PyG: RuntimeError: тензоры должны иметь одинаковое количество измерений: получили 2 и 3
Я используюTransformerConvи столкнулся с этой ошибкой:
Traceback (most recent call last):
File "pipeline_model_gat.py", line 1018, in <module>
output = model(
File"/mount/arbeitsdaten61/studenten3/advanced-ml/2022/gogirlspower/nicole/conda/envs/new_gvqa/lib/python3.8/sitepackages/torch/nn/modules/module.py", line 727, in _call_impl
result = self.forward(*input, **kwargs)
File "pipeline_model_gat.py", line 881, in forwardquestions_encoded = self.question_encoder(question_graphs)
File "/mount/arbeitsdaten61/studenten3/advanced-ml/2022/gogirlspower/nicole/conda/envs/new_gvqa/lib/python3.8/site-packages/torch/nn/modules/module.py", line 727, in _call_impl
result = self.forward(*input, **kwargs)
File "pipeline_model_gat.py", line 628, in forward= self.conv1(x, question_graphs.edge_index, edge_attr)
File "/mount/arbeitsdaten61/studenten3/advanced-ml/2022/gogirlspower/nicole/conda/envs/new_gvqa/lib/python3.8/site-packages/torch/nn/modules/module.py", line 727, in _call_impl
result = self.forward(*input, **kwargs)
File "/mount/arbeitsdaten61/studenten3/advanced-ml/2022/gogirlspower/nicole/conda/envs/new_gvqa/lib/python3.8/site-packages/torch_geometric/nn/conv/transformer_conv.py", line 190, in forward
beta = self.lin_beta(torch.cat([out, x_r, out - x_r], dim=-1))
RuntimeError: Tensors must have same number of dimensions: got 2 and 3
Может кто-нибудь, пожалуйста, скажите мне, что могло пойти не так?