oMarquess / rgt_assessment_2

Document-Understanding Home Assessment (DUHA)

Overview

This application leverages advanced AI to understand and answer questions based on PDF or CSV files. It utilizes LangChain, ChromaDB, and OpenAI's models.

Features

• Custom file handling command for PDF/CSV files.
• Django Rest API Backend with a frontend application.
• Integration with LangChain and OpenAI's models.
• ChromaDB for embedding data storage and retrieval.
• LangChain for conversational and user feedback mechanism based on memory.
• Context limited to the provided document.

Running the Application

0. pip install -r requirements.txt
1. cd duha

Backend :: [Django-Rest Framework - API Engine]

2. Run python manage.py runserver 8000 to start the Django server on port 8000.

Frontend (Port 8001)

3. Open a new terminal and start a separate (frontend) server on python manage.py runserver 8001
4. Alternatively, you may navigate to the "templates/index.html" and run a live server from VS Code.

File Handling

4. Drop your PDF or CSV file in the library folder which has been listed in .gitignore file [kindly recreate it in the duha/ directory] and run the command python manage.py file_handler .\library\<filename>
5. Alternatively, run the command python manage.py file_handler <path to your preferred PDF / CSV>

• The file_handler custom command handles the splitting, embedding and saving of the file chunks to a persistent ChromaDB.

API Interaction

The application is secured with authentication and uses Django's REST framework.

Contributing

Contributions are welcome in the form of comments.