IbnuEyni/Amharic-Hate-Speech-Detector

# Amharic Hate Speech Detector This application is designed to detect hate speech in Amharic text. It uses a fine-tuned BERT-based model to classify input text into two categories: "Hate Speech" or "Free Speech." The model is loaded dynamically from Hugging Face, and the application is implemented using FastAPI for efficient and scalable API services. ## Features - **Dynamic Model Loading**: The model can be dynamically selected by specifying its name in an environment variable. - **RESTful API Endpoint**: The application provides a single endpoint (`/predict`) to classify text. - **Customizable Labels**: Outputs are mapped to human-readable labels (`free` and `hate`). - **Web-Based Interface**: Includes a simple HTML template to interact with the API and test the model. ## Getting Started ### Prerequisites Ensure the following are installed on your system: - Python 3.8 or later - pip (Python package manager) - `virtualenv` (optional, but recommended) ### Installation 1. Clone the repository: ```bash git clone github.com cd amharic-hate-speech-detector ``` 2. Create a virtual environment (optional): ```bash python3 -m venv venv source venv/bin/activate # On Windows: venv\Scripts\activate ``` 3. Install dependencies: ```bash pip install -r requirements.txt ``` 4. Set up environment variables: - Create a `.env` file in the root directory: ```env MODEL_NAME=amharic-hate-speech-detection-mBERT ``` - Replace `amharic-hate-speech-detection-mBERT` with your preferred model name on Hugging Face if needed. ## Usage ### Running the Application 1. Start the FastAPI application: ```bash uvicorn main:app --reload ``` 2. Open your browser and visit: - API Docs: 127.0.0.1 - Web Interface: Open `index.html` in a browser (manually or host it using a simple HTTP server). ### API Endpoints #### **POST** `/predict` Predicts whether the given Amharic text is hate speech or free speech. - ** …