Dynamic AI Chatbot with Custom Training Sources

Customizable-gpt-chatbot

This project is a dynamic AI chatbot that can be trained from various sources, such as PDFs, documents, websites, and YouTube videos. It uses a user system with social authentication through Google, and the Django REST framework for its backend. The chatbot leverages OpenAI's GPT-3.5 language model to conduct conversations and is designed for scalability and ease of use.

Features

• Train chatbot from multiple sources (PDFs, documents, websites, YouTube videos)
• User system with social authentication through Google
• Connect with OpenAI GPT-3.5 language model for conversation
• Use Pinecone and FAISS for vector indexing
• Employ OpenAI's text-embedding-ada-002 for text embedding
• Python Langchain library for file processing and text conversion
• Scalable architecture with separate settings for local, staging, and production environments
• Dynamic site settings for title and prompt updates
• Multilingual support
• PostgreSQL database support
• Celery task scheduler with Redis and AWS SQS options
• AWS S3 bucket support for scalable hosting
• Easy deployment on Heroku or AWS

Technologies

• Language: Python
• Framework: Django REST Framework
• Database: PostgreSQL

Major Libraries:

• Celery
• Langchain
• OpenAI
• Pinecone
• FAISS

Requirements

• Python 3.8 or above
• Django 4.1 or above
• Pinecone API Key
• API key from OpenAI
• Redis or AWS SQS
• PostgreSQL database

Future Scope

• Integration with more third-party services for authentication
• Support for additional file formats and media types for chatbot training
• Improved context-awareness in conversations
• Enhanced multilingual support with automatic language detection
• Integration with popular messaging platforms and chat applications

How to run

• Clone the repository. git clone https://github.com/shamspias/customizable-gpt-chatbot
• Install the required packages by running pip install -r requirements.txt
• Run celery celery -A config worker --loglevel=info
• Run the command python manage.py runserver
• Open http://127.0.0.1:8000/ in your browser

In linux and mac need to install 'sudo apt install python3-dev -y`

1. Make sure that you have the development libraries for libcurl installed on your system. You can install them by running the following command: sudo apt-get install libcurl4-openssl-dev gcc libssl-dev -y
2. Make sure that you have the latest version of pip and setuptools installed by running the following command: pip install --upgrade pip setuptools
3. pip install pycurl

Deployment

The chatbot can be deployed on Heroku or AWS by following the standard procedures for Django deployment on these platforms.

Issues

• If you don't use AWS SQS then no need to install pycurl and boto3 packages.
• If you don't use AWS S3 then no need to install django-storages package.

Note

Make sure that you have API key from OpenAI before running the project.

This is just a basic implementation of the project, you can always add more features and customization according to your requirement.

Enjoy!