# VisRAG2.0: Mitigating Visual Hallucinations via Evidence-Guided Multi-Image Reasoning in Visual Retrieval-Augmented Generation

arXiv: https://arxiv.org/abs/2510.09733 · HTML: https://mlsanity.com/paper/2510.09733/

## Key facts

- Cited by 10 papers in the map (data through 14 Oct 2026).
- Placed in Multimodal reasoning, one of 17 directions in LLM reasoning.
- Builds on 6 of its 49 references that are in the map.
- First posted on arXiv 10 Oct 2025.

## Placed in

- LLM reasoning / [Multimodal reasoning](https://mlsanity.com/lr/multimodal-reasoning/)
- Multimodal LLMs / vision-language models / [Document understanding and multimodal RAG](https://mlsanity.com/mllm/document-understanding-and-multimodal-rag/)
- Retrieval-augmented generation / [Multimodal RAG](https://mlsanity.com/rag/multimodal-rag/)

## Builds on (most cited first, 6 of 6)

- [MiniCPM-V: A GPT-4V Level MLLM on Your Phone](https://mlsanity.com/paper/2408.01800/) (2024)
- [ColPali: Efficient Document Retrieval with Vision Language Models](https://mlsanity.com/paper/2407.01449/) (2024)
- [Fine-grained Hallucination Detection and Editing for Language Models](https://mlsanity.com/paper/2401.06855/) (2024)
- [MiMo-VL Technical Report](https://mlsanity.com/paper/2506.03569/) (2025)
- [jina-embeddings-v4: Universal Embeddings for Multimodal Multilingual Retrieval](https://mlsanity.com/paper/2506.18902/) (2025)
- [Visual Haystacks: A Vision-Centric Needle-In-A-Haystack Benchmark](https://mlsanity.com/paper/2407.13766/) (2024)

## Built on by (most cited first, 10 of 10)

- [VimRAG: Navigating Massive Visual Context in Retrieval-Augmented Generation via Multimodal Memory Graph](https://mlsanity.com/paper/2602.12735/) (2026)
- [AgenticOCR: Parsing Only What You Need for Efficient Retrieval-Augmented Generation](https://mlsanity.com/paper/2602.24134/) (2026)
- [Lang2Act: Fine-Grained Visual Reasoning through Self-Emergent Linguistic Toolchains](https://mlsanity.com/paper/2602.13235/) (2026)
- [VISOR: Agentic Visual Retrieval-Augmented Generation via Iterative Search and Over-horizon Reasoning](https://mlsanity.com/paper/2604.09508/) (2026)
- [MAGE-RAG: Multigranular Adaptive Graph Evidence for Agentic Multimodal RAG in Long-Document QA](https://mlsanity.com/paper/2606.15906/) (2026)
- [EvoGraph-R1: Self-Evolving Multimodal Knowledge Hypergraphs for Agentic Retrieval](https://mlsanity.com/paper/2607.12764/) (2026)
- [Index Light, Reason Deep: Deferred Visual Ingestion for Visual-Dense Document Question Answering](https://mlsanity.com/paper/2602.14162/) (2026)
- [LITTA: Late-Interaction and Test-Time Alignment for Visually-Grounded Multimodal Retrieval](https://mlsanity.com/paper/2603.26683/) (2026)
- [VLD-RAG: Agentic Vision-Language Retrieval-Augmented Generation for Long, Visually-Rich Multi-Page Documents](https://mlsanity.com/paper/2607.24748/) (2026)
- [NeoMME: A Single-Tower Multimodal-Native Multilingual Foundation Encoder for Efficient Fine-Tuning and Inference](https://mlsanity.com/paper/2609.01657/) (2026)

## Abstract (arXiv)

Visual Retrieval-Augmented Generation (VRAG) has emerged as a promising paradigm for equipping Vision-Language Models (VLMs) with external visual evidence, enabling them to go beyond parametric knowledge when answering visually grounded questions. However, in such multi-image settings, VLMs still often suffer from visual hallucinations and struggle to accurately identify the question-relevant evidence needed for reliable reasoning. Existing methods usually lack an explicit cross-image evidence collection process, and also provide limited credit assignment when jointly optimizing perception and reasoning. To address this issue, we propose EVisRAG, an evidence-guided visual retrieval-augmented framework for multi-image reasoning. EVisRAG first observes the retrieved images, records question-relevant visual evidence from each image, and then performs reasoning and answer generation based on the aggregated evidence. We further introduce RS-GRPO, which aligns reward signals with token spans from different stages, improving training stability and strengthening the joint optimization of evidence localization and reasoning. Experiments on multiple visual question answering benchmarks show that EVisRAG consistently outperforms the backbone VLM by an average of about 19%, while substantially reducing visual hallucinations. These results demonstrate that explicit evidence collection and scoped reward design are effective for improving visual grounding and reasoning reliability in multi-image settings. Codes and data are available at https://github.com/OpenBMB/VisRAG