项目描述
ViDoRAG is a novel framework for visual document retrieval and augmented generation using dynamic iterative reasoning agents. It features a multi-agent, actor-critic paradigm for iterative reasoning and a GMM-based multi-modal hybrid retrieval strategy to integrate visual and textual pipelines.
Project Information
Created on 2/25/2025
Updated on 6/29/2025
Categories
ai-content-generation
machine-learning-framework
search-and-retrieval
Tags
open-source-community
data-processing
algorithm-model
multimodal
research-frontier