PersonaGPT

Overview

This project is a Brand Identity Report Generator that utilizes Retrieval-Augmented Generation (RAG) to create detailed brand identity reports. It processes PDF documents, extracts relevant insights, and generates structured reports using Llama 3 via Ollama.

Features

• Loads PDF files from a designated directory.
• Extracts text using PyPDFLoader.
• Splits text into smaller, manageable chunks for efficient processing.
• Embeds text using SentenceTransformerEmbeddings (all-MiniLM-L6-v2).
• Stores and retrieves data using ChromaDB.
• Processes user input via Chainlit chatbot.
• Generates comprehensive brand reports using Llama 3.

Installation

Ensure you have Python installed and run the following command to install dependencies:

pip install chainlit langchain langchain_community chromadb langchain-ollama sentence-transformers

Setup

1. Place your PDF files inside:

   /teamspace/studios/this_studio/data_personlity
   

2. Install and run Ollama:

   ollama pull llama3
   

Running the Application

Start the Chainlit chatbot:

chainlit run app.py

Usage

1. Input brand details in the following format:

   Brand Name | Industry | Core Values | Target Audience | Competitors | Vision | Tone | Visuals
   

2. The system retrieves relevant brand-related information from ChromaDB.
3. A structured brand identity report is generated and sent to the user.

File Structure

/app
 ├── app.py                 # Main script
 ├── requirements.txt       # Dependencies
 ├── /data_personlity       # Folder for PDF files
 ├── /chromadb              # Vector database storage

Demo

Watch the YouTube demo: Demo Video

Presentation

View the project presentation: Presentation