Swahili Hate Speech Detection Model

A deep learning bert model for detecting hate speech in Swahili text, deployed as a Flask web application with Docker support.

This project implements a hate speech detection system specifically designed for the Swahili language. The model can classify Swahili text as either containing hate speech or being neutral, helping to moderate content and promote safer online environments in Swahili-speaking communities.

Features

Swahili Language Support: Specifically trained and optimized for Swahili text REST API: Easy-to-use Flask API for integration Docker Support: Containerized deployment for easy scaling Real-time Detection: Fast inference for live content moderation High Accuracy: Trained on curated Swahili hate speech datasets

Clone the repository

git clone https://github.com/codeshujaa/swahili-model.git

cd swahili-model

Install dependencies

pip install -r requirements.txt

Build the Docker image

docker build -t swahili-hate-speech .

Run the container

docker run -p 5000:5000 swahili-hate-speech