I am constructing a simple MLP model and trying to use 2 different inputs (A and B) to predict y. Before training a model, I need to feed inputs to the model correctly. I define a first layer to be nn.Linear(2,next_dim). I would like to ask whether I need to flatten out all inputs by concatenate(A,B) before feeding to a model? I believe that I have to but not really confident about this! Can anyone confirm this for me by chance, please?
Secondly, if I want to train 2 different models (more like multi model), can I train each model with different type of sending data to a model differently? what I meant is that I use a dataloader to train model A and use a normal batch (dataloader does not involved) to train model B
Thank you in advances,