A simple RNN built from scratch using numpy. Inspired by a blog post by Andrej Karpathy from 2015. Can be used to train on simple data like dataset of dinosaur names. Can be used to generate new dinosaur names after training.
Read Andrej Karpathy's blog post here.
clone this repo and run the code to start training a simple RNN with the dinosaur names dataset thats provided in the dataset folder. The model is a simple RNN using numpy in python, hence can not be trained on large text datasets like shakespeare texts, which require capturing long range dependancies within the data. Feel free to fork this repo and try using your own datasets to see what works and what does not.