REAL-MM-RAG: A Real-World Multi-Modal Retrieval Benchmark
Navve Wasserman, Roi Pony, Oshri Naparstek, Adi Raz Goldfarb, Eli, Schwartz, Udi Barzelay, Leonid Karlinsky

TL;DR
REAL-MM-RAG is a new benchmark for multi-modal document retrieval that captures real-world challenges, evaluates model understanding at multiple difficulty levels, and helps improve retrieval performance in RAG systems.
Contribution
The paper introduces REAL-MM-RAG, a comprehensive, realistic benchmark with multi-difficulty queries and new datasets to enhance multi-modal retrieval evaluation and training.
Findings
Models struggle with table-heavy documents and query rephrasing.
Fine-tuning on curated datasets improves retrieval performance.
Benchmark reveals significant weaknesses in current models.
Abstract
Accurate multi-modal document retrieval is crucial for Retrieval-Augmented Generation (RAG), yet existing benchmarks do not fully capture real-world challenges with their current design. We introduce REAL-MM-RAG, an automatically generated benchmark designed to address four key properties essential for real-world retrieval: (i) multi-modal documents, (ii) enhanced difficulty, (iii) Realistic-RAG queries and (iv) accurate labeling. Additionally, we propose a multi-difficulty-level scheme based on query rephrasing to evaluate models' semantic understanding beyond keyword matching. Our benchmark reveals significant model weaknesses, particularly in handling table-heavy documents and robustness to query rephrasing. To mitigate these shortcomings, we curate a rephrased training set and introduce a new finance-focused, table-heavy dataset. Fine-tuning on these datasets enables models to…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Code & Models
Videos
Taxonomy
TopicsAdvanced Image and Video Retrieval Techniques
MethodsRefunds@Expedia|||How do I get a full refund from Expedia? · Attention Is All You Need · Byte Pair Encoding · Adam · Softmax · Dropout · Weight Decay · BART · Linear Layer · WordPiece
