Skip to content

MiniRAG

Run RAG on small language models with graph indexing

Visit official site

What minirag is

The framework is described as a retrieval augmented generation system for small language models, built on heterogeneous graph indexing and lightweight topology-enhanced retrieval.

What you can do with it

  • Clone the repository to a local machine
  • Install from source or through PyPI
  • Index the dataset with the provided script

What to watch out for

  • The LiHua-World dataset is downloaded separately
  • Resource-constrained devices are the stated target
  • Open source code is available to the community

Pros & cons

✓ What we like

  • Heterogeneous graphs combine text chunks and named entities
  • Install from source or via PyPI

! What to watch out for

  • The repository is cloned before installation
  • The dataset is downloaded and placed in a set folder

FAQ

How much storage does it need?

The material states only 25 percent of the storage is required.

What dataset is provided?

The material names LiHua-World for evaluating lightweight RAG on real devices.

Last reviewed: 2026-11-15

More AI coding tools tools

View all →

How we review