This project implements a multilingual chatbot using Llama 3 integrated with Ollama. It supports multiple languages and evaluates the chatbot's responses using BLEU and ROUGE metrics.
# Low-Resource Language Chatbot
This project implements a multilingual chatbot using **Llama 3** integrated with **Ollama**. It supports multiple languages and evaluates the chatbot's responses using BLEU and ROUGE metrics.
## Table of Contents
- Overview
- Prerequisites
- Installation
- Usage
- Project Structure
- Dependencies
- Contributing
- License
## Overview
The **Low-Resource Language Chatbot** project aims to create a chatbot that supports multiple languages, including English, Spanish, French, Amharic, and Hindi. It uses **Llama 3** via **Ollama API** to generate responses and `transformers` library for BLEU and ROUGE score evaluations.
### Features:
- **Multilingual Support**: Select and interact with the chatbot in various languages.
- **Response Evaluation**: Uses BLEU and ROUGE metrics to evaluate the quality of chatbot responses.
- **User Interface**: A simple web interface using **Flask** to interact with the chatbot.
## Prerequisites
Before running the project, ensure you have the following installed:
- Python (version 3.8 or higher)
- `pip` package manager
- ngrok (to expose the local server publicly)
### Recommended Setup:
- **Python**: Download Python
- **Pip**: Comes bundled with Python by default.
- **Ngrok**: Download Ngrok for exposing local development servers.
## Installation
1. Clone the repository:
```bash
git clone
github.com
cd low-resource-chatbot
```
2. Install the required dependencies:
```bash
pip install -r requirements.txt
```
3. Start the Flask application:
```bash
python app.py
```
4. Expose the local server using ngrok:
```bash
ngrok http 5000
```
Note: Replace `5000` with the port the Flask app is running on if it’s different.
5. Access the chatbot:
- You can now open the chatbot at `
127.0.0.1` or the ngrok URL.
## Usage
1. **Start the Chatbot**:
- Run the Flask application with `python app.py`.
- Open `
127.0.0.1` in your web browser.
2. **Inte …