OpenBMB/VisRAG
A parsing-free RAG system that leverages vision-language models for visual document retrieval and multi-image reasoning.
Not currently ranked — collecting fresh signals.
star history
VisRAG 2.0 is a retrieval-augmented generation system designed for visual documents that operates without traditional text parsing. It uses vision-language models to directly retrieve and reason over visual content including images and documents. The system includes specialized retrieval models (VisRAG-Ret) and generation models (EVisRAG), enabling evidence-guided multi-image reasoning for visual question answering and document understanding tasks.
Frequently asked
- What is OpenBMB/VisRAG?
- A parsing-free RAG system that leverages vision-language models for visual document retrieval and multi-image reasoning.
- Is VisRAG open source?
- Yes — OpenBMB/VisRAG is open source, released under the Apache-2.0 license.
- What language is VisRAG written in?
- OpenBMB/VisRAG is primarily written in Python.
- How popular is VisRAG?
- OpenBMB/VisRAG has 972 stars on GitHub.
- Where can I find VisRAG?
- OpenBMB/VisRAG is on GitHub at https://github.com/OpenBMB/VisRAG.