RuntimeError: Expected object of scalar type Long but got scalar type Float for argument #2 'mat2' in call to _th_mm

1

Can’t get this GCN to work. Any help is appreciated.

  class Net(torch.nn.Module):                                                        
      def __init__(self):                                                            
          super(Net, self).__init__()                                                
          self.conv1 = GCNConv(dataset.num_features, 16, cached=True)                
          self.conv2 = GCNConv(16, int(dataset.num_classes), cached=True)            
          # self.conv1 = ChebConv(data.num_features, 16, K=2)                        
          # self.conv2 = ChebConv(16, data.num_features, K=2)                        
                                                                                     
          self.reg_params = self.conv1.parameters()                                  
          self.non_reg_params = self.conv2.parameters()                              
                                                                                     
      def forward(self):                                                             
          x, edge_index, edge_weight = data.x, data.edge_index, data.edge_attr       
          x = F.relu(self.conv1(x, edge_index, edge_weight))                         
          x = F.dropout(x, training=self.training)                                   
          x = self.conv2(x, edge_index, edge_weight)                                 
          return F.log_softmax(x, dim=1)                                             
                                                                                     
  def train():                                                                       
      model.train()                                                                  
      optimizer.zero_grad()                                                          
      F.nll_loss(model()[data.train_mask], data.y[data.train_mask].long()).backward()
      optimizer.step()

Error:

Traceback (most recent call last):
  File "src/meta_gnn.py", line 395, in <module>
    train()
  File "src/meta_gnn.py", line 307, in train
    F.nll_loss(model()[data.train_mask], data.y[data.train_mask].long()).backward()
  File "/home/ppandey/.local/lib/python3.6/site-packages/torch/nn/modules/module.py", line 532, in __call__
    result = self.forward(*input, **kwargs)
  File "src/meta_gnn.py", line 299, in forward
    x = F.relu(self.conv1(x, edge_index, edge_weight))
  File "/home/ppandey/.local/lib/python3.6/site-packages/torch/nn/modules/module.py", line 532, in __call__
    result = self.forward(*input, **kwargs)
  File "/home/ppandey/.local/lib/python3.6/site-packages/torch_geometric/nn/conv/gcn_conv.py", line 87, in forward
    x = torch.matmul(x, self.weight)
RuntimeError: Expected object of scalar type Long but got scalar type Float for argument #2 'mat2' in call to _th_mm
2

The error is raised, because you are feeding LongTensors to a matmul (e.g. via nn.Linear) somewhere in your model.
Based on your code I cannot see, how the input data is fed to the model, and it appears you are not passing anything to the model at all:

model()[data.train_mask]

but try to index the output?

Anyway, try to check the input.type() and make sure its float32.
You can transform tensors to this type via:

input = input.float()
1 Like