Hi!
Below is an example of the simplest linear multi-input network in Mathematica 14.1 (in fact it was suggested by Grok 3, xAI chatbot):
In[]:=
createNetwork[]:=NetGraph[​​<|​​"Input1"->LinearLayer[3],​​"Input2"->LinearLayer[3],​​"Input3"->LinearLayer[3],​​"ThreeToOne"->FunctionLayer[​​Function[{v1,v2,v3},v1+2v2+3v3],​​"Input"->{3,3,3},​​"Output"->LinearLayer[3]​​]​​|>,​​{​​NetPort["Input"]->"Input1",​​NetPort["Input"]->"Input2",​​NetPort["Input"]->"Input3",​​{"Input1","Input2","Input3"}->"ThreeToOne"->"Output"​​}​​];
But something is wrong:
In[]:=
net=createNetwork[];
FunctionLayer
:Part Input does not exist.
NetGraph
:$Failed is not a layer, a net, or a valid specification for one.
I have tried many obvious ways to fix the code, but all my efforts have led to nothing. Where is the error and how to fix it? I understand that this training task can be solved more simply (and without using three inputs), but my goal is to understand how NetGraph and FunctuionLayer work, so that later I can build more complex neural networks with more inputs.
​
Vladimir.