannot be multiplied

[bii-dpi]

Running the AttentionSiteDTI-based example on the Human dataset as found within README.md produces the following error in the "testing" phase after data processing: RuntimeError: mat1 and mat2 shapes cannot be multiplied (29x405 and 666x50)

The full printed output is as follows; note the Desired structure doesn't exist warnings which might be related to the issue:

Processing: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 138/138 [00:03<00:00, 42.52it/s]
Preprocessing SMILES from smile to graph is done.
Processing:   5%|██████                                                                                                                | 5/98 [00:08<01:54,  1.23s/it](_preprocess_and_save_data_point pid=123783) Desired structure doesn't exist
Processing:  15%|█████████████████▉                                                                                                   | 15/98 [00:14<00:49,  1.68it/s](_preprocess_and_save_data_point pid=123783) Desired structure doesn't exist
Processing:  84%|█████████████████████████████████████████████████████████████████████████████████████████████████▉                   | 82/98 [00:48<00:08,  1.94it/s](_preprocess_and_save_data_point pid=123785) Desired structure doesn't exist
Processing: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 98/98 [01:00<00:00,  1.61it/s]
Preprocessing pdb_id from pdb_id to binding_pocket_graph is done.
Preprocessing Label done.
Processing: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 22/22 [00:03<00:00,  7.06it/s]
Preprocessing SMILES from smile to graph is done.
Processing: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 6/6 [00:06<00:00,  1.04s/it]
Preprocessing pdb_id from pdb_id to binding_pocket_graph is done.
Preprocessing Label done.
Processing: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 24/24 [00:03<00:00,  7.66it/s]
Preprocessing SMILES from smile to graph is done.
Processing: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 13/13 [00:09<00:00,  1.37it/s]
Preprocessing pdb_id from pdb_id to binding_pocket_graph is done.
Preprocessing Label done.
Testing:   0%|                                                                                                                                  | 0/2 [00:01<?, ?it/s]
Traceback (most recent call last):
  File "path/to/your/project/tmp.py", line 49, in <module>
    print(model.evaluate(data_loader_val, device,
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "path/to/your/project/attentionsitedti.py", line 448, in evaluate
    out = self.forward(d, p)
          ^^^^^^^^^^^^^^^^^^
  File "path/to/your/project/attentionsitedti.py", line 333, in forward
    ligand_rep = self.drug_encoder(drug)
                 ^^^^^^^^^^^^^^^^^^^^^^^
  File "path/to/env/lib/python3.11/site-packages/torch/nn/modules/module.py", line 1553, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "path/to/env/lib/python3.11/site-packages/torch/nn/modules/module.py", line 1562, in _call_impl
    return forward_call(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "path/to/your/project/graph_conv.py", line 63, in forward
    graph = layer(graph)
            ^^^^^^^^^^^^
  File "path/to/env/lib/python3.11/site-packages/torch/nn/modules/module.py", line 1553, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "path/to/env/lib/python3.11/site-packages/torch/nn/modules/module.py", line 1562, in _call_impl
    return forward_call(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "path/to/env/lib/python3.11/site-packages/dgl/nn/pytorch/conv/tagconv.py", line 145, in forward
    rst = self.lin(th.cat(fstack, dim=-1))
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "path/to/env/lib/python3.11/site-packages/torch/nn/modules/module.py", line 1553, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "path/to/env/lib/python3.11/site-packages/torch/nn/modules/module.py", line 1562, in _call_impl
    return forward_call(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "path/to/env/lib/python3.11/site-packages/torch/nn/modules/linear.py", line 117, in forward
    return F.linear(input, self.weight, self.bias)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
RuntimeError: mat1 and mat2 shapes cannot be multiplied (29x405 and 666x50)

(tmp.py refers to the script I created directly based on the example from README.md)

Python Version: 3.11
Relevant Packages:
dgl: 2.4.0
torch: 2.4.0