How to add a valid blank graph

[Limee-ee]

Hello, I've got a series of graph lists, but they don't have the same length, in order to use our dataloader, I'd like to pad the shorter lists to the same length. Now, I use the following blank graph to pad.

from torch_geometric.data import Data
x=torch.tensor([], dtype=torch.float)
edge_index=torch.tensor([[], []], dtype=torch.long)
edge_attr = torch.tensor([])

data = Data(x=x, edge_index=edge_index,edge_attr=edge_attr)

It works with the dataloader. But when I put it to the Weight_GCN layer, it shows that:

Traceback (most recent call last):
  File "C:\Program Files\JetBrains\PyCharm 2021.3\plugins\python\helpers\pydev\pydevd.py", line 1483, in _exec
    pydev_imports.execfile(file, globals, locals)  # execute the script
  File "C:\Program Files\JetBrains\PyCharm 2021.3\plugins\python\helpers\pydev\_pydev_imps\_pydev_execfile.py", line 18, in execfile
    exec(compile(contents+"\n", file, 'exec'), glob, loc)
  File "/train.py", line 88, in <module>
    prediction = model(data[0][0], data[1][10], data[0][1], data[2])
  File "C:\Users\anaconda3\envs\pytorch\lib\site-packages\torch\nn\modules\module.py", line 1102, in _call_impl
    return forward_call(*input, **kwargs)
  File "/model.py", line 130, in forward
    graph_embedding=self.GNN_layer(g)
  File "C:\Users\anaconda3\envs\pytorch\lib\site-packages\torch\nn\modules\module.py", line 1102, in _call_impl
    return forward_call(*input, **kwargs)
  File "/model.py", line 65, in forward
    x = graph_data.x  
AttributeError: 'tuple' object has no attribute 'x'

So, how should I pad a blank graph correctly😔Thank you very much~

ps: I use this Weight_GCN

class Weight_GCN(MessagePassing):
    def __init__(self, x_dim):
        super(Weight_GCN, self).__init__(aggr='add')
        self.message_layer = nn.Linear(x_dim, x_dim, bias=False)
        self.update_mlp = nn.Sequential(
            nn.Linear(2 * x_dim, x_dim),
            nn.ReLU(),
            nn.Linear(x_dim, x_dim)
        )

    def forward(self, x, edge_index, edge_attr):
        receive_message = self.propagate(edge_index, x=x, edge_attr=edge_attr)
        update_x=self.update_mlp(torch.cat([x, receive_message], dim=1))
        return update_x

    def message(self, x_j, edge_attr):
        return edge_attr.view(-1, 1) * self.message_layer(x_j)

[rusty1s]

Note that x needs to have the correct shape. In your case, you can initialize it as

x = torch.empty((0, x_dim), dtype=torch.float)