MultAttnAttrib: Training-Free Multimodal Attribution in Long Document Question Answering
Quick Answer
MultAttnAttrib introduces a training-free method for multimodal attribution in long document QA, outperforming existing methods and matching GPT 5.4.
Quick Take
It enhances attribution accuracy significantly while reducing inference latency to one-seventh of prompting methods. The complementary benchmark dataset, MultAttrEval, provides fine-grained attributions for evaluation.
Key Points
- MultAttnAttrib leverages prefill passes and attention heads for evidence attribution.
- Outperforms various attribution methods, including strong prompting-based approaches.
- Matches performance of advanced models like GPT 5.4 in attribution accuracy.
- Reduces inference latency to one-seventh compared to traditional prompting.
- Introduces MultAttrEval, the first dataset for multimodal attribution in long documents.
Paper Resources
📖 Reader Mode
~2 min readAbstract:As grounded QA systems are increasingly deployed in AI assistants, accurately attributing generated answers to evidence is critical for user trust and model safety. While unimodal attributions have been explored in depth, the multimodal setting remains relatively under-researched. As a result, we introduce MultAttnAttrib, a training-free attribution-generation method that leverages a model's prefill pass, selected attention heads, and calibrated thresholds to locate source evidence within a document. To establish baseline results for the method, we introduce MultAttrEval, a complementary benchmark dataset annotated with fine-grained, ground-truth attributions for answer components grounded in multimodal source documents. To our knowledge, this is the first evaluation dataset designed specifically for multimodal attribution in long-form documents. Experimental results show that MultAttnAttrib consistently outperforms a variety of attribution-generation methods, including several strong prompting-based approaches and matches the latest frontier models such as GPT 5.4. Our method not only substantially improves attribution accuracy for both unimodal and multimodal attribution types, but also produces attributions at up to one-seventh of the direct inference latency compared to prompting on the same base model.
| Comments: | 25 pages (8 main, 17 references + appendix), 15 figures, Submitted to EMNLP 2026 Conference (Long Paper) |
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV) |
| Cite as: | arXiv:2607.01420 [cs.CL] |
| (or arXiv:2607.01420v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2607.01420 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Vinamra Tyagi [view email]
[v1]
Wed, 1 Jul 2026 19:29:19 UTC (481 KB)
— Originally published at arxiv.org
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.CL
See more →TriAgent: Divergence-Aware Committees for Cost-Efficient Financial Sentiment Analysis
TriAgent introduces a cost-efficient multi-agent system for financial sentiment analysis, combining VADER, FinBERT, and Qwen2.5. It achieves an F1 score of ~0.87 with significant savings of $9.3M/year at a 10M-user scale compared to GPT-4o-mini, while also detecting hallucinations with an AUC of 0.90.