incidentfox

incidentfox / OpenRag

Public

Multi-strategy RAG system achieving 74% Recall@10 on MultiHop-RAG. Combines RAPTOR hierarchical retrieval, knowledge graphs, HyDE, BM25, and Cohere neural reranking.

ai-sre cohere knowledge-graph llm rag
36
5
100% credibility
Found Feb 04, 2026 at 26 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

OpenRag is an open-source retrieval-augmented generation system that combines hierarchical summaries, knowledge graphs, hybrid search, and neural reranking to achieve state-of-the-art performance on multi-hop question answering benchmarks.

How It Works

1
🔍 Discover Smart Search

You hear about a tool that makes AI search through documents super accurate, beating other methods on tough tests.

2
📥 Get the Tool

Download the ready-to-use search assistant that works right away on your computer.

3
🔗 Connect Your AI Brain

Link a smart AI service so your assistant can understand and reason about information.

4
📄 Feed in Your Info

Upload your documents or articles, and the assistant organizes them into a smart knowledge map.

5
🚀 Launch Your Assistant

Start the search service with one simple action, and it's ready on your local web address.

6
🧪 Test and Query

Ask questions or run built-in tests to see how well it finds answers across connected facts.

✅ Perfect Results

Your assistant delivers spot-on answers with sources, making complex searches feel effortless.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 26 to 36 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is OpenRag?

OpenRag is a Python-based open rag github project delivering a multi-strategy RAG system that achieves 74% Recall@10 on the MultiHop-RAG benchmark. It combines hierarchical retrieval, knowledge graphs, HyDE query expansion, BM25 hybrid search, and Cohere neural reranking into a production-ready FastAPI server. Users ingest document batches via API, build hierarchies on-the-fly, and query in fast, standard, or thorough modes for top-k results.

Why is it gaining traction?

It outperforms RAPTOR's ~70% on multi-hop retrieval through smart fusions like Cohere reranking (+9% lift) and BM25 hybrids, proven in ablations. Developers get a full stack—ingest, persist trees, benchmark scripts, even AWS one-click deploys— at ~$0.0025 per query. Privacy options like local rerankers appeal for sensitive data.

Who should use this?

AI engineers tackling knowledge-intensive QA where single retrievers fail, like multi-hop news analysis or enterprise search. RAG benchmarkers comparing strategies on CRAG or MultiHop-RAG datasets. Teams needing quick API prototypes before scaling to custom LLMs.

Verdict

Grab it for multi-strategy RAG baselines that actually beat SOTA—worth benchmarking your corpus despite 33 stars and 1.0% credibility signaling early days. Docs and AWS scripts ease starts, but expect tweaks for prod.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.