Multi-modal Retrieval Augmented Multi-modal Generation: Datasets, Evaluation Metrics and Strong Baselines
This work addresses the lack of resources and analysis for M^2RAG, enabling better multi-modal content processing and generation, though it is incremental in building upon existing foundation models.
The paper tackles the understudied task of Multi-modal Retrieval Augmented Multi-modal Generation (M^2RAG) by establishing a benchmark with curated datasets and evaluation metrics, and shows that fine-tuned 7B-8B models outperform GPT-4o and approach OpenAI o3-mini in performance.
We present a systematic investigation of Multi-modal Retrieval Augmented Multi-modal Generation (M$^2$RAG), a novel task that enables foundation models to process multi-modal web content and generate multi-modal responses, which exhibits better information density and readability. Despite its potential impact, M$^2$RAG remains understudied, lacking comprehensive analysis and high-quality data resources. To address this gap, we establish a comprehensive benchmark through a rigorous data curation pipeline, and employ text-modal metrics and multi-modal metrics based on foundation models for evaluation. We further propose several strategies for foundation models to process M$^2$RAG task effectively and construct a training set by filtering high-quality samples using our designed metrics. Our extensive experiments demonstrate the reliability of our proposed metrics, a landscape of model performance within our designed strategies, and show that our fine-tuned 7B-8B models outperform the GPT-4o model and approach the state-of-the-art OpenAI o3-mini. Additionally, we perform fine-grained analyses across diverse domains and validate the effectiveness of our designs in data curation pipeline. All resources, including codes, datasets, and model weights, will be publicly released.