I am using both Torch and Caffe for deep learning. I wonder if there is a way to output nn model into caffemodel file. It may involve protobuf, which I am not familiar with.
It is kinda easy for the other way, where there are libraries to read caffemodel file into torch.