PDF-GPT

• It is a RAG application where I have used GoogleGenerativeAI LLM model to generate text.
• PDF-GPT helps to extract data and answer the questions related to given pdf.
• It has chat architecture where we can chat with LLM in different context at the same time.
• Whole project is made using OOPS design method.

Technologies used

• Retrieval Augmented Generation(RAG)
• Langchain framework for working with LLM's
• Tried GooglePalm, GoogleGenerativeAI, OpenAI model and their embeddings
• FAISS(Facebook AI Similarity Search) as vector database
• PyPDF for extracting data from pdf's
• PostgreSQL(psycopg2) for storing chats
• FastAPI for making python server
• Vite + TS + React for making frontend application
• React-Hot-Toast for toast notification
• React-Markdown for showing markdown format of otuput
• React-Router-Dom for making routes for different chats.
• React-To-Print for extracting chats in pdf form.

Start Application

• Frontend npm run dev
• Backend uvicorn app:app --reload --port 3000
• Build Frontend npm run build
• Start Frontend after build npm run preview
• Frontend Dev URL: http://localhost:5173
• Frontend Prod URL: http://localhost:4173

Project Demo Video

Video Link