Pith. sign in

REVIEW 10 cited by

MM-R5: MultiModal Reasoning-Enhanced ReRanker via Reinforcement Learning for Document Retrieval

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2506.12364 v2 pith:Y2TAAOIB submitted 2025-06-14 cs.AI cs.CLcs.CV

MM-R5: MultiModal Reasoning-Enhanced ReRanker via Reinforcement Learning for Document Retrieval

classification cs.AI cs.CLcs.CV
keywords mm-r5multimodalreasoningretrievaldocumentlearningreasoning-enhancedreinforcement
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Multimodal document retrieval systems enable information access across text, images, and layouts, benefiting various domains like document-based question answering, report analysis, and interactive content summarization. Rerankers improve retrieval precision by reordering retrieved candidates. However, current multimodal reranking methods remain underexplored, with significant room for improvement in both training strategies and overall effectiveness. Moreover, the lack of explicit reasoning makes it difficult to analyze and optimize these methods further. In this paper, We propose MM-R5, a MultiModal Reasoning-Enhanced ReRanker via Reinforcement Learning for Document Retrieval, aiming to provide a more effective and reliable solution for multimodal reranking tasks. MM-R5 is trained in two stages: supervised fine-tuning (SFT) and reinforcement learning (RL). In the SFT stage, we focus on improving instruction-following and guiding the model to generate complete and high-quality reasoning chains. To support this, we introduce a novel data construction strategy that produces rich, high-quality reasoning data. In the RL stage, we design a task-specific reward framework, including a reranking reward tailored for multimodal candidates and a composite template-based reward to further refine reasoning quality. We conduct extensive experiments on MMDocIR, a challenging public benchmark spanning multiple domains. MM-R5 achieves state-of-the-art performance on most metrics and delivers comparable results to much larger models on the remaining ones. Moreover, compared to the best retrieval-only method, MM-R5 improves recall@1 by over 4%. These results validate the effectiveness of our reasoning-enhanced training pipeline. Our code is available at https://github.com/i2vec/MM-R5 .

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 10 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Very Efficient Listwise Multimodal Reranking for Long Documents

    cs.IR 2026-05 unverdicted novelty 7.0

    ZipRerank delivers state-of-the-art multimodal listwise reranking accuracy for long documents at up to 10x lower latency via early interaction and single-pass scoring.

  2. Purifying Multimodal Retrieval: Fragment-Level Evidence Selection for RAG

    cs.IR 2026-04 unverdicted novelty 7.0

    FES-RAG reframes multimodal RAG as fragment-level selection using Fragment Information Gain to outperform document-level methods with up to 27% relative CIDEr gains on M2RAG while shortening context.

  3. ELVA: Exploring Ranking-Driven Universal Multimodal Retrieval

    cs.IR 2026-06 unverdicted novelty 6.0

    ELVA applies ranking-driven RLVR to multimodal retrieval to reduce grain blindness in contrastive learning, reporting SOTA results and a 13.1% gain on the new MRBench benchmark.

  4. miniReranker: Efficient Multimodal Reranking through Visual Cache Reuse and Interaction Sparsity

    cs.IR 2026-06 unverdicted novelty 6.0

    miniReranker reduces multimodal reranking runtime to under 1% of the dense baseline under high-reuse conditions while retaining over 96% of performance via vision-first prompting, early exit, sparse cross-segment atte...

  5. MEG-RAG: Quantifying Multi-modal Evidence Grounding for Evidence Selection in RAG

    cs.CL 2026-04 unverdicted novelty 6.0

    MEG-RAG defines a new MEG metric based on Semantic Certainty Anchoring and trains a multimodal reranker to select evidence aligned with ground-truth semantic anchors, yielding higher accuracy and consistency on the M²...

  6. GR2 Technical Report

    cs.IR 2026-06 unverdicted novelty 5.0

    GR2 applies mid-training on semantic IDs, reasoning distillation, RL with conditional verifiable rewards, and a context compressor to re-ranking in industrial recsys, reporting +18.7% R@1 over baselines.

  7. ELVA: Exploring Ranking-Driven Universal Multimodal Retrieval

    cs.IR 2026-06 unverdicted novelty 5.0

    ELVA uses rule-based RL rewards to rank negatives by similarity, reducing grain blindness in universal multimodal retrieval and reporting a 13.1% gain on a new multi-grain benchmark.

  8. DocRetriever: A Plug-and-Play Framework for Multimodal Document Retrieval with Comprehensive Benchmark

    cs.CV 2026-05 unverdicted novelty 5.0

    DocRetriever introduces a framework using layout-aware sparse embeddings for hybrid encoding without OCR and a generalizable reasoning-augmented reranker for few-shot settings, plus the MultiDocR benchmark for evaluation.

  9. UniRank: End-to-End Domain-Specific Reranking of Hybrid Text-Image Candidates

    cs.IR 2026-02 conditional novelty 5.0

    UniRank natively scores hybrid text-image candidates with a VLM, then SFT plus hard-negative RLHF adapts it to a domain; it outperforms baselines on scientific literature and design patent retrieval.

  10. Rich-Media Re-Ranker: A User Satisfaction-Driven LLM Re-ranking Framework for Rich-Media Search

    cs.IR 2026-02 unverdicted novelty 5.0

    A re-ranking system for rich-media search that plans query intents from sessions, adds visual signals from VLMs, and uses an LLM to score results on multiple facets before multi-task RL adaptation, with reported gains...