DeepSets Keras Architecture

I am interested in the Deepsets Model but I am not quite sure how to implement it in Keras/Tensorflow for a normal regression task. More specifically the DeepSet model should take an arbitrary-sized set and perform a regression task.

My question is the following: In the Deepest model, the encoder part should be shared across the elements in the set and it should work for an arbitrary number of elements (different sized). How to enable this in Keras? How to make a shared encoder ? How to make it work on different set sizes?