Notebook LLM Project

Overview

The Notebook LLM Project is a Spring AI-based application that allows users to upload PDF documents and ask questions about the content. The application leverages two large language models (LLMs) for its functionality: Groq OpenAI and Ollama. It is built using Spring Boot and follows a Gradle project structure.

Features

• Upload PDFs: Users can upload PDF documents for processing.
• Ask Questions: Users can interact with the application to query the content of the uploaded PDF.
• Multiple Model Support: The application integrates with two LLMs:
  • Groq OpenAI
  • Ollama
• Spring AI Integration: Simplifies interaction with the language models.
• Cassandra: As a VectorStore and to store user's chat.
• PostgreSQL: To store user's information

Prerequisites

To run this project, ensure you have the following installed:

• Java 21 or higher
• Gradle 7.6 or higher
• A valid API key for OpenAI and Ollama models

Setup Instructions

1. Clone the Repository:

   git clone https://github.com/KushParsaniya/notebook-llm
   cd notebook-llm

2. Add API key: Add api key for openaAI, and install ollama in your local machine.

3. Build the Project: Use Gradle to build the project:

./gradlew build

4. Run the Application: Start the Spring Boot application:

./gradlew bootRun

Project Structure

• src/main/java: Contains the Java source code for the application.
• src/main/resources: Contains configuration files like application.yml.
• build.gradle: The Gradle build script.

How It works:

1. PDF Upload:

   • The user uploads a PDF document via the UI.
   • The application processes the uploaded file and extracts its content.
   • and store it in vector store with metadata of user.

2. Query Handling:

   • The user inputs a question related to the PDF content.
   • The question is processed by the integrated LLMs (Groq OpenAI and Ollama).

3. Information Storage:

   • Ollama extracts useful information from user interactions and stores it in the Cassandra vector store.
   • User information and chat history are stored in PostgreSQL.

4. Response Generation:

   • The application uses the LLMs to generate a response based on the PDF content.
   • The response is displayed to the user.