RAG Me Up is a generic, configurable framework for applying Retrieval-Augmented Generation (RAG) to custom datasets. It provides a lightweight server and multiple UI options, targeting developers and researchers who need to easily integrate LLMs with their own data. The framework aims to simplify RAG pipeline setup and customization without extensive coding.

How It Works

RAG Me Up employs a hybrid search strategy, combining dense vector retrieval (via Milvus or PostgreSQL with pgvector) and sparse BM25 retrieval for enhanced accuracy. Documents are loaded, chunked, and embedded. During inference, it can optionally use an LLM to determine if new documents need fetching, rerank retrieved documents using cross-encoders, and even use the LLM to assess answer quality or rewrite queries for better retrieval. Provenance attribution is supported through reranking, attention weights (for OS LLMs), similarity scores, or LLM-based scoring.

Quick Start & Requirements

• Server: cd server && pip install -r requirements.txt && python server.py
• UI: Download and run compiled binary or sbt run from server/scala.
• Prerequisites: JDK 17+ for Scala UI. GPU with 16GB+ VRAM recommended for default instruct model. Ollama compatibility is available.
• PostgreSQL: Requires manual index creation for pgvector after initial setup.
• Docs: README

Highlighted Details

• Supports multiple LLM backends (LLaMa3, Ollama, OpenAI, Gemini, Azure OpenAI).
• Offers various provenance attribution methods (rerank, attention, similarity, LLM).
• Includes evaluation frameworks like RAGAS and DeepEval.
• Supports hybrid search with Milvus and PostgreSQL (pgvector/pg_search).
• Features query rewriting (self-inflection) and Re2 (re-reading) capabilities.

Maintenance & Community

The project is actively updated, with recent additions including evaluation frameworks, Ollama compatibility, and improved Postgres hybrid search. Links to community channels are not explicitly provided in the README.

Licensing & Compatibility

The README does not specify a license. Compatibility for commercial use or closed-source linking is not detailed.

Limitations & Caveats

The README does not specify a license, which may impact commercial adoption. Configuration for specific document types like JSON and XML requires detailed schema or XPath definitions. Provenance attribution methods have varying levels of LLM dependency and OS LLM requirements.