Skip to content

Latest commit

 

History

History
89 lines (58 loc) · 3.89 KB

README.md

File metadata and controls

89 lines (58 loc) · 3.89 KB
Raw

R2R

R2R (RAG to Riches) is a Python framework designed for the rapid construction and deployment of production-ready Retrieval-Augmented Generation (RAG) systems. This semi-opinionated framework accelerates the transition from experimental stages to production-grade RAG systems.

Developer Installation

Fast Install:

  1. Install R2R directly using pip:

    pip install r2r

Full Install:

Follow these steps to ensure a smooth setup:

  1. Install Poetry:

    • Before installing the project, make sure you have Poetry on your system. If not, visit the official Poetry website for installation instructions.

  2. Clone and Install Dependencies:

    • Clone the project repository and navigate to the project directory: 
      git clone [email protected]:SciPhi-AI/r2r.git
cd r2r
    • Install the project dependencies with Poetry: 
      poetry install

  3. Configure Environment Variables:

    • You need to set up cloud provider secrets in your .env file for the project to work properly. At a minimum, you will need an OpenAI key and a vector database provider.
    • For a fast setup, we recommend creating a project on Supabase, enabling the vector extension, and then updating the .env.example with the necessary details.
    • Other providers are also available, such as Qdrant for vector database support.
    • Once updated, copy the .env.example to .env to apply your configurations: 
      cp .env.example .env

Demonstration

qt_exp_1_720p.mp4

Community

Join our Discord server!

Core Abstractions

The framework primarily revolves around three core abstractions:

  • The Ingestion Pipeline: Facilitates the preparation of embeddable 'Documents' from various data formats (json, txt, pdf, html, etc.). The abstraction can be found in ingestion.py.

  • The Embedding Pipeline: Manages the transformation of text into stored vector embeddings, interacting with embedding and vector database providers through a series of steps (e.g., extract_text, transform_text, chunk_text, embed_chunks, etc.). The abstraction can be found in embedding.py.

  • The RAG Pipeline: Works similarly to the embedding pipeline but incorporates an LLM provider to produce text completions. The abstraction can be found in rag.py.

Each pipeline incorporates a logging database for operation tracking and observability.

Running the Examples

The project includes several basic examples that demonstrate application deployment and standalone usage of the embedding and RAG pipelines:

  1. app.py: This example runs the main application, which includes the ingestion, embedding, and RAG pipelines served via FastAPI.

    poetry run uvicorn examples.basic.app:app

  2. test_client.py: This example should be run after starting the main application. It demonstrates a test of the user client.

    poetry run python -m examples.client.test_client

  3. rag_pipeline.py: This standalone example demonstrates the usage of the RAG pipeline. It takes a query as input and returns a completion generated by the OpenAI API.

    poetry run python -m examples.basic.rag_pipeline

  4. embedding_pipeline.py: This standalone example demonstrates the usage of the embedding pipeline. It loads datasets from HuggingFace, generates embeddings for the data using the OpenAI API, and stores the embeddings in a PostgreSQL vector database.

    poetry run python -m examples.basic.embedding_pipeline