Retrieval-Augmented Generation (RAG) APP

Overview

This application allows users to upload a document and ask questions against the document. The application uses Langchain, GroQ LLM, Azure OpenAI embeddings and Retrieval-Augmented Generation (RAG).

APP Features

• Load: Document ingestion
• Split: Document chunking using semantic chuncker, and embedding function from Azure openai.
• Store: This is done using a vector store (chromadb) and Embeddings model
• Retrieve: Given a user input, relevant splits are retrieved from storage using a retriever.
• Generate: A chatmodel LLM(GroQ) produces an answer using a prompt that includes both the question with the retrieved data.
• Chat: Streamlit web interface for seamless user interaction

Tools Used

• Python -> Backend development
• Langchain -> LLM framework
• ChromaDB -> vector store
• Streamlit -> Frontend/Chat Builder
• Docker -> Application packaging tool
• Docker Compose -> Application deployment

Steps to Run this Application

1. Have the following installed on your machine.

• Docker
• Docker-compose
• git

2. Clone the Repository

git clone https://github.com/erickasakya/langchain-rag-app.git

cd langchain-rag-app

3. Create .env from .env.template and update the api keys

mv .env.template .env

4. Build Docker Containers

docker compose up -d --build

Access

Open the browser and enter the url below.

http://localhost:8502

Test application.

1. Load the sample text document in context_data folder.
2. After loading use the below sample prompts to test

• Who is the writer of the book The Man Who Was Thursday: A Nightmare?
• Which year was it realased
• Which language