Multi-output regression problem

Hi,

My model has several outputs from the forward method:

def forward(self, x):
       ---code---
       return ClCd, angle

This returns a tuple, which LR finder does not like. I get the following error message:

if not (target.size() == input.size()):
AttributeError: 'tuple' object has no attribute 'size'

Is there a way for LR finder to work with tuples? Alternatively, should I be structuring the output from my forward method differently (i.e. using a single output tensor)? I tried outputting a single tensor with two columns from my forward method (each column representing an output), but this gave significantly worse results in training.

[Stoops-ML]

This worked perfectly. Thank you very much to the both of you for all your time, patience and clear instructions!

[NaleRaphael]

@davidtvs Agree with that. Making a wrapper for multiple loss functions is way more elegant for such case!

@Stoops-ML As it’s shown in the link provided by @davidtvs, loss function is also a subclass of nn.Module. Therefore, you can simply implement a wrapper like the following one:

class MyLossFunc(nn.Module):
    def __init__(self, ...):
        self.loss_func_1 = ...
        self.loss_func_2 = ...

    def forward(self, inputs, labels):
        loss_1 = self.loss_func_1(inputs, labels)
        loss_2 = self.loss_func_2(inputs, labels)
        return loss_1 + loss_2

See also this post for the reason why it works.