Prebuilt Pipelines

R2R Pipeline Templates

This repository contains a collection of templates for deploying Retrieval Augmented Generation (RAG) pipelines using the R2R framework with the SciPhi cloud platform. Each template provides a starting point for building RAG pipelines with different capabilities and configurations.

Available Templates

  1. Q&A RAG (GitHub Repository (opens in a new tab))

    • A basic RAG pipeline template that serves as a foundation for building more complex pipelines.
    • Out of the box API for ingesting documents, streaming RAG completions, deleting documents, and more.
    • Supports multiple document, multiple user question & answer RAG.
    • Provides configuration options for vector database provider, evaluation settings, embedding settings, text splitter settings, and language model provider.
  2. Synthetic Queries (GitHub Repository (opens in a new tab))

    • An extension of the basic RAG pipeline that incorporates synthetic query generation capabilities.
    • Allows for generating synthetic queries based on the initial user query to retrieve relevant information.
    • Provides additional configuration options specific to synthetic query generation.
  3. Web RAG (GitHub Repository (opens in a new tab))

    • Integrates with web search clients, such as Serper or Exa to perform web searches and retrieve relevant information.
    • Supports streaming of search results, context, and completion for real-time updates.
    • Readily extensible to combine with your existing data sources and more.