GloVe Model for Igala Language

Overview

This repository contains the code and resources for training a GloVe (Global Vectors for Word Representation) model on Igala, a low-resource language. The project includes scripts for preprocessing the text data, training the GloVe model, evaluating the resulting word embeddings, and visualizing the embeddings using techniques such as t-SNE.

Features

• Data Preprocessing: Tools for cleaning and tokenizing Igala text data.
• GloVe Model Training: Implementation of a GloVe model using PyTorch.
• Evaluation: Scripts to evaluate word embeddings using cosine similarity, intrinsic evaluation methods, and extrinsic tasks.
• Visualization: Interactive visualization of word embeddings using t-SNE and Plotly.

Getting Started

Prerequisites

Before running the code, ensure you have the following dependencies installed:

• Python 3.x
• PyTorch
• NumPy
• Matplotlib
• Scikit-learn
• Plotly
• Pandas
• GPU

You can install these dependencies using pip:

pip install torch numpy matplotlib scikit-learn plotly pandas