Input tensors with different size

autograd
1

I have a network, whose inputs are 2 tensors, like:

def forward(self, inp, rgb):
     #  
    return model

the inp and rgb have different size(both 4D)
while I run model.forward(x1, x1), it went wrong:


    if input is not None and input.dim() != 4:
AttributeError: 'list' object has no attribute 'dim'

is that the input of module must have be one argument?
thx

2

can you give a full example reproducing this?

Also you should do: model(x1, x1), not model.forward(x1, x1).