It is possible to make a concat layer with other simplest?

1

Hi, I’m now working in a project where the concat layer in low level implementation is a big problem, but I need it to add this operation on my model architecture.

I’m trying to implement this concat layer with other primitive operation using PyTorch, but it’s a bit tricky for me. Anyone had some possible solution to my problem?

Thank you in advance :slight_smile:

2

Could you describe a bit more what exactly this concat layer should perform and why torch.cat wouldn’t work?