How dose work class of Network

1

I have this class.
class Net(torch.nn.Module):
def init(self, D_in, H, D_out):
super(Net, self).init()
self.fc1 = nn.Linear(D_in, H)
self.fc2 = nn.Linear(H, H)
self.fc3 = nn.Linear(H,H)
self.fc4 = nn.Linear(H, D_out)
self.active_1 = torch.nn.Tanh()
self.active_2 = torch.nn.Tanh()
self.active_3 = torch.nn.Tanh()
self.fc11 = nn.Linear(D_in, H)
self.fc22 = nn.Linear(H, H)
self.fc33 = nn.Linear(H, H)
self.fc_tot = nn.Linear(D_out, D_out)
##################################

def forward(self, x):
    """
    In the forward function we accept a Tensor of input data and we must return
    a Tensor of output data. We can use Modules defined in the constructor as
    well as arbitrary operators on Tensors.
    """
    y = self.active_1(self.fc1(x))
    y = self.active_2(self.fc2(y))
    y = self.active_3(self.fc3(y))
    y = self.fc4(y)
    y = y*y########################
    y = self.fc_tot(y)
    return y

As you see I have not to use these three functions( self.fc11 = nn.Linear(D_in, H),self.fc22 = nn.Linear(H, H),self.fc33 = nn.Linear(H, H)) in forward part, but when I delete from the class, the result gets completely change? why?

2

This looks quite odd. Is the complete model you are running?

3

yes, and the result is good but I do not know, how can I explain it.
the only parameter that, get change in class is the " _cdata".