Elementary architecture implementation problem

I am new to pytorch and I was wondering how I can create the neural network described in figure 2 of this paper [1909.07481] Deep Neural Networks for Choice Analysis: Architectural Design with Alternative-Specific Utility Functions. I can create figure 1 but I am confused on how to go about creating figure 2. I tried structural pruning by setting weights and biases to zero but I believe it should be much simpler than that.

A simple example would be highly appreciated.

Thank you.