Detect humor with recurrent neural networks.
This repository contains an exploration of recurrent neural networks (RNNs) and how they can be trained to detect humor. The Yelp Open Dataset is used for training and demonstration. The machine learning model is built with the help of PyTorch and spaCy.
Open the notebook right here in GitHub. If you want to view the notebook locally, install Conda and Jupyter with this guide. Then run:
$ jupyter notebook yelp_humor_detection.ipynb
and follow the initial instructions inside. These directions have been tested on macOS Mojave.
If you want to train the model yourself, grab the dataset from Yelp and store the reviews JSON file (named review.json) in a new folder called yelp_dataset within the working directory before running through the notebook.