Since this can be a classification drawback, we’ll use a "many to one" RNN.
Later in this post, we’ll build a "many to one" RNN from scratch to carry out basic Sentiment Evaluation.