CarnegieBin/GlobalRAG

This is the Ofiicial repository for paper: GlobalRAG: Enhancing Global Reasoning in Multi-hop Question Answering via Reinforcement Learning

32
/ 100
Emerging

This helps researchers and developers who work with large language models to answer complex, multi-step questions more accurately. It takes a question and a body of text (corpus) as input, then produces a well-reasoned answer by breaking down the question, retrieving relevant information, and iteratively refining its approach. This is for those building or evaluating advanced AI systems that need to perform sophisticated information retrieval and reasoning.

Use this if you are developing or studying AI systems that need to answer multi-hop questions by reasoning across multiple pieces of information, and you want to improve reasoning accuracy and planning capabilities.

Not ideal if you're looking for an off-the-shelf chatbot or a simple single-query search tool.

AI-research natural-language-processing question-answering information-retrieval reasoning-systems
No Package No Dependents
Maintenance 6 / 25
Adoption 5 / 25
Maturity 13 / 25
Community 8 / 25

How are scores calculated?

Stars

9

Forks

1

Language

Python

License

Apache-2.0

Last pushed

Nov 04, 2025

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/rag/CarnegieBin/GlobalRAG"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.