Custom loss function

1

Hey! I have already to define my own loss function. It does work. But I am not sure whether it is correct because I don’t define the backward(). I am not sure whether I need to define backward(), and I don’t know how to define backward().

class _Loss(nn.Module):
    def __init__(self, size_average=True):
        super(_Loss, self).__init__()
        self.size_average = size_average
class MyLoss(_Loss):
    def forward(self, input, target):
        loss = 0
        weight = np.zeros((BATCH_SIZE,BATCH_SIZE))
        for a in range(BATCH_SIZE):
            for b in range(BATCH_SIZE):
                weight[a][b] = Censor(target.data[a][0])
        for i in range(BATCH_SIZE):
            for j in range(BATCH_SIZE):
                a_ij= (input[i]-input[j]-target[i]+target[j])*weight[i,j]
                loss += F.relu(a_ij)
        return loss
2

Sorry but I can’t read your code. Try format your code using

```Python
your code here
```

Every op in the forward function need to be Variable
so it need to be

weight=Variable(t.zeros(Batch_size,Batch_size))

and maybe you shall not use for loop, try vectorize your code.

3

I am sorry for inconvenience.
Thanks for your reminder. This is my first time writing codes. So I do not know how to write it.

4

Hi, @11127

You do not need to define a class for your loss. A function is sufficient and the backward() comes for free (i.e. you do not need to write any code for it).