Weights dimensions mismatch when loading a saved .pth: d_con1_lin_src.weight

[usmancheema89]

Steps to reproduce:

• train model with default configuration one the given dataset (train.py)
• load saved model for testing (test.py)

Exception trace:

cd /home/usman/Work/GlobalMapper ; /usr/bin/env /home/usman/anaconda3/envs/SD201/bin/python /home/usman/.vscode-server/extensions/ms-python.debugpy-2024.0.0-linux-x64/bundled/libs/debugpy/adapter/../../debugpy/launcher 47893 -- /home/usman/Work/GlobalMapper/test.py
/home/usman/Work/GlobalMapper/model.py:129: UserWarning: nn.init.xavier_uniform is now deprecated in favor of nn.init.xavier_uniform_.
m.weight.data = nn.init.xavier_uniform(m.weight.data, gain=nn.init.calculate_gain('relu'))
/home/usman/Work/GlobalMapper/model.py:394: UserWarning: nn.init.xavier_uniform is now deprecated in favor of nn.init.xavier_uniform_.
m.weight.data = nn.init.xavier_uniform(m.weight.data, gain=nn.init.calculate_gain('relu'))
Traceback (most recent call last):
File "/home/usman/anaconda3/envs/SD201/lib/python3.10/runpy.py", line 196, in _run_module_as_main
return _run_code(code, main_globals, None,
File "/home/usman/anaconda3/envs/SD201/lib/python3.10/runpy.py", line 86, in _run_code
exec(code, run_globals)
File "/home/usman/.vscode-server/extensions/ms-python.debugpy-2024.0.0-linux-x64/bundled/libs/debugpy/adapter/../../debugpy/launcher/../../debugpy/main.py", line 39, in
cli.main()
File "/home/usman/.vscode-server/extensions/ms-python.debugpy-2024.0.0-linux-x64/bundled/libs/debugpy/adapter/../../debugpy/launcher/../../debugpy/../debugpy/server/cli.py", line 430, in main
run()
File "/home/usman/.vscode-server/extensions/ms-python.debugpy-2024.0.0-linux-x64/bundled/libs/debugpy/adapter/../../debugpy/launcher/../../debugpy/../debugpy/server/cli.py", line 284, in run_file
runpy.run_path(target, run_name="main")
File "/home/usman/.vscode-server/extensions/ms-python.debugpy-2024.0.0-linux-x64/bundled/libs/debugpy/_vendored/pydevd/_pydevd_bundle/pydevd_runpy.py", line 321, in run_path
return _run_module_code(code, init_globals, run_name,
File "/home/usman/.vscode-server/extensions/ms-python.debugpy-2024.0.0-linux-x64/bundled/libs/debugpy/_vendored/pydevd/_pydevd_bundle/pydevd_runpy.py", line 135, in _run_module_code
_run_code(code, mod_globals, init_globals,
File "/home/usman/.vscode-server/extensions/ms-python.debugpy-2024.0.0-linux-x64/bundled/libs/debugpy/_vendored/pydevd/_pydevd_bundle/pydevd_runpy.py", line 124, in _run_code
exec(code, run_globals)
File "/home/usman/Work/GlobalMapper/test.py", line 132, in
model.load_state_dict(torch.load(os.path.join(test_yaml, pth_name + ".pth"), map_location=device), strict=False)
File "/home/usman/anaconda3/envs/SD201/lib/python3.10/site-packages/torch/nn/modules/module.py", line 2041, in load_state_dict
raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(
RuntimeError: Error(s) in loading state_dict for AttentionBlockGenerator_independent_cnn:
While copying the parameter named "d_conv1.lin_src.weight", whose dimensions in the model are torch.Size([0]) and whose dimensions in the checkpoint are torch.Size([3072, 376]), an exception occurred : ('The size of tensor a (0) must match the size of tensor b (376) at non-singleton dimension 1',).

[Arking1995]

I suppose that is mostly the pytorch geometric version issue. If you do use the newer version than readme says, update the definition codes for graph attention layer will be helpful. Thanks for the feedbacks!