superlinear-ai/raglite

🥤 RAGLite is a Python toolkit for Retrieval-Augmented Generation (RAG) with DuckDB or PostgreSQL

What it solves

RAGLite is a lightweight Python toolkit designed to simplify the implementation of Retrieval-Augmented Generation (RAG) pipelines. It removes the complexity of heavy frameworks like LangChain and PyTorch, providing a streamlined way to integrate LLMs with vector and keyword search databases to ground AI responses in specific documents.

How it works

The toolkit allows users to configure an LLM provider (via LiteLLM) and a database (DuckDB or PostgreSQL). It processes documents by converting them to Markdown, applying semantic chunking, and creating multi-vector embeddings. For retrieval, it supports hybrid search (combining keyword and vector search) and reranking to find the most relevant content. It also features "adaptive retrieval," where the LLM determines if and what to retrieve based on the user's query.

Who it’s for

It is intended for developers who want a fast, permissive, and extensible RAG implementation without the overhead of large ML frameworks, as well as those looking to integrate their knowledge base with MCP clients like Claude Desktop.

Highlights

  • Flexible Backend: Supports DuckDB and PostgreSQL for storage and LiteLLM for LLM/embedding provider flexibility.
  • Advanced Chunking: Implements late chunking, contextual chunk headings, and optimal semantic splitting using binary integer programming.
  • Adaptive RAG: The LLM can autonomously decide when to trigger retrieval based on the prompt.
  • MCP Integration: Includes a built-in Model Context Protocol server to connect with AI clients like Claude Desktop.
  • Optimized Performance: Features prompt caching-aware structures and a closed-form linear query adapter to improve retrieval quality.

Related

  • Project
    Bessouat40/RAGLight

    A lightweight and modular Python library for implementing Retrieval-Augmented Generation (RAG), allowing users to easily integrate various LLMs and vector stores to chat with their documents.

  • Project
    weiwill88/Local_Pdf_Chat_RAG

    A modular Python implementation of a RAG pipeline designed for learning and experimentation, featuring hybrid retrieval and support for multiple LLM backends.

  • Project
    FalkorDB/GraphRAG-SDK

    GraphRAG‑SDK is a Python SDK that builds a knowledge graph in FalkorDB from your documents and then answers questions by traversing that graph. It adds provenance edges, multi‑hop retrieval, and optional abstention to dramatically cut down LLM hallucinations, and it ships with benchmarks showing it outperforms standard vector‑RAG. Install via pip, run a FalkorDB container, ingest texts (or PDFs/markdown), call `finalize()`, and query with `completion()`. The library supports schemas, incremental updates, custom providers, and is Apache‑2.0 licensed.

  • Project
    DataScienceUIBK/Rankify

    A comprehensive Python toolkit for retrieval, re-ranking, and RAG, providing a unified interface to experiment with and deploy state-of-the-art models.

  • Project
    NirDiamant/Controllable-RAG-Agent

    A controllable autonomous RAG agent that uses a deterministic graph and adaptive planning to solve complex, multi-hop questions from custom datasets while preventing hallucinations.