Modeling Encoder-Decoder according to instructions from a paper

Question

I am new to this field and I was reading a paper "Predicting citation counts based on deep neural network learning techniques". There the authors describe the code