Simple RAG

Run RAG (Retreival Augment Generation) for any documents!

Hey there! 👋 RAG-in-a-box is your go-to tool for quickly setting up Retrieval Augmented Generation (RAG) on your docs. It's perfect for when you're tired of uploading docs to chatgpt or want everything to be saved locally.

• 📚 Handles PDFs and Markdown files
• 🤖 Works with OpenAI and Ollama models
• 🖥️ Easy-to-use interface s/o Gradio
• 💻 CLI for you command-line lovers

Prerequisites

1. Python 3.10 or higher
2. pip or rye package manager

Installation

Choose one of the following methods:

• Using pip:

  pip install -r requirements.lock
  

• Using rye:

  rye sync
  

API Keys

• For OpenAI models: Set the OPENAI_API_KEY environment variable with your OpenAI API key.
• For Ollama models: Ensure Ollama is installed and running on your system.

Using the Gradio Interface

python3 src/rag_in_a_box/interface.py

Using the CLI

To use RAG-in-a-box from the command line, you can utilize the main.py script. Here’s how you can get started:

1. Basic Usage:

   python3 src/rag_in_a_box/main.py --path <path_to_your_documents>
   

   Replace <path_to_your_documents> with the path to the directory containing your PDF or Markdown files.

2. Specify Document Loader Type: By default, the script assumes you are loading PDF documents. If you want to load Markdown files, use the --loader_type argument:

   python3 src/rag_in_a_box/main.py --loader_type md --path <path_to_your_documents>
   

3. Persisting the Vector Database: You can specify a custom path to persist the vector database using the --persist_path argument:

   python3 src/rag_in_a_box/main.py --path <path_to_your_documents> --persist_path <path_to_persist_db>
   

4. Choosing the Model: The script supports both OpenAI and Ollama models. You can specify the model to use with the --model argument:

   python3 src/rag_in_a_box/main.py --path <path_to_your_documents> --model <model_name>
   

5. Interactive Q&A: Once the documents are loaded and stored, the script enters an interactive Q&A loop. Simply type your questions, and the system will provide answers based on the loaded documents. Type quit to exit the loop.

Example command:

python3 src/rag_in_a_box/main.py --loader_type pdf --path ./docs --persist_path ./chroma_db --model gpt-4o-mini-2024-07-18

This command loads PDF documents from the ./docs directory, stores them in the ./chroma_db vector database, and uses the gpt-4o-mini-2024-07-18 model for Q&A.