Word prediction with RNN

I’m looking for a detailed tutorial / explanation about building a RNN for predicting the next word of a phrase. I have the embeddings of each word obtained with Word2Vec. Now I’m trying to understand how to build the network for the prediction of the next word given a phrase of length N, for example.

INPUT —> The cat is on the
OUTPUT --> Softmax of some vectors (Table, Sofà, Kitchen etc…)

the word language example might help. https://github.com/pytorch/examples/tree/master/word_language_model