Something wrong with my forward function

vision
1

At first, I write a samll net named Tnet, and it works well, but now I want to use it in a bigger net named Mnet, so I write

self.tnet=Tnet()

in Mnet’s __init__ function, but it catches a error during forward process like this:

out, _ = cls(data)
  File "/home/zqk/anaconda2/envs/torch/lib/python2.7/site-packages/torch/nn/modules/module.py", line 489, in __call__
    result = self.forward(*input, **kwargs)
  File "mynet.py", line 109, in forward
    tout = self.tnet(x)
  File "/home/zqk/anaconda2/envs/torch/lib/python2.7/site-packages/torch/nn/modules/module.py", line 535, in __getattr__
    type(self).__name__, name))
AttributeError: 'Mnet' object has no attribute 'tnet'

what’s wrong with it?

2

can anyone help me? thanks for your help!

3

Could you post the whole model code so that we could have a look?
Also, check for typos and the right indent in your __init__.

4 
class TNet(nn.Module):
    def __init__(self,number = 3):
        super(TNet,self).__init__()
        self.number = number
        self.conv1 = torch.nn.Conv1d(self.number,64,1)
        self.conv2 = torch.nn.Conv1d(64,128,1)
        self.conv3 = torch.nn.Conv1d(128,1024,1)

        self.fc1 = nn.Linear(1024,512)
        self.fc2 = nn.Linear(512,256)

        if self.number == 3:
            self.fc3 = nn.Linear(256,9)
        elif self.number == 64:
            self.fc3 = nn.Linear(256,64*64)

        self.relu = nn.ReLU()

        self.bn1 = nn.BatchNorm1d(64)
        self.bn2 = nn.BatchNorm1d(128)
        self.bn3 = nn.BatchNorm1d(1024)
        self.bn4 = nn.BatchNorm1d(512,affine=False)
        self.bn5 = nn.BatchNorm1d(256,affine=False)
    
    def forward(self, x):
        batchsize = x.size()[0]

        x = self.conv1(x)
        x = F.relu(self.bn1(x))

        x = self.conv2(x)
        x = F.relu(self.bn2(x))

        x = self.conv3(x)
        x = F.relu(self.bn3(x))

        x = torch.max(x,2,keepdim=True)[0]
        x = x.view(-1,1024)

        x = F.relu(self.fc1(x))
        x = F.relu(self.fc2(x))
        x = self.fc3(x)

        if self.number == 3:
            iden = Variable(torch.from_numpy(np.array([1,0,0,0,1,0,0,0,1]).astype(np.float32))).view(1,9).repeat(batchsize,1)
        elif self.number == 64:
            arr = generate_dim()
            iden = Variable(torch.from_numpy(np.array(arr).astype(np.float32))).view(1,64*64).repeat(batchsize,1)
        
        if x.is_cuda:
            iden = iden.cuda()

        x = x + iden
        x = x.view(-1, self.number, self.number)
        return x

class Mnet(nn.Module):
    def __init(self):
        super(Mnet,self).__init__()
        self.tnet1 = TNet()
        self.tnet2= TNet(64)
        self.conv1 = torch.nn.Conv1d(3,64,1)
        self.conv2 = torch.nn.Conv1d(64,128,1)
        self.conv3 = torch.nn.Conv1d(128,1024,1)

        self.bn1 = nn.BatchNorm1d(64)
        self.bn2 = nn.BatchNorm1d(128)
        self.bn3 = nn.BatchNorm1d(1024)
    
    def forward(self,x):
        batchsize = x.size()[0]
        number = x.size()[2]

        tranx1 = self.tnet1(x)

        x = x.transpose(2,1)
        x = torch.bmm(x,tranx1)
        x = x.transpose(2,1)
        
        x = self.conv1(x)
        x = F.relu(self.bn1(x))

        tranx2 = self.tnet2(x)
        x = x.tranpose(2,1)
        x = torch.bmm(x,tranx2)
        x = x.transpose(2,1)

        feat = x

        x = self.conv2(x)
        x = F.relu(self.bn2(x))

        x = self.conv3(x)
        x = F.relu(self.bn3(x))
        x = torch.max(x,2,keepdim=True)[0]
        x = x.view(-1,1024)

        x = x.view(-1,1024,1).repeat(1,1,number)
        return torch.cat([x,feat],1), tranx2

the error said that Mnet has no tnet1

5

Thanks for the code.
The __init__ method in MNet has a typo (the second two underscores are missing).

7

:sweat_smile: thanks for your careful inspection. I made a fool mistake :（