OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
RAG & Knowledge Systems

Rag Fusion

Rag Fusion is a Python project that demonstrates a retrieval-augmented generation search method: it generates multiple query variants and combines the resulting ranked search lists with Reciprocal Rank Fusion. Its repository also includes an evaluation harness using NFCorpus from BEIR.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Rag Fusion — OpenViST project profile in RAG & Knowledge Systems.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Consider the method for exploratory or specialist-corpus search when terminology mismatch is common and recall matters more than minimizing latency. The repository cautions that multiple rewrites add latency and that the technique is a weaker fit for latency-critical or precision-dominated retrieval.

Who it is for

  • Developers building or studying retrieval-augmented generation systems.
  • Practitioners searching corpora where user terminology may differ from indexed text.
  • Researchers comparing retrieval strategies on a judged dataset.

Documented uses

  • Generate alternate query phrasings to broaden retrieval.
  • Merge ranked results from multiple searches using Reciprocal Rank Fusion.
  • Compare retrieval approaches with an NFCorpus/BEIR evaluation harness.
Before you choose

Requirements & limits

The repository is public and licensed under MIT. The documented demo setup requires an OpenAI API key.

  • The core demo uses vector search and RRF; hybrid BM25-plus-vector search and reranking are covered in the evaluation experiments.
  • The README reports retrieval gains for tested hybrid-plus-rerank configurations, but generated-answer results are described as less consistent.
  • Query rewriting adds latency; the README gives about 1.5 seconds for the rewrite call with a small model.
  • The repository's benchmark uses NFCorpus, a medical and nutrition dataset, so its reported measurements do not by themselves establish results for other corpora.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Set up and run the demo

Install the listed dependencies, configure the OpenAI key, and launch the repository's demo.

Before you start

  • Python environment
  • OpenAI API key
  • Dependencies listed in the repository README

Steps

  1. Install the listed packages with pip: openai, chromadb, python-dotenv, tqdm, tabulate and rank_bm25.

  2. Copy .env.example to .env and replace the placeholder with an OpenAI API key.

  3. Run python main.py.

Expected outcome

The repository's RAG-Fusion demo runs using its configured model and dependencies.

Guide limitations

  • The README says the default LLM is gpt-6-luna and supports model and reasoning-effort overrides through environment variables.
  • JEV_API_KEY is needed only for Jev experiments, not the basic demo.
Documentation 1 ↗
COMPARE OPTIONS

Compare options for Retrieval-augmented applications

These projects document a common task and share a project role. This is a source-based comparison, not an execution benchmark or a guarantee of interchangeability.

ProjectDocumented purposePlatformAccess and requirements
Rag FusionGenerate alternate query phrasings to broaden retrieval.CLI, Self-hostedThe repository is public and licensed under MIT. The documented demo setup requires an OpenAI API key.
LangsearchCrawl website or local data for a knowledge base.CLI, Self-hosted, Server
Hybrid Search RagCombine document retrieval and language-model generation in a configurable workflow.
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
9 October 2026
Original website title
Raudaschl/rag-fusion · Official homepage title · retained in its source language
How we review sourcesSuggest a correction