lwm412 / stfgnn-pytorch Goto Github PK

What version of Pytorch you were using for this code? I use 1.12.1+cu113, but there was an error:
Traceback (most recent call last):
File "main.py", line 33, in
executor.train(train_data, valid_data)
File "/work/users/s/h/shuaishu/Transformer/STFGNN-Pytorch/STFGNN/executor/multi_step_executor.py", line 82, in train
output = self.model(trainx)
File "/work/users/s/h/shuaishu/.conda/STFGNNpy37pytorch/lib/python3.7/site-packages/torch/nn/modules/module.py", line 1130, in _call_impl
return forward_call(*input, **kwargs)
File "/work/users/s/h/shuaishu/Transformer/STFGNN-Pytorch/STFGNN/model/STFGNN.py", line 416, in forward
x = model(x, self.mask)
File "/work/users/s/h/shuaishu/.conda/STFGNNpy37pytorch/lib/python3.7/site-packages/torch/nn/modules/module.py", line 1130, in _call_impl
return forward_call(*input, **kwargs)
File "/work/users/s/h/shuaishu/Transformer/STFGNN-Pytorch/STFGNN/model/STFGNN.py", line 204, in forward
data_left = torch.sigmoid(self.conv1(data_temp))
File "/work/users/s/h/shuaishu/.conda/STFGNNpy37pytorch/lib/python3.7/site-packages/torch/nn/modules/module.py", line 1130, in _call_impl
return forward_call(*input, **kwargs)
File "/work/users/s/h/shuaishu/.conda/STFGNNpy37pytorch/lib/python3.7/site-packages/torch/nn/modules/conv.py", line 307, in forward
return self._conv_forward(input, self.weight, self.bias)
File "/work/users/s/h/shuaishu/.conda/STFGNNpy37pytorch/lib/python3.7/site-packages/torch/nn/modules/conv.py", line 304, in _conv_forward
self.padding, self.dilation, self.groups)
RuntimeError: Expected 2D (unbatched) or 3D (batched) input to conv1d, but got input of size: [64, 64, 358, 12]

I have searched on the Internet about this issue and it is probably due to the version of Pytorch. For 1.12.1+cu113 I use, I need to modify to Conv2d, with the same parameters. Do they do the same things? If so, I will not downgrade my Pytorch version. But I still want to know what version you were using. Thank you so much!

When I run it in another datasets, it achieve an incredible result. Preds is almost the same with the groud_truth in test_data. And, the groud_truth is changed when self.scaler.inverse_transform is used. Maybe it's not float32? I'm confused

您好，请问一下邻接矩阵是如何生成的，您那边方便提供一下PEMS04，07,08的adj_mx.pkl数据集吗？万分感谢