RNN Implementation for point classification


I understand most implementation of RNN’s are used with NLP, at least all the ones I have seen online. Has anyone found an implementation that uses classification with a group of points?

So the sequence length will be the number of points in a group, but I am confused on such parameters like “dict_size” that I have frequently seen in RNN implementations. I am not sure if the “dict_size” comes from the number of possible inputs (infinite because the points are on a continuous scale) or if it comes from the number of possible outputs (2 here because)I am currently only classifying between two classes.)

I am using an RNN because I want the choice of classification for one point to affect the classification of subsequent points.

Thank you so much for the help!