Pith. sign in

REVIEW 3 major objections 5 minor 163 references

Locating Failure in Multi-Page Visually Rich Document Understanding: An Empirical Attribution

T0 review · 3 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper attributes multi-page document QA failures to three loci and shows that reasoners fail to integrate evidence across pages even when every required page is supplied.

desk verdict A careful, well-controlled attribution study whose headline reasoning claim rests on an unmatched question-pool comparison; the fix is achievable and the rest holds up. read the letter →

arxiv 2608.07943 v1 pith:F7MCXVZR submitted 2026-08-08 cs.AI

Lewei Xu , Yihao Ding , Zihan Xu , Daniel Yitian Su , Daochang Liu , Siwen Luo , Yifan Peng , Wei Liu This is my paper · ORCID
classification cs.AI
keywords multi-pagedocumentunderstandingvisuallyrichdocumentsfailureattributionevidenceselectioncross-pagereasoningchain-of-thoughtpromptingretrievalrecallquestionanswering
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper tries to locate where multi-page visually rich document understanding fails by attributing incorrect answers to three stages: representation (does the encoding preserve the evidence?), selection (is the right evidence delivered?), and reasoning (can the model use it?). Using a controlled intervention design on the MMLongBench-Doc benchmark, it finds that vision is necessary but does not replace text extraction, that omitting one required page sharply caps accuracy while adding distractor pages costs little, and that reasoners fail to integrate evidence across pages even when every required page is supplied. The cross-page integration drop is not explained by context length, because chain-of-thought prompting narrows it while increasing model scale does not. The paper converts these findings into concrete design guidance for building MP-VRDU systems under a fixed compute budget.

What carries the argument

The central object is the three-locus attribution framework, which decomposes an incorrect answer into representation, selection, and reasoning, each with two mechanisms: modality ceiling and conversion fidelity for representation; evidence coverage and distractor exposure for selection; evidence integration and response calibration for reasoning. The framework is made experimental by 'attribution by construction': each intervention changes only the condition a failure mode depends on while all other stages are held fixed. The load-bearing comparison is the single-page versus multi-page accuracy contrast at the same encoding, which isolates cross-page integration from mere input length.

What would settle it

Take a set of questions whose evidence chain can be placed either on one page or split across two pages while keeping topic, answer type, and judged difficulty matched; if accuracy does not drop noticeably in the two-page condition under the combined encoding, the claimed cross-page integration failure is not confirmed.

Watch

Extended reading notes

Core claim

The paper's central claim is that incorrect answers in multi-page visually rich document understanding can be attributed to three hierarchical failure modes—representation, selection, and reasoning—and that once representation and selection are controlled, the dominant remaining failure is cross-page evidence integration. On MMLongBench-Doc, the default reasoner at the combined text-layout-vision encoding answers 64.6% of single-page questions but only 38.6% of two-page questions, a pooled multi-page deficit of 25.9 points despite the context being well within the model's window. Chain-of-thought prompting narrows the deficit to 16.5 points, whereas scaling the reasoner from 2B to 32B parameters leaves it essentially unchanged. The paper reads this as evidence that current reasoners fail to assemble evidence across pages even when every required page is fully supplied—a reasoning failure, not a capacity or selection failure.

Load-bearing premise

The comparison between single-page and multi-page questions assumes the two question pools are otherwise comparable, so the accuracy gap measures integration ability rather than intrinsic difficulty; if multi-page questions are simply harder, the deficit is partly a task-difficulty effect.

Editorial extensions

If this is right

  • Retrieval should be tuned for recall over precision: removing one required gold page drops multi-page accuracy from 38.6 to about 18.5 at the combined encoding, while adding three distractor pages costs only a few points.
  • The combined text-layout-vision encoding is the safe default, but routing representation by document class can match vision-only accuracy at less than half the input tokens, saving compute where questions do not need images.
  • Chain-of-thought prompting should be applied selectively to multi-page questions: it narrows the integration deficit by roughly 8–9 points while leaving single-page accuracy near baseline.
  • Under a fixed memory budget, quantization is the first thing to trade for parameters: four-bit weights cost about 0.6 accuracy points at the combined encoding while shrinking the reasoner footprint by 61%.
  • Abstention instructions should be enabled when a wrong answer costs more than no answer, because they raise correct refusal on unanswerable questions to about three-quarters while adding false refusals on answerable ones.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the cross-page deficit is truly a reasoning-stage failure, then training objectives that force explicit multi-page evidence assembly—for example, requiring the model to state which page contributes which fact—may close a gap that prompting and scale only partially address; the paper itself notes that multi-page integration is rarely a training target.
  • The deficit widening as representation improves (from 11.7 points on text to 25.9 on text-layout-vision) suggests a visual-integration bottleneck specifically; a testable extension would use interleaved per-page text-image ordering or visual grounding supervision to see whether the image-based gap shrinks.
  • Because the paper does not demonstrate that single-page and multi-page question pools are matched in difficulty, an exact attribution would require rewriting the same evidence chain into single-page and multi-page variants; such a control could revise the magnitude of the reasoning deficit.
  • The distractor-tolerance result is measured at modest context lengths; extending retrieval depth or moving to learned rankers could reverse the asymmetry if long enough contexts eventually impair evidence use, which the paper acknowledges as a limit.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper proposes an attribution framework for multi-page visually rich document understanding (MP-VRDU) that decomposes incorrect answers into three failure modes—representation, selection, and reasoning—each with two mechanisms, and instantiates the framework in a controlled single-pass retrieval-generation pipeline evaluated on MMLongBench-Doc. The empirical study intervenes on each locus while holding others fixed: representation is varied across text, layout, and vision encodings; selection is controlled by withholding gold pages or adding distractor pages; reasoning is probed by supplying complete gold evidence and varying prompting, scale, and model family. The main findings are that vision is necessary but does not replace text extraction, that missing evidence sharply bounds accuracy while distractor pages are largely tolerated, and that reasoners fail to integrate evidence across pages even when all required pages are supplied. The paper further reports that chain-of-thought prompting narrows the cross-page deficit more than increasing model scale does, and derives design guidance for building MP-VRDU systems under fixed compute budgets.

Significance. If the findings hold, the paper makes a useful contribution: it offers a common vocabulary for MP-VRDU failures, a controlled experimental design that separates selection from reasoning, and a concrete, falsifiable claim that cross-page evidence integration—not context length or evidence selection—is a major remaining bottleneck. The paper ships a careful experimental apparatus: oracle gold pages, per-question verdict transitions that distinguish systematic loss from churn, document-level bootstrap confidence intervals, and a rich set of worked examples. The practical guidance on routing, recall-oriented retrieval, quantization, and abstention is grounded in measured trade-offs. The central empirical claim, however, rests on comparing naturally occurring single-page and multi-page question pools, and that comparability is not established; this is the key risk to the paper's main conclusion and to its training recommendation.

major comments (3)
  1. [§5.3, Table 3] The central claim that the reasoner 'fails to integrate evidence across pages' is supported by comparing accuracy on single-page questions (TLV: 64.6) with accuracy on multi-page questions (TLV: 38.6). This comparison assumes the two question pools are comparable in reasoning difficulty aside from the page spread of the evidence. No matching, difficulty rating, or single-page control with the same number of evidence pieces is reported. The distractor experiment in §5.2 (Figure 3b) rules out context length, but its single-gold-page condition does not reproduce the multi-evidence composition of multi-page questions. Chain-of-thought narrowing the gap (38.6 to 44.9) and the 32B result leaving the deficit at −25.1 are both consistent with multi-page questions being intrinsically harder (requiring combination of more evidence units), independent of page boundaries. The manuscript should provide a control that removes the page boundary while keeping the evidence set unchanged, or otherwise demonstrate that the single-page and multi-page pools are matched on evidence-unit count and type. Without this, the conclusion that the loss is specifically a cross-page reasoning failure is not uniquely supported.
  2. [§5.3, Table 3 and §D.1, Table 13] The claim that a stronger representation 'enlarges the gap' and 'points to a visual reasoning failure specifically' relies on the same pool-comparability assumption, now made sharper: the deficit grows from −11.7 at T to −25.9 at TLV. Table 13 shows that the deficit varies wildly by domain, from +8.5 (Academic paper) to −54.1 (Financial report), and by evidence source, from −0.4 (Chart at T) to −34.5 (Table at TLV). This variation suggests that the composition of the single-page and multi-page pools strongly determines the measured gap. If multi-page questions draw disproportionately on evidence sources or domains where the image helps less, or where questions are intrinsically harder and hence less responsive to better inputs, the widening gap need not indicate a visual integration failure. The manuscript should demonstrate that the single-page and multi-page pools are balanced on evidence source, domain, and evidence-unit count before attributing the widening to a modality-specific integration deficit.
  3. [§5.2 and §4.1] The 'missing evidence bounds accuracy' result is presented as an empirical finding, but it follows largely by construction: the gold-page annotation defines the withheld evidence as required, so removing it removes information the question needs. The paper partially acknowledges this framing, but the practical recommendation to 'favour recall' would be more convincing if the coverage mechanism had a quantitative baseline—for example, an estimate of how often the remaining pages already contain enough evidence to answer the question, or a judged upper bound on accuracy given only the remaining pages. As reported, the near-collapse in accuracy when one gold page is withheld (Figure 3a) confirms the annotation, but does not by itself tell us how much of the drop is due to evidential necessity versus the reasoner's inability to use partial evidence. A control that presents the non-gold pages as the context, or scores the question with the withheld page's content paraphrased into another page, would strengthen the attribution without changing the framework.
minor comments (5)
  1. [§4.3 and Appendix A.8] The main text states that Gemini 2.5 Flash is used as the judge, while Appendix A.8 describes two judges (GPT-4o-mini and Gemini-2.5-flash) run at temperature 0. Please clarify which judge produced the main-text results and whether the two judges were combined or compared for agreement.
  2. [§4.3 and Tables 2–5] The paper states that 95% confidence intervals are computed from a document-level bootstrap, but the main-text tables do not display intervals. Adding intervals to the key tables, or explicitly referencing the appendix where they appear, would help readers assess the stability of the reported gaps, especially the 25.9-point deficit in Table 3.
  3. [Figure 3 caption] The caption reports 'oracle 38.6, n=352' for the withholding panel and 'oracle 64.6, n=474' and 'oracle 38.6, n=241' for the distractor panel, but the relationship of these n values to the Table 1 counts (480 single-page, 246 two-page, 112 three-plus-page) is not explained on first reading. Please clarify whether the n values refer to the answerable subset used in each experiment and why the oracle for the two-page condition appears in both panels.
  4. [§6.1 and Appendix A.2] The representation 'TV' is recommended in the main text ('TV is the cheaper choice on born-digital pages') but is not defined until Appendix A.2. Define TV and TLVi at first use in §4.2 so the discussion in §6 is self-contained.
  5. [Appendix F, worked example M3] The worked example M3 is labelled 'PROVISIONAL: partial G2 pool.' If any example is provisional, either complete it and remove the label or omit it from the final version; a provisional example undermines the otherwise strong illustration of the recall trade-off.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: empirical findings are measured against an external benchmark; only a framework-internal definition and minor self-citations, neither load-bearing.

full rationale

The central attribution is not circular because it is not derived from fitted parameters or from the authors' own prior claims. The key numbers (representation ladder: T 31.9 vs TLV 52.5; selection: oracle 38.6 dropping to 18.5 after one withheld gold page, distractor additions at 63.1-64.6; reasoning: single-page 64.6 vs two-page 38.6 at TLV, CoT raising multi-page to 44.9) are measured on MMLongBench-Doc with a fixed Qwen3-VL pipeline and an external judge. No fitted coefficient is later relabelled as a prediction. The only definitional components are the framework axioms themselves: 'Evidence Coverage' is defined as the recall of required evidence units and is said to 'plac[e] an upper bound on downstream accuracy,' so the subsequent demonstration that withholding gold pages lowers accuracy is partly a consequence of the framework's own definition rather than a novel empirical discovery. Similarly, 'Evidence Integration' is operationalized as a single-page vs multi-page accuracy contrast, so the label 'cross-page integration failure' is a restatement of the observed gap; however, the gap itself (25.9 points) is an external measurement and could in principle have been absent. Self-citations to Xu et al. 2026 and Ding et al. 2025 support background statements about survey coverage and training targets; they do not carry any measured result. The single-page vs multi-page pools are not matched for intrinsic difficulty, which is a genuine validity threat to the reasoning-stage interpretation, but that is a confound, not circular reasoning. No load-bearing circular step was found.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

The paper introduces no new fitted parameters or invented physical entities. Its free parameters are experimental configuration choices, such as render resolution, retrieval depth, and prompt wording, which are not fitted to data. The main domain assumptions concern the completeness of the benchmark annotations and the comparability of single-page and multi-page question pools.

assumptions (5)
  • domain assumption MMLongBench-Doc gold evidence page annotations are complete and correct.
    All selection and reasoning interventions use annotated gold pages as ground truth (Section 4.3, Appendix A.7). If annotations miss evidence, the coverage and integration conclusions could shift.
  • domain assumption The three failure modes and two mechanisms per mode are sufficient to attribute answer failures.
    The framework in Section 3 constructs interventions around these mechanisms. Other error sources, such as prompt biases, decoding randomness, and judge noise, are held fixed but are not explicitly modeled.
  • domain assumption Single-page and multi-page question pools are comparable apart from evidence spread.
    The integration deficit in Section 5.3 compares naturally occurring question groups without matching or counterbalancing for intrinsic difficulty, which could confound the attributed reasoning failure.
  • domain assumption LLM judge scoring approximates semantic correctness.
    Accuracy is judge-scored by Gemini 2.5 Flash in the main text and a second judge is mentioned in Appendix A.8, but only one judge drives the reported main numbers. Judge errors would propagate to all accuracy figures.
  • domain assumption Page-level granularity matches retrieval and evidence units.
    Retrieval and oracle selection both operate on whole pages (Section 4.2, A.5). Sub-page evidence interactions and partial-page evidence are not measured.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Locating Failure in Multi-Page Visually Rich Document Understanding: An Empirical Attribution." pith.science (2026). https://pith.science/paper/F7MCXVZR

@misc{pith2026260807943,
  author       = {Pith},
  title        = {Pith review of: Locating Failure in Multi-Page Visually Rich Document Understanding: An Empirical Attribution},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/F7MCXVZR}},
  note         = {Machine review of arXiv:2608.07943}
}
read the original abstract

Multi-page visually-rich document understanding (MP-VRDU) requires managing evidence that is sparse, spread across pages, and often exceeds a model's context window. Prior work has produced competing, largely untested claims about how these systems should be built. We attribute incorrect answers to three failure modes, representation, selection, and reasoning, and isolate each over a multi-page document understanding dataset by intervening on one while holding the others fixed. We find that vision is necessary but does not replace text extraction, that missing pages bound accuracy while distractors cost little, and that reasoners fail to integrate evidence across pages even when it is fully supplied. Prompting can shift reasoning behaviour substantially, improving some outcomes at the expense of others. We translate these findings into guidance for building such systems under a fixed compute budget.

Figures

Figures reproduced from arXiv: 2608.07943 by the authors.

Figure 1
Figure 1. The three failure modes and their mechanisms. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Parser accuracy by text source, text-only vs. [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Selection at the TLV encoding, with page rank [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Calibration trade-off at TLV representation. [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: Average input tokens and accuracy by repre [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

163 extracted references · 37 canonical work pages

  1. [1]

    and Vandierendonck, Hans and John, Deepu and Ji, Bo , booktitle =

    Arif, Kazi Hasan Ibn and Yoon, JinYi and Nikolopoulos, Dimitrios S. and Vandierendonck, Hans and John, Deepu and Ji, Bo , booktitle =. 2025 , doi =. 2408.10945 , archivePrefix=

    arXiv 2025

  2. [2]

    Computer Vision -- ECCV 2022 , series =

    Reliable Visual Question Answering: Abstain Rather Than Answer Incorrectly , author =. Computer Vision -- ECCV 2022 , series =. 2022 , publisher =. doi:10.1007/978-3-031-20059-5\_9 , eprint =

    doi:10.1007/978-3-031-20059-5 2022

  3. [3]

    2026 , eprint=

    Unlimited OCR Works , author=. 2026 , eprint=

    2026

  4. [4]

    arXiv preprint arXiv:2504.10479 , year=

    InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Models , author=. arXiv preprint arXiv:2504.10479 , year=

    arXiv

  5. [7]

    International conference on machine learning , pages=

    Large language models can be easily distracted by irrelevant context , author=. International conference on machine learning , pages=. 2023 , organization=

    2023

  6. [8]

    When Good OCR Is Not Enough: Benchmarking OCR Robustness for Retrieval-Augmented Generation

    Sun, Lin and Wang, Dexian and Huang, Jingang and Zhang, Linglin and Jia, Change and Cheng, Zhengwei and Zhang, Xiangzheng , year =. When Good. 2605.00911 , archivePrefix=

    work page Pith review arXiv

  7. [9]

    Text or Pixels? It Takes Half: On the Token Efficiency of Visual Text Inputs in Multimodal

    Li, Yanhong and Lan, Zixuan and Zhou, Jiawei , booktitle =. Text or Pixels? It Takes Half: On the Token Efficiency of Visual Text Inputs in Multimodal. 2025 , eprint =

    2025

  8. [10]

    2026 , eprint =

    Hu, Lianyu and Ma, Xiaoyu and Liao, Zeqin and Liu, Yang , booktitle =. 2026 , eprint =

    2026

Show all 163 references
  1. [11]

    2507.17262 , archivePrefix=

    Saadat, Asir and Aziz, Syem and Mahmud, Shahriar and Mahi, Abdullah Ibne Masud and Ahmed, Sabbir , year =. 2507.17262 , archivePrefix=

    arXiv

  2. [12]

    arXiv preprint arXiv:2606.21553 , year=

    Dissecting Agentic RAG: A Component Ablation for Multi-Hop QA with a Local 7B Model , author=. arXiv preprint arXiv:2606.21553 , year=

    arXiv

  3. [13]

    arXiv preprint arXiv:2606.28337 , year=

    A Systems-Level Analysis of Sensitivity, Robustness, and Stability in Retrieval-Augmented Generation , author=. arXiv preprint arXiv:2606.28337 , year=

    arXiv

  4. [14]

    arXiv preprint arXiv:2605.08888 , year=

    DocScope: Benchmarking Verifiable Reasoning for Trustworthy Long-Document Understanding , author=. arXiv preprint arXiv:2605.08888 , year=

    arXiv

  5. [15]

    Managing Evidence at Document Scale: A Survey of Multi-Page Visually Rich Document Understanding , author=

  6. [16]

    2025 , url =

    Cui, Cheng and Sun, Ting and Liang, Suyin and Gao, Tingquan and Zhang, Zelun and Liu, Jiaxuan and Wang, Xueqing and Zhou, Changda and Liu, Hongen and Lin, Manhui and Zhang, Yue and Zhang, Yubo and Zheng, Handong and Zhang, Jing and Zhang, Jun and Liu, Yi and Yu, Dianhai and Ma...

    2025

  7. [17]

    2025 , url =

    Niu, Junbo and Liu, Zheng and Gu, Zhuangcheng and Wang, Bin and Ouyang, Linke and Zhao, Zhiyuan and Chu, Tao and He, Tianyao and Wu, Fan and Zhang, Qintong and Jin, Zhenjiang and others , journal =. 2025 , url =

    2025

  8. [18]

    2025 , url =

    Bai, Shuai and Cai, Yuxuan and Chen, Ruizhe and Chen, Keqin and Chen, Xionghui and Cheng, Zesen and Deng, Lianghao and Ding, Wei and Gao, Chang and Ge, Chunjiang and others , journal =. 2025 , url =

    2025

  9. [19]

    The Reasoning Bottleneck in

    Zarrinkia, Yasaman and Srinivasan, Venkatesh and Thomo, Alex , journal =. The Reasoning Bottleneck in. 2026 , eprint =

    2026

  10. [20]

    2026 , eprint =

    Sun, Jiashuo and Xie, Yixuan and Shi, Jimeng and Wang, Shaowen and Han, Jiawei , journal =. 2026 , eprint =

    2026

  11. [21]

    Facet-Level Tracing of Evidence Uncertainty and Hallucination in

    Elchafei, Passant and Swain, Monorama and Masoudian, Shahed and Schedl, Markus , journal =. Facet-Level Tracing of Evidence Uncertainty and Hallucination in. 2026 , eprint =

    2026

  12. [22]

    arXiv preprint arXiv:2605.14192 , year =

    Why Retrieval-Augmented Generation Fails: A Graph Perspective , author =. arXiv preprint arXiv:2605.14192 , year =. 2605.14192 , archivePrefix=

    arXiv

  13. [23]

    2026 , eprint =

    Li, Xingyu and Wang, Rongguang and Wang, Yuying and Guo, Mengqing and Li, Chenyang and Sheng, Tao and Ravi, Sujith and Roth, Dan , journal =. 2026 , eprint =

    2026

  14. [24]

    Most, Alexander and Jones, Shawn and Winjum, Joseph and Ranasinghe, Nishath Rajiv and Biswas, Ayan , journal =. Lost in. 2025 , eprint =

    2025

  15. [25]

    arXiv preprint arXiv:2512.02660 , year =

    Spatially-Grounded Document Retrieval via Patch-to-Region Relevance Propagation , author =. arXiv preprint arXiv:2512.02660 , year =. 2512.02660 , archivePrefix=

  16. [26]

    2025 , eprint =

    Zhang, Junyuan and Zhang, Qintong and Wang, Bin and Ouyang, Linke and Wen, Zichen and Li, Ying and Chow, Ka-Ho and He, Conghui and Zhang, Wentao , booktitle =. 2025 , eprint =

    2025

  17. [27]

    Assessing the Impact of

    van Strien, Daniel and Beelen, Kaspar and Coll Ardanuy, Mariona and Hosseini, Kasra and McGillivray, Barbara and Colavizza, Giovanni , booktitle =. Assessing the Impact of. 2020 , publisher =

    2020

  18. [28]

    arXiv preprint arXiv:2311.03731 , year =

    A Survey of Large Language Models Attribution , author =. arXiv preprint arXiv:2311.03731 , year =. 2311.03731 , archivePrefix=

    arXiv

  19. [29]

    2024 , eprint =

    Wang, Changyuan and Wang, Ziwei and Xu, Xiuwei and Tang, Yansong and Zhou, Jie and Lu, Jiwen , booktitle =. 2024 , eprint =

    2024

  20. [30]

    2025 , eprint =

    Yuan, Zhengqing and Sun, Weixiang and Liu, Yixin and Zhou, Huichi and Zhou, Rong and Li, Yiyang and Zhang, Zheyuan and Song, Wei and Huang, Yue and Jia, Haolong and Murugesan, Keerthiram and Wang, Yu and He, Lifang and Gao, Jianfeng and Sun, Lichao and Ye, Yanfang , journal =....

    2025

  21. [31]

    2025 , eprint =

    Peng, Xiangyu and Qin, Can and Chen, Zeyuan and Xu, Ran and Xiong, Caiming and Wu, Chien-Sheng , journal =. 2025 , eprint =

    2025

  22. [32]

    arXiv preprint arXiv:2602.14162 , year =

    Index Light, Reason Deep: Deferred Visual Ingestion for Visual-Dense Document Question Answering , author =. arXiv preprint arXiv:2602.14162 , year =. 2602.14162 , archivePrefix=

  23. [33]

    Proceedings of the 33rd ACM International Conference on Multimedia , pages=

    DREAM: Integrating Hierarchical Multimodal Retrieval with Multi-page Multimodal Language Model for Documents VQA , author=. Proceedings of the 33rd ACM International Conference on Multimedia , pages=

  24. [34]

    Information Fusion , pages=

    SynJAC: Synthetic-data-driven Joint-granular Adaptation and Calibration for Domain Specific Scanned Document Key Information Extraction , author=. Information Fusion , pages=. 2025 , publisher=

    2025

  25. [35]

    arXiv preprint arXiv:2604.13579 , year=

    MM-Doc-R1: Training Agents for Long Document Visual Question Answering through Multi-turn Reinforcement Learning , author=. arXiv preprint arXiv:2604.13579 , year=

    arXiv

  26. [36]

    The Thirteenth International Conference on Learning Representations , year=

    SV-RAG: LoRA-Contextualizing Adaptation of MLLMs for Long Document Understanding , author=. The Thirteenth International Conference on Learning Representations , year=

  27. [37]

    arXiv preprint arXiv:2511.22850 , year=

    Resolving Evidence Sparsity: Agentic Context Engineering for Long-Document Understanding , author=. arXiv preprint arXiv:2511.22850 , year=

  28. [38]

    Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing , pages=

    MMDocIR: Benchmarking Multimodal Retrieval for Long Documents , author=. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing , pages=

    2025

  29. [39]

    Proceedings of the AAAI Conference on Artificial Intelligence , year=

    MM-R5: MultiModal Reasoning-Enhanced ReRanker via Reinforcement Learning for Document Retrieval , author=. Proceedings of the AAAI Conference on Artificial Intelligence , year=

  30. [40]

    Proceedings of the 2025 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers) , year=

    VisDoM: Multi-Document QA with Visually Rich Elements Using Multimodal Retrieval-Augmented Generation , author=. Proceedings of the 2025 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Paper...

    2025

  31. [41]

    Advances in Neural Information Processing Systems , volume=

    Vrag-rl: Empower vision-perception-based rag for visually rich information understanding via iterative reasoning with reinforcement learning , author=. Advances in Neural Information Processing Systems , volume=

  32. [42]

    Proceedings of the AAAI Conference on Artificial Intelligence , year=

    URaG: Unified Retrieval and Generation in Multimodal LLMs for Efficient Long Document Understanding , author=. Proceedings of the AAAI Conference on Artificial Intelligence , year=

  33. [43]

    arXiv preprint , year=

    DocSeeker: Structured Visual Reasoning with Evidence Grounding for Long Document Understanding , author=. arXiv preprint , year=

  34. [44]

    Proceedings of the ACM Web Conference 2026 , pages=

    DMAP: Human-Aligned Structural Document Map for Multimodal Document Understanding , author=. Proceedings of the ACM Web Conference 2026 , pages=

    2026

  35. [45]

    Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing , pages=

    MultiDocFusion: Hierarchical and Multimodal Chunking Pipeline for Enhanced RAG on Long Industrial Documents , author=. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing , pages=

    2025

  36. [46]

    arXiv preprint arXiv:2510.15253 , year=

    Scaling beyond context: A survey of multimodal retrieval-augmented generation for document understanding , author=. arXiv preprint arXiv:2510.15253 , year=

    arXiv

  37. [47]

    International Journal on Document Analysis and Recognition , volume=

    Deep learning approaches for information extraction from visually rich documents: datasets, challenges and methods , author=. International Journal on Document Analysis and Recognition , volume=. 2025 , publisher=

    2025

  38. [48]

    Artificial Intelligence Review , year=

    Deep learning based visually rich document content understanding: A survey , author=. Artificial Intelligence Review , year=

  39. [49]

    arXiv preprint arXiv:2011.13534 , year=

    A survey of deep learning approaches for ocr and document understanding , author=. arXiv preprint arXiv:2011.13534 , year=

    2011 arXiv

  40. [50]

    arXiv preprint arXiv:2604.13731 , year=

    Doc-V*: Coarse-to-Fine Interactive Visual Reasoning for Multi-Page Document VQA , author=. arXiv preprint arXiv:2604.13731 , year=

    arXiv

  41. [51]

    arXiv preprint arXiv:2503.13964 , year=

    Mdocagent: A multi-modal multi-agent framework for document understanding , author=. arXiv preprint arXiv:2503.13964 , year=

    arXiv

  42. [52]

    Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing , pages=

    Vidorag: Visual document retrieval-augmented generation via dynamic iterative reasoning agents , author=. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing , pages=

    2025

  43. [53]

    arXiv preprint arXiv:2411.04952 , year=

    M3docrag: Multi-modal retrieval is what you need for multi-page multi-document understanding , author=. arXiv preprint arXiv:2411.04952 , year=

    arXiv

  44. [54]

    International Conference on Document Analysis and Recognition , pages=

    Multi-page document visual question answering using self-attention scoring mechanism , author=. International Conference on Document Analysis and Recognition , pages=. 2024 , organization=

    2024

  45. [55]

    Proceedings of the Computer Vision and Pattern Recognition Conference , pages=

    Vdocrag: Retrieval-augmented generation over visually-rich documents , author=. Proceedings of the Computer Vision and Pattern Recognition Conference , pages=

  46. [56]

    Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing , pages=

    Molorag: Bootstrapping document understanding via multi-modal logic-aware retrieval , author=. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing , pages=

    2025

  47. [57]

    Proceedings of the 7th ACM International Conference on Multimedia in Asia , pages=

    AVIR: Adaptive Visual In-Document Retrieval for Efficient Multi-Page Document Question Answering , author=. Proceedings of the 7th ACM International Conference on Multimedia in Asia , pages=

  48. [58]

    Proceedings of the AAAI conference on artificial intelligence , volume=

    Knowledge graph prompting for multi-document question answering , author=. Proceedings of the AAAI conference on artificial intelligence , volume=

  49. [59]

    Proceedings of the Computer Vision and Pattern Recognition Conference , pages=

    A simple yet effective layout token in large language models for document understanding , author=. Proceedings of the Computer Vision and Pattern Recognition Conference , pages=

  50. [60]

    Proceedings of the AAAI conference on artificial intelligence , volume=

    Instructdoc: A dataset for zero-shot generalization of visual document understanding with instructions , author=. Proceedings of the AAAI conference on artificial intelligence , volume=

  51. [61]

    arXiv preprint arXiv:2410.05970 , year=

    Wukong: A large multimodal model for efficient long pdf reading with end-to-end sparse sampling , author=. arXiv preprint arXiv:2410.05970 , year=

    arXiv

  52. [62]

    arXiv preprint arXiv:2410.05261 , year=

    Texthawk2: A large vision-language model excels in bilingual ocr and grounding with 16x fewer tokens , author=. arXiv preprint arXiv:2410.05261 , year=

    arXiv

  53. [63]

    Proceedings of the 32nd ACM International Conference on Multimedia , pages=

    CREAM: coarse-to-fine retrieval and multi-modal efficient tuning for document VQA , author=. Proceedings of the 32nd ACM International Conference on Multimedia , pages=

  54. [64]

    arXiv preprint arXiv:2410.01744 , year=

    Leopard: A vision language model for text-rich multi-image tasks , author=. arXiv preprint arXiv:2410.01744 , year=

    arXiv

  55. [65]

    Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) , pages=

    mplug-docowl2: High-resolution compressing for ocr-free multi-page document understanding , author=. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) , pages=

  56. [66]

    Proceedings of the Computer Vision and Pattern Recognition Conference , pages=

    Docopilot: Improving multimodal models for document-level understanding , author=. Proceedings of the Computer Vision and Pattern Recognition Conference , pages=

  57. [67]

    Pattern Recognition , volume=

    Hierarchical multimodal transformers for multipage docvqa , author=. Pattern Recognition , volume=. 2023 , publisher=

    2023

  58. [68]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    Gram: Global reasoning for multi-page vqa , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  59. [69]

    business document understanding at sub-billion scale , author=

    Arctic-tilt. business document understanding at sub-billion scale , author=. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 6: Industry Track) , pages=

  60. [70]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    Layoutllm: Layout instruction tuning with large language models for document understanding , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

  61. [71]

    Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) , pages=

    Docllm: A layout-aware generative language model for multimodal document understanding , author=. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) , pages=

  62. [72]

    Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers) , pages=

    Doc-react: Multi-page heterogeneous document question-answering , author=. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers) , pages=

  63. [73]

    Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing , pages=

    Docagent: An agentic framework for multi-modal long-context document understanding , author=. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing , pages=

    2025

  64. [74]

    arXiv preprint arXiv:2511.11313 , year=

    DocSLM: A Small Vision-Language Model for Long Multimodal Document Understanding , author=. arXiv preprint arXiv:2511.11313 , year=

  65. [75]

    arXiv preprint arXiv:2508.00579 , year=

    MHier-RAG: Multi-Modal RAG for Visual-Rich Document Question-Answering via Hierarchical and Multi-Granularity Reasoning , author=. arXiv preprint arXiv:2508.00579 , year=

  66. [76]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    Doclayllm: An efficient multi-modal extension of large language models for text-rich document understanding , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  67. [77]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    Docr1: Evidence page-guided grpo for multi-page document understanding , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

  68. [78]

    End-to-End Document Understanding via Chain-of-Reading , author=

  69. [79]

    Electronics , volume=

    Accurate Multi-Page Document Retrieval by Effectively Fusing Context Information Across Pages , author=. Electronics , volume=. 2026 , publisher=

    2026

  70. [80]

    arXiv preprint arXiv:2601.05163 , year=

    DocDancer: Towards Agentic Document-Grounded Information Seeking , author=. arXiv preprint arXiv:2601.05163 , year=

  71. [81]

    International Workshop on Document Analysis Systems , pages=

    Multi-page document VQA with recurrent memory transformer , author=. International Workshop on Document Analysis Systems , pages=. 2024 , organization=

    2024

  72. [82]

    arXiv preprint arXiv:2511.11552 , year=

    Doclens: A tool-augmented multi-agent framework for long visual document understanding , author=. arXiv preprint arXiv:2511.11552 , year=

  73. [83]

    Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing , pages=

    SimpleDoc: Multi-Modal Document Understanding with Dual-Cue Page Retrieval and Iterative Refinement , author=. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing , pages=

    2025

  74. [84]

    arXiv preprint arXiv:2508.03404 , year=

    Visual document understanding and reasoning: A multi-agent collaboration framework with agent-wise adaptive test-time scaling , author=. arXiv preprint arXiv:2508.03404 , year=

  75. [85]

    arXiv preprint arXiv:2603.02438 , year=

    ORCA: Orchestrated Reasoning with Collaborative Agents for Document Visual Question Answering , author=. arXiv preprint arXiv:2603.02438 , year=

  76. [86]

    arXiv preprint arXiv:2602.10271 , year=

    MLDocRAG: Multimodal Long-Context Document Retrieval Augmented Generation , author=. arXiv preprint arXiv:2602.10271 , year=

  77. [87]

    International Conference on Document Analysis and Recognition , pages=

    Enhancing Document VQA Models via Retrieval-Augmented Generation , author=. International Conference on Document Analysis and Recognition , pages=. 2025 , organization=

    2025

  78. [88]

    International Conference on Neural Information Processing , pages=

    M2RAG: A Multi-agent and Multimodal Fusion Framework for Retrieval-Augmented Document QA , author=. International Conference on Neural Information Processing , pages=. 2025 , organization=

    2025

  79. [89]

    Australasian Joint Conference on Artificial Intelligence , pages=

    MultiRAG: An Agentic Multi-Modal and Multi-Source Retrieval-Augmented Generation Framework for Scientific Research , author=. Australasian Joint Conference on Artificial Intelligence , pages=. 2025 , organization=

    2025

  80. [90]

    Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

    Docthinker: Explainable multimodal large language models with rule-based reinforcement learning for document understanding , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

  81. [91]

    arXiv preprint arXiv:2603.07931 , year=

    BRIDGE: Benchmark for multi-hop Reasoning In long multimodal Documents with Grounded Evidence , author=. arXiv preprint arXiv:2603.07931 , year=

  82. [92]

    European Conference on Computer Vision Workshops (Text in Everything) , year=

    Biten, Ali Furkan and Tito, Rub. European Conference on Computer Vision Workshops (Text in Everything) , year=

  83. [93]

    International Conference on Document Analysis and Recognition (ICDAR) , pages=

    Turski, Micha. International Conference on Document Analysis and Recognition (ICDAR) , pages=. 2023 , organization=

    2023

  84. [94]

    2019 , organization=

    Jaume, Guillaume and Ekenel, Hazim Kemal and Thiran, Jean-Philippe , booktitle=. 2019 , organization=

    2019

  85. [95]

    2019 , organization=

    Mishra, Anand and Shekhar, Shashank and Singh, Ajeet Kumar and Chakraborty, Anirban , booktitle=. 2019 , organization=

    2019

  86. [96]

    Park, Seunghyun and Shin, Seung and Lee, Bado and Lee, Junyeop and Surh, Jaeheung and Seo, Minjoon and Lee, Hwalsuk , booktitle=

  87. [97]

    Yang, Zhilin and Qi, Peng and Zhang, Saizheng and Bengio, Yoshua and Cohen, William W and Salakhutdinov, Ruslan and Manning, Christopher D , booktitle=

  88. [98]

    Ferguson, James and Gardner, Matt and Hajishirzi, Hannaneh and Khot, Tushar and Dasigi, Pradeep , booktitle=

  89. [99]

    Constructing a Multi-hop

    Ho, Xanh and Duong Nguyen, Anh-Khoa and Sugawara, Saku and Aizawa, Akiko , booktitle=. Constructing a Multi-hop

  90. [100]

    Trivedi, Harsh and Balasubramanian, Niranjan and Khot, Tushar and Sabharwal, Ashish , journal=

  91. [101]

    Stray, Jonathan and Svetlichnaya, Stacey , year=

  92. [102]

    Hendrycks, Dan and Burns, Collin and Chen, Anya and Ball, Spencer , booktitle=

  93. [103]

    Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

    Document understanding dataset and evaluation (dude) , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

  94. [104]

    Advances in Neural Information Processing Systems , volume=

    Mmlongbench-doc: Benchmarking long-context document understanding with visualizations , author=. Advances in Neural Information Processing Systems , volume=

  95. [105]

    Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) , pages=

    Longdocurl: a comprehensive multimodal long document benchmark integrating understanding, reasoning, and locating , author=. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) , pages=

  96. [106]

    , author=

    MMVQA: A Comprehensive Dataset for Investigating Multipage Multimodal Information Retrieval in PDF-based Visual Question Answering. , author=. IJCAI , pages=

  97. [107]

    Joint European Conference on Machine Learning and Knowledge Discovery in Databases , pages=

    VQA: A new dataset for real-world VQA on PDF documents , author=. Joint European Conference on Machine Learning and Knowledge Discovery in Databases , pages=. 2023 , organization=

    2023

  98. [108]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    Slidevqa: A dataset for document visual question answering on multiple images , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

  99. [109]

    Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

    M3docvqa: Multi-modal multi-page multi-document understanding , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

  100. [110]

    Advances in Neural Information Processing Systems , volume=

    Needle in a multimodal haystack , author=. Advances in Neural Information Processing Systems , volume=

  101. [111]

    Findings of the Association for Computational Linguistics: ACL 2022 , pages=

    DuReadervis: A Chinese dataset for open-domain document visual question answering , author=. Findings of the Association for Computational Linguistics: ACL 2022 , pages=

    2022

  102. [112]

    Advances in Neural Information Processing Systems , volume=

    Uda: A benchmark suite for retrieval augmented generation in real-world document analysis , author=. Advances in Neural Information Processing Systems , volume=

  103. [113]

    arXiv preprint arXiv:2406.11633 , year=

    Docgenome: An open large-scale scientific document benchmark for training and testing multi-modal large language models , author=. arXiv preprint arXiv:2406.11633 , year=

  104. [114]

    arXiv preprint arXiv:2408.12637 , year=

    Building and better understanding vision-language models: insights and future directions , author=. arXiv preprint arXiv:2408.12637 , year=

    arXiv

  105. [115]

    Proceedings of the IEEE/CVF winter conference on applications of computer vision , pages=

    Docvqa: A dataset for vqa on document images , author=. Proceedings of the IEEE/CVF winter conference on applications of computer vision , pages=

  106. [116]

    Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision , pages=

    Infographicvqa , author=. Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision , pages=

  107. [117]

    Findings of the association for computational linguistics: ACL 2022 , pages=

    Chartqa: A benchmark for question answering about charts with visual and logical reasoning , author=. Findings of the association for computational linguistics: ACL 2022 , pages=

    2022

  108. [118]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    Towards vqa models that can read , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

  109. [119]

    Proceedings of the IEEE/CVF international conference on computer vision , pages=

    Scene text visual question answering , author=. Proceedings of the IEEE/CVF international conference on computer vision , pages=

  110. [120]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    Visualmrc: Machine reading comprehension on document images , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

  111. [121]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    PubTables-1M: Towards comprehensive table extraction from unstructured documents , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  112. [122]

    Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing , pages=

    M-longdoc: A benchmark for multimodal super-long document understanding and a retrieval-aware tuning framework , author=. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing , pages=

    2025

  113. [124]

    arXiv preprint arXiv:2507.09861 , year=

    A Survey on MLLM-based Visually Rich Document Understanding: Methods, Challenges, and Emerging Trends , author=. arXiv preprint arXiv:2507.09861 , year=

    arXiv

  114. [125]

    Proceedings of the 30th ACM international conference on multimedia , pages=

    Layoutlmv3: Pre-training for document ai with unified text and image masking , author=. Proceedings of the 30th ACM international conference on multimedia , pages=

  115. [126]

    Proceedings of the AAAI conference on artificial intelligence , volume=

    Docformerv2: Local features for document understanding , author=. Proceedings of the AAAI conference on artificial intelligence , volume=

  116. [127]

    International Conference on Machine Learning , pages=

    Pix2struct: Screenshot parsing as pretraining for visual language understanding , author=. International Conference on Machine Learning , pages=. 2023 , organization=

    2023

  117. [128]

    International conference on machine learning , pages=

    Learning transferable visual models from natural language supervision , author=. International conference on machine learning , pages=. 2021 , organization=

    2021

  118. [129]

    arXiv preprint arXiv:2111.15664 , volume=

    Donut: Document understanding transformer without ocr , author=. arXiv preprint arXiv:2111.15664 , volume=

    arXiv

  119. [130]

    Proceedings of the 30th ACM international conference on multimedia , pages=

    Dit: Self-supervised pre-training for document image transformer , author=. Proceedings of the 30th ACM international conference on multimedia , pages=

  120. [131]

    5-VL Technical Report (No

    Qwen2. 5-VL Technical Report (No. arXiv: 2502.13923). arXiv , author=

    arXiv

  121. [132]

    Hu, Anwen and Xu, Haiyang and Ye, Jiabo and Yan, Ming and Zhang, Liang and Zhang, Bo and Zhang, Ji and Jin, Qin and Huang, Fei and Zhou, Jingren , booktitle=

  122. [133]

    Transactions of the association for computational linguistics , volume=

    Lost in the middle: How language models use long contexts , author=. Transactions of the association for computational linguistics , volume=

  123. [134]

    Advances in neural information processing systems , volume=

    Attention is all you need , author=. Advances in neural information processing systems , volume=

  124. [135]

    Advances in neural information processing systems , volume=

    Chain-of-thought prompting elicits reasoning in large language models , author=. Advances in neural information processing systems , volume=

  125. [136]

    arXiv preprint arXiv:2210.03629 , year=

    React: Synergizing reasoning and acting in language models , author=. arXiv preprint arXiv:2210.03629 , year=

    arXiv

  126. [137]

    Proceedings of the 29th annual international ACM SIGIR conference on Research and development in information retrieval , pages=

    Building a test collection for complex document information processing , author=. Proceedings of the 29th annual international ACM SIGIR conference on Research and development in information retrieval , pages=

  127. [138]

    Journal of Artificial Intelligence Research , volume=

    Agentic large language models, a survey , author=. Journal of Artificial Intelligence Research , volume=

  128. [139]

    Frontiers of Computer Science , volume=

    A survey on large language model based autonomous agents , author=. Frontiers of Computer Science , volume=. 2024 , publisher=

    2024

  129. [140]

    Retrieval-Augmented Generation for Knowledge-Intensive

    Lewis, Patrick and Perez, Ethan and Piktus, Aleksandra and Petroni, Fabio and Karpukhin, Vladimir and Goyal, Naman and K. Retrieval-Augmented Generation for Knowledge-Intensive. Advances in Neural Information Processing Systems , volume=

  130. [141]

    2020 , publisher=

    Guu, Kelvin and Lee, Kenton and Tung, Zora and Pasupat, Panupong and Chang, Ming-Wei , booktitle=. 2020 , publisher=

    2020

  131. [142]

    Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: Main Volume , pages=

    Leveraging Passage Retrieval with Generative Models for Open Domain Question Answering , author=. Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: Main Volume , pages=. 2021 , publisher=

    2021

  132. [143]

    Journal of Machine Learning Research , volume=

    Atlas: Few-shot Learning with Retrieval Augmented Language Models , author=. Journal of Machine Learning Research , volume=

  133. [144]

    Transactions of the Association for Computational Linguistics , volume=

    In-Context Retrieval-Augmented Language Models , author=. Transactions of the Association for Computational Linguistics , volume=

  134. [145]

    Nakano, Reiichiro and Hilton, Jacob and Balaji, Suchir and Wu, Jeff and Ouyang, Long and Kim, Christina and Hesse, Christopher and Jain, Shantanu and Kosaraju, Vineet and Saunders, William and Jiang, Xu and Cobbe, Karl and Eloundou, Tyna and Krueger, Gretchen and Button, Kevin...

  135. [146]

    Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) , pages=

    Internet-Augmented Dialogue Generation , author=. Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) , pages=. 2022 , publisher=

    2022

  136. [147]

    Advances in Neural Information Processing Systems , volume=

    Toolformer: Language Models Can Teach Themselves to Use Tools , author=. Advances in Neural Information Processing Systems , volume=

  137. [148]

    , booktitle=

    Chen, Wenhu and Hu, Hexiang and Chen, Xi and Verga, Pat and Cohen, William W. , booktitle=. 2022 , publisher=

    2022

  138. [149]

    and Fathi, Alireza , booktitle=

    Hu, Ziniu and Iscen, Ahmet and Sun, Chen and Wang, Zirui and Chang, Kai-Wei and Sun, Yizhou and Schmid, Cordelia and Ross, David A. and Fathi, Alireza , booktitle=

  139. [150]

    Yu, Shi and Tang, Chaoyue and Xu, Bokai and Cui, Junbo and Ran, Junhao and Yan, Yukun and Liu, Zhenghao and Wang, Shuo and Han, Xu and Liu, Zhiyuan and Sun, Maosong , journal=

  140. [151]

    and Shen, Yelong and Wallis, Phillip and Allen-Zhu, Zeyuan and Li, Yuanzhi and Wang, Shean and Wang, Lu and Chen, Weizhu , booktitle=

    Hu, Edward J. and Shen, Yelong and Wallis, Phillip and Allen-Zhu, Zeyuan and Li, Yuanzhi and Wang, Shean and Wang, Lu and Chen, Weizhu , booktitle=

  141. [152]

    Khattab, Omar and Zaharia, Matei , booktitle=

  142. [153]

    2024 , note=

    Faysse, Manuel and Sibille, Hugues and Wu, Tony and Omrani, Bilel and Viaud, Gautier and Hudelot, C. 2024 , note=

    2024

  143. [154]

    Proceedings of the 30th ACM International Conference on Multimedia , pages=

    Towards Complex Document Understanding by Discrete Reasoning , author=. Proceedings of the 30th ACM International Conference on Multimedia , pages=

  144. [155]

    Compositional Semantic Parsing on Semi-Structured Tables , author=. Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing (Volume 1: Long Papers) , pages=. 2015 , publisher=

    2015

  145. [156]

    Chen, Wenhu and Wang, Hongmin and Chen, Jianshu and Zhang, Yunkai and Wang, Hong and Li, Shiyang and Zhou, Xiyou and Wang, William Yang , booktitle=

  146. [157]

    Transactions of the Association for Computational Linguistics , volume=

    Nan, Linyong and Hsieh, Chiachun and Mao, Ziming and Lin, Xi Victoria and Verma, Neha and Zhang, Rui and Kry. Transactions of the Association for Computational Linguistics , volume=

  147. [158]

    Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT) , pages=

    A Dataset of Information-Seeking Questions and Answers Anchored in Research Papers , author=. Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL-HLT) , pages=. 2021 , publisher=

    2021

  148. [159]

    Document Analysis and Recognition -- ICDAR 2021 , pages=

    Kleister: Key Information Extraction Datasets Involving Long Documents with Complex Layouts , author=. Document Analysis and Recognition -- ICDAR 2021 , pages=. 2021 , publisher=

    2021

  149. [160]

    Zou, Anni and Yu, Wenhao and Zhang, Hongming and Ma, Kaixin and Cai, Deng and Zhang, Zhuosheng and Zhao, Hai and Yu, Dong , journal=

  150. [161]

    Wang, Zhaowei and Yu, Wenhao and Ren, Xiyu and Zhang, Jipeng and Zhao, Yu and Saxena, Rohit and Cheng, Liang and Wong, Ginny and See, Simon and Minervini, Pasquale and Song, Yangqiu and Steedman, Mark , booktitle=

  151. [162]

    2022 , publisher=

    Zhao, Yilun and Li, Yunxiang and Li, Chenying and Zhang, Rui , booktitle=. 2022 , publisher=

    2022

  152. [163]

    ColPali : Efficient document retrieval with vision language models

    Faysse, Manuel and Sibille, Hugues and Wu, Tony and Omrani, Bilel and Viaud, Gautier and Hudelot, C \'e line and Colombo, Pierre. ColPali : Efficient document retrieval with vision language models. arXiv:2407.01449

    arXiv

  153. [164]

    Advances in neural information processing systems , volume=

    Reflexion: Language agents with verbal reinforcement learning , author=. Advances in neural information processing systems , volume=

  154. [165]

    Advances in neural information processing systems , volume=

    Tree of thoughts: Deliberate problem solving with large language models , author=. Advances in neural information processing systems , volume=

  155. [166]

    arXiv preprint arXiv:2203.11171 , year=

    Self-consistency improves chain of thought reasoning in language models , author=. arXiv preprint arXiv:2203.11171 , year=

    arXiv

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.