I’ve a case where I need to call the forward method like this:
class Net(nn.Module):
#This defines the structure of the NN.
def __init__(self):
super(Net, self).__init__()
self.conv1 = nn.Conv2d(1, 10, kernel_size=5)
self.conv2 = nn.Conv2d(10, 20, kernel_size=5)
self.conv2_drop = nn.Dropout2d() #Dropout
self.fc1 = nn.Linear(320, 50)
self.fc2 = nn.Linear(50, 10)
def forward(self, x):
x = F.relu(F.max_pool2d(self.conv1(x), 2))
x = F.relu(F.max_pool2d(self.conv2_drop(self.conv2(x)), 2))
x = x.view(-1, 320)
x = F.relu(self.fc1(x))
x = F.dropout(x, training=self.training)
x = self.fc2(x)
return F.log_softmax(x, dim=1)
def sampler(self,x):
print("*",end=" ",flush=True)
y = self.forward(x)
# Do something to y!
....
return y
model = Net()
y = model.sampler(x)
loss = F.nll_loss()(y,target)
loss.backward()
Will the loss backpropagation perform normally without breaking any internal hooks? If it is breaking it, what’s the alternative without changing the model structure?
Thanks in advance.