/rag_pdf

This project is a basic implementation of a Retrieval-Augmented Generation (RAG) system for PDF documents using LangChain, Chroma, and Ollama.

Primary LanguageJupyter Notebook

RAG for PDFs

This project is a basic implementation of a Retrieval-Augmented Generation (RAG) system for PDF documents using LangChain, Chroma, and Ollama. The system loads a PDF, splits the text into manageable chunks, creates embeddings, and then allows for retrieval-based question answering using a language model.

Project Structure

  • PDF Loading: Uses PyPDFLoader to load the content of a PDF file.
  • Text Splitting: Utilizes RecursiveCharacterTextSplitter to split the document into chunks for processing.
  • Embeddings: Embeddings are generated using the SentenceTransformerEmbeddings.
  • Vector Store: Chroma is used as the vector store to persist and search embeddings.
  • RAG Implementation: Combines the embedding-based retrieval with a language model (OllamaLLM) to answer questions based on the content of the PDF.

Getting Started

Prerequisites

  • Python 3.9 or higher
  • Poetry for dependency management
  • Ollama installed on your system

Installation

  1. Clone the Repository: 
    git clone <repository-url>
cd rag-pdf