JonoMopp/NOUNDIS

UCT Computer Science Honours Project 2024 showcasing the ability of a Multi-Layer Perceptron & a fine-tuned version of Serengeti to classify nouns in Sepedi and isiZulu. # MLP and Serengeti Noun Classification for Sepedi and isiZulu ## Project Overview This project focuses on classifying nouns in Sepedi and isiZulu using a Multi-Layer Perceptron (MLP) combined with the Serengeti language model. ## DataSets & WordVectors The datasets & word embeddings that were used in this project can be found in the DataSets & WordVecs folders respectively. ## How to Use ### Prerequisites Ensure the following prerequisites are met before proceeding: - Python 3.8+ - Required Python libraries (specified in `requirements.txt`) To install the dependencies, use: ```bash pip install -r requirements.txt ``` ## Managing Model Hyperparameters Both the MLP and Serengeti are handled using the `config.json` file which is found in the `utils` folder. In order to modify the various hyperparameters, change the corresponding values in the `config` file and it will automatically update on the next run of the program. ## Steps to Use the Classifier 1. Run the decoupledTesting.py file 2. Follow the commandline instructions ## Connect With Me