Pith. sign in

REVIEW 3 major objections 5 minor 100 references

What Is That Talk About? A Video-to-Text Summarization Dataset for Scientific Presentations

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

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

Pith's one-line read This paper introduces VISTA, 18,599 conference talk videos paired with paper abstracts, and shows that an explicit question plan before summarizing improves quality and factual consistency over end-to-end models, with humans still ahead.

desk verdict VISTA is a genuinely useful dataset for scientific video summarization with a solid but not novel plan-based method; the abstract-as-summary proxy needs firmer validation and the zero-shot labeling is misleading, but neither sinks the paper. read the letter →

arxiv 2502.08279 v4 pith:6XX3AHPI submitted 2025-02-12 cs.CL cs.AIcs.CV

Dongqi Liu , Chenxi Whitehouse , Xi Yu , Louis Mahon , Rohit Saxena , Zheng Zhao , Yifu Qiu , Mirella Lapata
show 1 more author
Vera Demberg
This is my paper · ORCID
classification cs.CLcs.AIcs.CV
keywords VISTAdatasetvideo-to-textsummarizationscientificpresentationsplan-basedQuestionUnderDiscussionmultimodallargelanguagemodelsbenchmarkfactualconsistency
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

Scientific talks are hard to summarize automatically: they mix speech, slides, and technical visuals, and current models struggle to turn any of that into a faithful written summary. The paper introduces VISTA, 18,599 recorded AI conference presentations paired with the abstracts of the papers they present, so that this task can be measured at scale. It then argues that explicit planning helps: if a model first predicts an ordered list of questions and then writes a summary whose sentences answer those questions, both summary quality and factual consistency improve over direct end-to-end generation, across zero-shot and fine-tuned settings. Even so, every model tested still trails human-written abstracts by a wide margin, which the paper takes as evidence that the benchmark is not yet solved.

What carries the argument

The load-bearing object is the question plan: an ordered list $p = (q_1, \dots, q_m)$ of questions, one per sentence of the target abstract, generated before any summary text is written. The pipeline trains a Plan Generation module to predict $p$ from the input video and a Summary Generation module to learn $P(s \mid v, p)$, so each summary sentence is meant to answer its corresponding question. The plan instantiates Question Under Discussion theory, the view that discourse is organized by the implicit questions it resolves, and silver-standard plans are produced by GPT-o1, one question per reference sentence using that sentence plus its predecessors as context. Plan quality is measurable and consequential: replacing the full set of plan questions with irrelevant ones drops Rouge-2 from 13.74 to 12.71, while deleting up to 60% of plan questions shortens summaries far more controllably than instruction-based prompting does.

What would settle it

Have annotators score a random sample of VISTA pairs for whether the abstract genuinely captures what the talk says, not merely what the paper claims; if a substantial fraction of talks omit, contradict, or go beyond their abstract, the proxy assumption fails. A second check: regenerate the plans with a weaker open-source question generator and rerun the comparison, since the reported gains partly inherit the quality of the GPT-o1 silver plans.

Watch

Extended reading notes

Core claim

On VISTA, the paper's central claim is that scientific video summarization is both measurable and unsolved, and that the structure of scientific abstracts is a lever for improving it. The strongest evidence is the plan-based pipeline: Plan-mPlug-Owl3, which first generates a question plan from the video and then produces the summary conditioned on that plan, outperforms the best end-to-end open-source model, mPLUG-Owl3, on every reported metric, with the largest gains in factual consistency (FactVC 75.41 versus 71.94 under full fine-tuning) and quality (Rouge-L 33.25 versus 32.91). Human evaluation confirms the ordering: three annotators rated human-written abstracts highest or lowest 81.7% and 0.1% of the time, respectively, versus 12.4% and 6.6% for Plan-mPlug-Owl3, with the plan-based model significantly better than other neural models on faithfulness, coherence, and informativeness. The same planning scaffold also improves transcript-, OCR-, and audio-only models across all metrics, indicating that the benefit comes from imposing discourse structure rather than from video processing alone.

Load-bearing premise

The premise everything rests on is that a paper's abstract is a faithful summary of what the speaker actually says in the video; if presenters deviate from their abstracts, the labels are noisy and part of the task cannot be learned from the video alone.

Editorial extensions

If this is right

  • VISTA gives the field a large public benchmark for scientific video summarization: 18,599 pairs, with in-domain fine-tuning and video input both clearly mattering, since video-based models beat transcript- and audio-only models in every learning setting.
  • The regular structure of scientific abstracts is a usable prior: a question-driven plan before generation improves summary quality and factual consistency across zero-shot, QLoRA, and full fine-tuning, with statistical significance.
  • Planning generalizes beyond video: the same scaffold improves transcript-, OCR-, and audio-only models on all metrics, so discourse structure helps summarization independently of the input modality.
  • The task is far from closed: the best model reaches FactVC 75.41 against 88.54 for human abstracts, and human judges ranked human abstracts best 81.7% of the time.

Reading between the lines

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

  • Read strictly, VISTA measures how well a model can recover a paper's abstract from its talk; a dataset built on independent human summaries of the videos themselves would likely be harder and would test a slightly different skill, and this distinction is left implicit in the paper.
  • Because the plan questions are silver labels from a top commercial model, the framework's measured gains inherit the quality of that generator; a cheaper open-source planner would probably narrow the advantage, so the practical value of the method depends on plan-generation cost.
  • The controlled-generation results hint at a product-level use the paper only sketches: editing plan questions could let end users steer a summary's reading level, length, and emphasis in a more reliable way than prompting the model to rewrite.
  • The same pairs could support neighbouring tasks beyond summarization, such as retrieving talks from abstracts, grounding claims in specific slides, or studying which visual and spoken cues carry the content of the abstract.
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 introduces VISTA, a large-scale dataset for video-to-text summarization of scientific presentations, containing 18,599 pairs of conference talk videos (from ACL Anthology venues, ICML, and NeurIPS) aligned with the corresponding paper abstracts. The authors benchmark closed- and open-source large multimodal models under zero-shot, QLoRA, and full fine-tuning settings, and propose a plan-based summarization framework in which a question-generation module produces a sequence of QUD-style questions (a plan) and a summary-generation module verbalizes answers to those questions. Automatic metrics and a human evaluation on 50 test instances are used to compare models. The paper reports that the plan-based method consistently improves summary quality and factual consistency over end-to-end baselines, while a considerable gap to human performance remains.

Significance. If the dataset labels are valid, VISTA fills a genuine gap: it is larger and more specialized for scientific talks than prior video-summarization resources, and its domain focus, venue diversity, and long-form summaries make it a potentially useful benchmark. The paper also contributes a transparent planning framework that is evaluated across multiple backbones and learning regimes, with both automatic and human judgments. The experimental design is generally careful—fixed splits, comparable hyperparameters, ablations on plan quality and modality, and human evaluation with reported inter-annotator agreement. The central claims, however, rest on two under-validated points: that a paper abstract is a reliable video-grounded summary, and that the 'zero-shot' plan-based row in Table 3 is a fair comparison. These are fixable within the manuscript's scope, but they affect the benchmark's primary validity and the interpretation of the planning gains.

major comments (3)
  1. The benchmark's central assumption—that the paper abstract is a faithful summary of the video—is not directly validated. The manual QC in Appendix B asked two annotators for a binary valid/invalid judgment on 500 pairs, and the automated GPT-o1 check flagged only 39 pairs that were later confirmed valid; neither procedure verifies that the content of each abstract sentence is actually grounded in the video. A talk can omit parts of an abstract, include framing absent from the paper, or present formal/theoretical statements only in the paper; the paper itself reports reference abstracts scoring 88.54 FactVC and 4.62 VideoScore (Table 3), which is consistent with imperfect video-grounding of the gold summaries. If a nontrivial fraction of abstract sentences is not recoverable from the video, then the task is partially unlearnable from video alone, the reported upper bounds are miscalibrated, and the relative gains of plan-based models may partly reflect fitting paper-style text rather than video summarization. Because the dataset is the primary contribution, the authors should provide direct sentence-level evidence of video grounding—for example, annotate a stratified sample of test pairs for whether each abstract sentence is stated or clearly implied by the video—and report the fraction of ungrounded sentences, filtering or discussing the ceiling they impose.
  2. The row labeled 'Plan-mPlug-Owl3⋇' under 'Zero-shot Learning' is not a zero-shot comparison: the footnote states that only the PG module is trained on the training set and then used to predict plans on the test set. This gives the plan-based variant in-domain supervision that the zero-shot baselines do not receive, so the reported improvements (e.g., FactVC 47.37 vs. 42.07 for mPLUG-Owl3) cannot be attributed to planning alone. The authors should either evaluate a truly zero-shot variant in which the PG module is not trained, or relabel this condition (e.g., 'PG-trained, SG zero-shot') and soften the abstract/introduction claim that the plan-based method improves zero-shot performance. This is load-bearing because the paper's central claim that explicit planning enhances summary quality is partly supported by this row.
  3. The silver-standard plans are generated by GPT-o1 from the reference abstract sentences and their preceding context, and then used to train both PG and SG. The manual validation in Appendix G checks only whether the generated questions are locally coherent and could be answered by the reference sentence; it does not check whether the plan questions can be answered from the video alone. Since the test-time PG is trained to imitate reference-derived plans, any systematic divergence between video content and abstract content (see the first major comment) will propagate through the plan. I would like to see a small analysis of how well PG-predicted plans are answerable from the corresponding video, and how plan quality varies when the reference abstract contains content not present in the talk.
minor comments (5)
  1. [Table 1, caption] The caption contains a typo: 'Comprison' should be 'Comparison'.
  2. [§4, Plan Generation] The notation leaves unspecified how the number of plan questions m is determined at inference time; please clarify whether the PG module predicts a variable-length sequence and how the maximum length is set.
  3. [Appendix D, Table 7] The version table lists 'VideoLLaMA2-7B-16F' while the main text uses 'Video-LLaMA'; please reconcile the model naming for reproducibility.
  4. [References] GPT-o1 is cited via Achiam et al. (2023), the GPT-4 technical report; if an o1-specific model card or technical report is available, it should be cited instead.
  5. [Figure 3] The x-axis label 'T okens' has a spacing error; minor but easy to fix.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: test-time summaries are generated from plans predicted from the video, not from gold abstracts; self-citations are motivational rather than load-bearing.

full rationale

The paper's derivation chain is self-contained. The central claims—(1) VISTA is a challenging video-to-text summarization benchmark and (2) a plan-based framework improves summary quality and factual consistency—are each supported by experiments run on the paper's own dataset with standard external metrics (ROUGE, SacreBLEU, METEOR, BERTScore, CIDEr-D, VideoScore, FactVC) and by human evaluation against gold abstracts. No prediction reduces to its input by construction. Although silver-standard plans are generated from reference abstracts (Section 4, Plan Generation: 'We leverage GPT-o1 (Achiam et al., 2023) to generate silver-standard plans based on reference summary sentences and their preceding context'), the test-time evaluation is clean: 'During inference, the trained PG module predicts plan p-hat for input v, and the tuple [v; p-hat] is fed into the SG module to generate the final summary'—the evaluated summaries depend on the video through a predicted plan, not on the gold abstract. The abstract-as-proxy assumption is stated transparently in the Limitations ('we consider the paper abstract as a proxy for the summary of the corresponding video') and is a data-validity assumption rather than a circular definition; the two-stage QC (Appendix B) validates it externally, and its imperfection is acknowledged in the paper. Self-citations (Liu et al., 2025; Mahon and Lapata, 2024a; Liu et al., 2023a) appear as related-work motivation, but the plan-based framework is attributed primarily to external prior work (Narayan et al., 2021, 2023; Roberts, 2012), and the paper's own experiments and ablations provide the evidence for the planning claim. No uniqueness theorem is imported from the authors' prior work, and no known result is renamed as organization. The principal caveat—that a proxy-validated dataset may underestimate task learnability from video alone—is a correctness/data-quality risk, not circularity.

Assumptions & free parameters 0 free parameters · 3 assumptions · 0 invented entities

No free parameters are fitted to the central claim; the plan length and hyperparameters are design choices. The load-bearing assumptions are the abstract-as-summary proxy, the quality of GPT-o1 silver plans, and the validity of automatic metrics. No new entities are introduced.

assumptions (3)
  • domain assumption The paper abstract is a faithful summary of the corresponding presentation video.
    This is the core labeling assumption for the dataset. It is explicitly acknowledged as a proxy in the Limitations section: "we consider the paper abstract as a proxy for the summary of the corresponding video." Quality control on 500 samples and GPT-o1 checks on all samples are used to support it, but the assumption is not independently verified for all 18,599 pairs.
  • domain assumption GPT-o1-generated plan questions are well-formed and QUD-aligned with the reference summary sentences.
    The plan-based method relies on silver-standard plans generated by GPT-o1 from reference summaries. Manual validation on 100 samples reports high Fleiss' kappa, but this is a limited check and the quality of plans at scale is assumed.
  • standard math Automatic metrics (ROUGE, BERTScore, VideoScore, FactVC) are valid proxies for summary quality and factual consistency in this task.
    These metrics are standard in the field and are used without establishing task-specific correlation to human judgments beyond the small human evaluation of 50 samples.

how reviews work

0 comments
Cite this review

Pith. "Pith review of What Is That Talk About? A Video-to-Text Summarization Dataset for Scientific Presentations." pith.science (2026). https://pith.science/paper/6XX3AHPI

@misc{pith2026250208279,
  author       = {Pith},
  title        = {Pith review of: What Is That Talk About? A Video-to-Text Summarization Dataset for Scientific Presentations},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6XX3AHPI}},
  note         = {Machine review of arXiv:2502.08279}
}
read the original abstract

Transforming recorded videos into concise and accurate textual summaries is a growing challenge in multimodal learning. This paper introduces VISTA, a dataset specifically designed for video-to-text summarization in scientific domains. VISTA contains 18,599 recorded AI conference presentations paired with their corresponding paper abstracts. We benchmark the performance of state-of-the-art large models and apply a plan-based framework to better capture the structured nature of abstracts. Both human and automated evaluations confirm that explicit planning enhances summary quality and factual consistency. However, a considerable gap remains between models and human performance, highlighting the challenges of our dataset. This study aims to pave the way for future research on scientific video-to-text summarization.

Figures

Figures reproduced from arXiv: 2502.08279 by the authors.

Figure 1
Figure 1. An example from VISTA: a conference pre [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Venue distribution of the VISTA dataset. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Distribution of summary sentences, summary tokens, video durations, and video shots in VISTA. [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (13 more)
Figure 4
Figure 4. Figure 4: GPT-o1 generates plans based on reference summaries. Each question qi corresponds to a summary sentence ti , which we assume constitutes its answer. Index i ranges from 1 to the number of summary sentences. posits that discourse often revolves around a set of questions…
Figure 5
Figure 5. Figure 5: Noise in plan generation impacts summariza [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Human evaluation results. Human-written summaries consistently outperform all neural models. model, along with the proportion of instances where models are rated best or worst. Fleiss’ Kappa scores for Faithfulness (κ = 0.767), Rele￾vance (κ = 0.842), Informativeness (…
Figure 7
Figure 7. Figure 7: A random sample from the VISTA dataset, originating from [PITH_FULL_IMAGE:figures/full_fig_p017_7.png]
Figure 8
Figure 8. Figure 8: Impact of text context for plan generation. [PITH_FULL_IMAGE:figures/full_fig_p019_8.png]
Figure 9
Figure 9. Figure 9: LMM-as-Judge evaluation results showing that human-written summaries consistently outperform neural models. evaluators and GPT-o1 achieve substantial levels of agreement across these dimensions. Following this, we expand the evaluation to include all sam￾ples in our te…
Figure 14
Figure 14. Figure 14: Prompt for SG model. Irrelevant Question Generation Randomly generate a question with a ques￾tion mark. Question Sentence [PITH_FULL_IMAGE:figures/full_fig_p023_14.png]
Figure 15
Figure 15. Figure 15: Prompt used by GPT-o1 to generate irrele [PITH_FULL_IMAGE:figures/full_fig_p023_15.png]
Figure 16
Figure 16. Figure 16: Summary readability modification. Summary Length Modification Rewrite the following text to further adjust the style or detail. Here is the text to be rewritten: {Text} Shorten the above text by about {10% / 30% / 60%}. Focus on the key points and remove less critical…
Figure 17
Figure 17. Figure 17: Summary length modification. Plan Readability Modification Rewrite the following questions to further adjust the style or detail. Here are the questions to be rewritten: 1. {Q1} 2. {Q2} ... Refine the above questions to be more {lay/expert} style. Modified Questions …
Figure 13
Figure 13. Figure 13: Prompt for PG model. Prompt for SG model Generate a summary for the following {Video/Audio/Transcript...} based on the plan questions. Content: {Video/Audio/Transcript...}. Plan Questions: {Questions} Ensure that the generated summary sequen￾tially answers the plan qu…
Figure 19
Figure 19. Figure 19: A snapshot of the experimental instructions provided to human evaluators. [PITH_FULL_IMAGE:figures/full_fig_p024_19.png]
Figure 20
Figure 20. Figure 20: Prompt for GPT-o1 to evaluate summary quality [PITH_FULL_IMAGE:figures/full_fig_p024_20.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

100 extracted references · 35 canonical work pages

  1. [1]

    Josh Achiam, Steven Adler, Sandhini Agarwal, Lama Ahmad, Ilge Akkaya, Florencia Leoni Aleman, Diogo Almeida, Janko Altenschmidt, Sam Altman, Shyamal Anadkat, et al. 2023. https://arxiv.org/abs/2303.08774 Gpt-4 technical report . arXiv preprint arXiv:2303.08774

    arXiv 2023

  2. [2]

    Daechul Ahn, Yura Choi, Youngjae Yu, Dongyeop Kang, and Jonghyun Choi. 2024. https://doi.org/10.18653/v1/2024.acl-long.52 Tuning large multimodal models for videos using reinforcement learning from AI feedback . In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 923--940, Bangkok, Thai...

    work page doi:10.18653/v1/2024.acl-long.52 2024

  3. [3]

    Reinald Kim Amplayo, Stefanos Angelidis, and Mirella Lapata. 2021. https://ojs.aaai.org/index.php/AAAI/article/view/17481 Unsupervised opinion summarization with content planning . In Proceedings of the AAAI conference on artificial intelligence, volume 35, pages 12489--12497

    2021

  4. [4]

    Anthropic. 2024. Claude 3.5 - sonnet. https://www.anthropic.com/news/claude-3-5-sonnet. Accessed: 2024-12-06

    2024

  5. [5]

    Dawit Mureja Argaw, Seunghyun Yoon, Fabian Caba Heilbron, Hanieh Deilamsalehy, Trung Bui, Zhaowen Wang, Franck Dernoncourt, and Joon Son Chung. 2024. https://ieeexplore.ieee.org/document/10656029/ Scaling up video summarization pretraining with large language models . In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pa...

    work page arXiv 2024

  6. [6]

    Zechen Bai, Pichao Wang, Tianjun Xiao, Tong He, Zongbo Han, Zheng Zhang, and Mike Zheng Shou. 2024. https://arxiv.org/abs/2404.18930 Hallucination of multimodal large language models: A survey . arXiv preprint arXiv:2404.18930

    arXiv 2024

  7. [7]

    Satanjeev Banerjee and Alon Lavie. 2005. https://aclanthology.org/W05-0909 METEOR : An automatic metric for MT evaluation with improved correlation with human judgments . In Proceedings of the ACL Workshop on Intrinsic and Extrinsic Evaluation Measures for Machine Translation and/or Summarization , pages 65--72, Ann Arbor, Michigan. Association for Comput...

    2005

  8. [8]

    Isabel Cachola, Kyle Lo, Arman Cohan, and Daniel Weld. 2020. https://doi.org/10.18653/v1/2020.findings-emnlp.428 TLDR : Extreme summarization of scientific documents . In Findings of the Association for Computational Linguistics: EMNLP 2020, pages 4766--4777, Online. Association for Computational Linguistics

    doi:10.18653/v1/2020.findings-emnlp.428 2020

Show all 100 references
  1. [9]

    Gerard Canal, Senka Krivi \'c , Paul Luff, and Andrew Coles. 2022. Planverb: Domain-independent verbalization and summary of task plans. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 36, pages 9698--9706

    2022

  2. [10]

    Brian Chen, Xiangyuan Zhao, and Yingnan Zhu. 2024 a . Personalized video summarization by multimodal video understanding. In Proceedings of the 33rd ACM International Conference on Information and Knowledge Management, pages 4382--4389

    2024

  3. [11]

    Dongping Chen, Ruoxi Chen, Shilin Zhang, Yaochen Wang, Yinuo Liu, Huichi Zhou, Qihui Zhang, Yao Wan, Pan Zhou, and Lichao Sun. 2024 b . Mllm-as-a-judge: assessing multimodal llm-as-a-judge with vision-language benchmark. In Proceedings of the 41st International Conference on M...

    2024

  4. [12]

    Zhe Chen, Heyang Liu, Wenyi Yu, Guangzhi Sun, Hongcheng Liu, Ji Wu, Chao Zhang, Yu Wang, and Yanfeng Wang. 2024 c . https://doi.org/10.18653/v1/2024.acl-long.489 M ^3 AV : A multimodal, multigenre, and multipurpose audio-visual academic lecture dataset . In Proceedings of the ...

    2024 doi

  5. [13]

    Siyuan Cheng, Bozhong Tian, Qingbin Liu, Xi Chen, Yongheng Wang, Huajun Chen, and Ningyu Zhang. 2023. https://doi.org/10.18653/v1/2023.emnlp-main.856 Can we edit multimodal large language models? In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Pr...

    2023 doi

  6. [14]

    Sangwoo Cho, Franck Dernoncourt, Tim Ganter, Trung Bui, Nedim Lipka, Walter Chang, Hailin Jin, Jonathan Brandt, Hassan Foroosh, and Fei Liu. 2021. https://doi.org/10.18653/v1/2021.emnlp-main.520 S tream H over: Livestream transcript summarization and annotation . In Proceeding...

    2021 doi

  7. [15]

    Yunfei Chu, Jin Xu, Qian Yang, Haojie Wei, Xipin Wei, Zhifang Guo, Yichong Leng, Yuanjun Lv, Jinzheng He, Junyang Lin, et al. 2024. https://arxiv.org/abs/2407.10759 Qwen2-audio technical report . arXiv preprint arXiv:2407.10759

    2024 arXiv

  8. [16]

    Aldan Creo, Manuel Lama, and Juan C Vidal. 2023. https://arxiv.org/abs/2312.08282 Prompting llms with content plans to enhance the summarization of scientific articles . arXiv preprint arXiv:2312.08282

    2023 arXiv

  9. [17]

    Tim Dettmers, Artidoro Pagnoni, Ari Holtzman, and Luke Zettlemoyer. 2024. https://proceedings.neurips.cc/paper_files/paper/2023/file/1feb87871436031bdc0f2beaa62a049b-Paper-Conference.pdf Qlora: Efficient finetuning of quantized llms . Advances in Neural Information Processing ...

    2024

  10. [18]

    Mehwish Fatima and Michael Strube. 2023. https://doi.org/10.18653/v1/2023.acl-long.103 Cross-lingual science journalism: Select, simplify and rewrite summaries for non-expert readers . In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (...

    2023 doi

  11. [19]

    Patrick Fernandes, Kayo Yin, Emmy Liu, Andr \'e Martins, and Graham Neubig. 2023. https://doi.org/10.18653/v1/2023.acl-long.36 When does translation require context? a data-driven, multilingual exploration . In Proceedings of the 61st Annual Meeting of the Association for Comp...

    2023 doi

  12. [20]

    Xingyu Fu, Yushi Hu, Bangzheng Li, Yu Feng, Haoyu Wang, Xudong Lin, Dan Roth, Noah A Smith, Wei-Chiu Ma, and Ranjay Krishna. 2025. Blink: Multimodal large language models can see but not perceive. In European Conference on Computer Vision, pages 148--166. Springer

    2025

  13. [21]

    Xiyan Fu, Jun Wang, and Zhenglu Yang. 2021. https://doi.org/10.18653/v1/2021.naacl-main.473 MM - AVS : A full-scale dataset for multi-modal summarization . In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Hum...

    2021 doi

  14. [22]

    Qi Gou, Zehua Xia, Bowen Yu, Haiyang Yu, Fei Huang, Yongbin Li, and Nguyen Cam-Tu. 2023. https://doi.org/10.18653/v1/2023.emnlp-main.104 Diversify question generation with retrieval-augmented style transfer . In Proceedings of the 2023 Conference on Empirical Methods in Natura...

    2023 doi

  15. [23]

    Mingfei Han, Linjie Yang, Xiaojun Chang, Lina Yao, and Heng Wang. 2025. https://openreview.net/forum?id=FZv3kPHTtB Shot2story: A new benchmark for comprehensive understanding of multi-shot videos . In The Thirteenth International Conference on Learning Representations

    2025

  16. [24]

    Bo He, Jun Wang, Jielin Qiu, Trung Bui, Abhinav Shrivastava, and Zhaowen Wang. 2023. Align and attend: Multimodal summarization with dual contrastive losses. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 14867--14878

    2023

  17. [25]

    Xuan He, Dongfu Jiang, Ge Zhang, Max Ku, Achint Soni, Sherman Siu, Haonan Chen, Abhranil Chandra, Ziyan Jiang, Aaran Arulraj, Kai Wang, Quy Duc Do, Yuansheng Ni, Bohan Lyu, Yaswanth Narsupalli, Rongqi Fan, Zhiheng Lyu, Bill Yuchen Lin, and Wenhu Chen. 2024. https://doi.org/10....

    2024 doi

  18. [26]

    Anwen Hu, Yaya Shi, Haiyang Xu, Jiabo Ye, Qinghao Ye, Ming Yan, Chenliang Li, Qi Qian, Ji Zhang, and Fei Huang. 2024. https://openreview.net/forum?id=S7pieMItch m PLUG -paperowl: Scientific diagram analysis with the multimodal large language model . In ACM Multimedia 2024

    2024

  19. [27]

    Hang Hua, Yunlong Tang, Chenliang Xu, and Jiebo Luo. 2024. V2xum-llm: Cross-modal video summarization with temporal prompt instruction tuning. arXiv preprint arXiv:2404.12353

    2024

  20. [28]

    Kung-Hsiang Huang, Hou Pong Chan, Yi R Fung, Haoyi Qiu, Mingyang Zhou, Shafiq Joty, Shih-Fu Chang, and Heng Ji. 2024. From pixels to insights: A survey on automatic chart understanding in the era of large foundation models. IEEE Transactions on Knowledge and Data Engineering

    2024

  21. [29]

    Fantine Huot, Joshua Maynez, Chris Alberti, Reinald Kim Amplayo, Priyanka Agrawal, Constanza Fierro, Shashi Narayan, and Mirella Lapata. 2024. https://aclanthology.org/2024.eacl-long.131 PLAN : Summarizing using a content plan as cross-lingual bridge . In Proceedings of the 18...

    2024

  22. [30]

    Md Mohaiminul Islam, Ngan Ho, Xitong Yang, Tushar Nagarajan, Lorenzo Torresani, and Gedas Bertasius. 2024. Video recap: Recursive captioning of hour-long videos. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 18198--18208

    2024

  23. [31]

    Jiaxin Ju, Ming Liu, Huan Yee Koh, Yuan Jin, Lan Du, and Shirui Pan. 2021. https://doi.org/10.18653/v1/2021.findings-emnlp.345 Leveraging information bottleneck for scientific document summarization . In Findings of the Association for Computational Linguistics: EMNLP 2021, pa...

    2021 doi

  24. [32]

    Jing Yu Koh, Ruslan Salakhutdinov, and Daniel Fried. 2023. https://proceedings.mlr.press/v202/koh23a.html Grounding language models to images for multimodal inputs and outputs . In International Conference on Machine Learning, pages 17283--17300. PMLR

    2023

  25. [33]

    Mateusz Krubi \'n ski and Pavel Pecina. 2023. https://doi.org/10.18653/v1/2023.findings-eacl.67 MLASK : Multimodal summarization of video-based news articles . In Findings of the Association for Computational Linguistics: EACL 2023, pages 910--924, Dubrovnik, Croatia. Associat...

    2023 doi

  26. [34]

    Mateusz Krubi \'n ski and Pavel Pecina. 2024. https://aclanthology.org/2024.findings-eacl.30 Towards unified uni- and multi-modal news headline generation . In Findings of the Association for Computational Linguistics: EACL 2024, pages 437--450, St. Julian ' s, Malta. Associat...

    2024

  27. [35]

    u bler, Wittawat Jitkrittum, Bernhard Sch \

    Jonas K \"u bler, Wittawat Jitkrittum, Bernhard Sch \"o lkopf, and Krikamol Muandet. 2020. Learning kernel tests without data splitting. Advances in Neural Information Processing Systems, 33:6245--6255

    2020

  28. [36]

    Guy Lev, Michal Shmueli-Scheuer, Jonathan Herzig, Achiya Jerbi, and David Konopnicki. 2019. https://doi.org/10.18653/v1/P19-1204 T alk S umm: A dataset and scalable annotation method for scientific paper summarization based on conference talks . In Proceedings of the 57th Annu...

    2019 doi

  29. [37]

    Bohao Li, Yuying Ge, Yixiao Ge, Guangzhi Wang, Rui Wang, Ruimao Zhang, and Ying Shan. 2024 a . Seed-bench: Benchmarking multimodal large language models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 13299--13308

    2024

  30. [38]

    Feng Li, Renrui Zhang, Hao Zhang, Yuanhan Zhang, Bo Li, Wei Li, Zejun MA, and Chunyuan Li. 2025. https://openreview.net/forum?id=oSQiao9GqB LL a VA -ne XT -interleave: Tackling multi-image, video, and 3d in large multimodal models . In The Thirteenth International Conference o...

    2025

  31. [39]

    Haoran Li, Junnan Zhu, Cong Ma, Jiajun Zhang, and Chengqing Zong. 2017. https://doi.org/10.18653/v1/D17-1114 Multi-modal summarization for asynchronous collection of text, image, audio and video . In Proceedings of the 2017 Conference on Empirical Methods in Natural Language P...

    2017 doi

  32. [40]

    Jiwei Li, Michel Galley, Chris Brockett, Jianfeng Gao, and Bill Dolan. 2016. https://doi.org/10.18653/v1/N16-1014 A diversity-promoting objective function for neural conversation models . In Proceedings of the 2016 Conference of the North A merican Chapter of the Association f...

    2016 doi

  33. [41]

    Junnan Li, Yongkang Wong, Qi Zhao, and Mohan S Kankanhalli. 2019. Video storytelling: Textual summaries for events. IEEE Transactions on Multimedia, 22(2):554--565

    2019

  34. [42]

    Lei Li, Yuqi Wang, Runxin Xu, Peiyi Wang, Xiachong Feng, Lingpeng Kong, and Qi Liu. 2024 b . https://doi.org/10.18653/v1/2024.acl-long.775 Multimodal A r X iv: A dataset for improving scientific comprehension of large vision-language models . In Proceedings of the 62nd Annual ...

    2024 doi

  35. [43]

    Mingzhe Li, Xiuying Chen, Shen Gao, Zhangming Chan, Dongyan Zhao, and Rui Yan. 2020. https://doi.org/10.18653/v1/2020.emnlp-main.752 VMSMO : Learning to generate multimodal summary for video-based news articles . In Proceedings of the 2020 Conference on Empirical Methods in Na...

    2020 doi

  36. [44]

    Yanwei Li, Chengyao Wang, and Jiaya Jia. 2024 c . https://doi.org/10.1007/978-3-031-72952-2_19 Llama-vid: An image is worth 2 tokens in large language models . In Computer Vision – ECCV 2024: 18th European Conference, Milan, Italy, September 29–October 4, 2024, Proceedings, Pa...

    2024 doi

  37. [45]

    Paul Pu Liang, Amir Zadeh, and Louis-Philippe Morency. 2024. Foundations & trends in multimodal machine learning: Principles, challenges, and open questions. ACM Computing Surveys, 56(10):1--42

    2024

  38. [46]

    Bin Lin, Yang Ye, Bin Zhu, Jiaxi Cui, Munan Ning, Peng Jin, and Li Yuan. 2024 a . https://doi.org/10.18653/v1/2024.emnlp-main.342 Video- LL a VA : Learning united visual representation by alignment before projection . In Proceedings of the 2024 Conference on Empirical Methods ...

    2024 doi

  39. [47]

    Chin-Yew Lin. 2004. https://www.aclweb.org/anthology/W04-1013 ROUGE : A package for automatic evaluation of summaries . In Text Summarization Branches Out, pages 74--81, Barcelona, Spain. Association for Computational Linguistics

    2004

  40. [48]

    Jingyang Lin, Hang Hua, Ming Chen, Yikang Li, Jenhao Hsiao, Chiuman Ho, and Jiebo Luo. 2023. Videoxum: Cross-modal visual and textural summarization of videos. IEEE Transactions on Multimedia

    2023

  41. [49]

    Jingyang Lin, Hang Hua, Ming Chen, Yikang Li, Jenhao Hsiao, Chiuman Ho, and Jiebo Luo. 2024 b . https://doi.org/10.1109/TMM.2023.3335875 Videoxum: Cross-modal visual and textural summarization of videos . IEEE Transactions on Multimedia, 26:5548--5560

    2024

  42. [50]

    Danyang Liu, Mirella Lapata, and Frank Keller. 2023 a . https://doi.org/10.18653/v1/2023.findings-emnlp.386 Visual storytelling with question-answer plans . In Findings of the Association for Computational Linguistics: EMNLP 2023, pages 5800--5813, Singapore. Association for C...

    2023 doi

  43. [51]

    Dongqi Liu and Vera Demberg. 2023. https://doi.org/10.18653/v1/2023.acl-srw.1 C hat GPT vs human-authored text: Insights into controllable text summarization and sentence style transfer . In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistic...

    2023 doi

  44. [52]

    Dongqi Liu and Vera Demberg. 2024. https://doi.org/10.18653/v1/2024.naacl-long.121 RST - L o RA : A discourse-aware low-rank adaptation for long document abstractive summarization . In Proceedings of the 2024 Conference of the North American Chapter of the Association for Comp...

    2024 doi

  45. [53]

    Dongqi Liu, Xudong Hong, Pin-Jie Lin, Ernie Chang, and Vera Demberg. 2022. https://aclanthology.org/2022.creativesumm-1.9/ Two-stage movie script summarization: An efficient method for low-resource long document summarization . In Proceedings of the Workshop on Automatic Summa...

    2022

  46. [54]

    Dongqi Liu, Yifan Wang, and Vera Demberg. 2023 b . https://doi.org/10.18653/v1/2023.acl-long.306 Incorporating distributions of discourse structure for long document abstractive summarization . In Proceedings of the 61st Annual Meeting of the Association for Computational Ling...

    2023 doi

  47. [55]

    Dongqi Liu, Yifan Wang, Jia Loy, and Vera Demberg. 2024 a . https://aclanthology.org/2024.lrec-main.1258 S ci N ews: From scholarly complexities to public narratives -- a dataset for scientific news report generation . In Proceedings of the 2024 Joint International Conference ...

    2024

  48. [56]

    Dongqi Liu, Xi Yu, Vera Demberg, and Mirella Lapata. 2025. https://arxiv.org/abs/2504.19339 Explanatory summarization with discourse-driven planning . Preprint, arXiv:2504.19339

    2025

  49. [57]

    Hui Liu and Xiaojun Wan. 2021. https://doi.org/10.18653/v1/2021.acl-short.9 Video paragraph captioning as a text summarization task . In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natur...

    2021 doi

  50. [58]

    Hui Liu and Xiaojun Wan. 2023. https://doi.org/10.18653/v1/2023.emnlp-main.723 Models see hallucinations: Evaluating the factuality in video captioning . In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 11807--11823, Singapore. A...

    2023 doi

  51. [59]

    Ran Liu, Ming Liu, Min Yu, He Zhang, Jianguo Jiang, Gang Li, and Weiqing Huang. 2024 b . https://doi.org/10.18653/v1/2024.findings-acl.574 S um S urvey: An abstractive dataset of scientific survey papers for long document summarization . In Findings of the Association for Comp...

    2024 doi

  52. [60]

    Yinhong Liu, Han Zhou, Zhijiang Guo, Ehsan Shareghi, Ivan Vuli \'c , Anna Korhonen, and Nigel Collier. 2024 c . https://openreview.net/forum?id=9gdZI7c6yr Aligning with human judgement: The role of pairwise preference in large language model evaluators . In First Conference on...

    2024

  53. [61]

    Zhengyuan Liu and Nancy Chen. 2021. https://doi.org/10.18653/v1/2021.emnlp-main.8 Controllable neural dialogue summarization with personal named entity planning . In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 92--106, Online a...

    2021 doi

  54. [62]

    Adian Liusie, Potsawee Manakul, and Mark Gales. 2024. https://aclanthology.org/2024.eacl-long.8 LLM comparative assessment: Zero-shot NLG evaluation through pairwise comparisons using large language models . In Proceedings of the 18th Conference of the European Chapter of the ...

    2024

  55. [63]

    Ilya Loshchilov and Frank Hutter. 2019. https://openreview.net/forum?id=Bkg6RiCqY7 Decoupled weight decay regularization . In International Conference on Learning Representations

    2019

  56. [64]

    Pan Lu, Hritik Bansal, Tony Xia, Jiacheng Liu, Chunyuan Li, Hannaneh Hajishirzi, Hao Cheng, Kai-Wei Chang, Michel Galley, and Jianfeng Gao. 2024. https://openreview.net/forum?id=KUNzEQMWU7 Mathvista: Evaluating mathematical reasoning of foundation models in visual contexts . I...

    2024

  57. [65]

    Muhammad Maaz, Hanoona Rasheed, Salman Khan, and Fahad Khan. 2024. https://doi.org/10.18653/v1/2024.acl-long.679 Video- C hat GPT : Towards detailed video understanding via large vision and language models . In Proceedings of the 62nd Annual Meeting of the Association for Comp...

    2024 doi

  58. [66]

    Louis Mahon and Mirella Lapata. 2024 a . https://doi.org/10.18653/v1/2024.acl-long.450 A modular approach for multimodal summarization of TV shows . In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 8272--...

    2024 doi

  59. [67]

    Louis Mahon and Mirella Lapata. 2024 b . Screenwriter: Automatic screenplay generation and movie summarisation. arXiv preprint arXiv:2410.19809

    2024 arXiv

  60. [68]

    Alex Mallen, Akari Asai, Victor Zhong, Rajarshi Das, Daniel Khashabi, and Hannaneh Hajishirzi. 2023. https://doi.org/10.18653/v1/2023.acl-long.546 When not to trust language models: Investigating effectiveness of parametric and non-parametric memories . In Proceedings of the 6...

    2023 doi

  61. [69]

    Yuning Mao, Ming Zhong, and Jiawei Han. 2022. https://doi.org/10.18653/v1/2022.emnlp-main.750 C ite S um: Citation text-guided scientific extreme summarization and domain adaptation with limited supervision . In Proceedings of the 2022 Conference on Empirical Methods in Natura...

    2022 doi

  62. [70]

    Shashi Narayan, Joshua Maynez, Reinald Kim Amplayo, Kuzman Ganchev, Annie Louis, Fantine Huot, Anders Sandholm, Dipanjan Das, and Mirella Lapata. 2023. https://doi.org/10.1162/tacl_a_00583 Conditional generation with a question-answering blueprint . Transactions of the Associa...

    2023 doi

  63. [71]

    Shashi Narayan, Yao Zhao, Joshua Maynez, Gon c alo Sim \ o es, Vitaly Nikolaev, and Ryan McDonald. 2021. https://doi.org/10.1162/tacl_a_00438 Planning with learned entity prompts for abstractive summarization . Transactions of the Association for Computational Linguistics, 9:1...

    2021 doi

  64. [72]

    Pinelopi Papalampidi and Mirella Lapata. 2023. https://doi.org/10.18653/v1/2023.findings-eacl.96 H ierarchical3 D adapters for long video-to-text summarization . In Findings of the Association for Computational Linguistics: EACL 2023, pages 1297--1320, Dubrovnik, Croatia. Asso...

    2023 doi

  65. [73]

    Vaidehi Patil, Leonardo Ribeiro, Mengwen Liu, Mohit Bansal, and Markus Dreyer. 2024. https://doi.org/10.18653/v1/2024.acl-long.743 REFINESUMM : Self-refining MLLM for generating a multimodal summarization dataset . In Proceedings of the 62nd Annual Meeting of the Association f...

    2024 doi

  66. [74]

    Matt Post. 2018. https://www.aclweb.org/anthology/W18-6319 A call for clarity in reporting BLEU scores . In Proceedings of the Third Conference on Machine Translation: Research Papers, pages 186--191, Belgium, Brussels. Association for Computational Linguistics

    2018

  67. [75]

    Shraman Pramanick, Rama Chellappa, and Subhashini Venugopalan. 2024. https://openreview.net/forum?id=h3lddsY5nf SPIQA : A dataset for multimodal question answering on scientific papers . In The Thirty-eight Conference on Neural Information Processing Systems Datasets and Bench...

    2024

  68. [76]

    Jielin Qiu, Jiacheng Zhu, William Han, Aditesh Kumar, Karthik Mittal, Claire Jin, Zhengyuan Yang, Linjie Li, Jianfeng Wang, Ding Zhao, et al. 2024. Mmsum: A dataset for multimodal summarization and thumbnail generation of videos. In Proceedings of the IEEE/CVF Conference on Co...

    2024

  69. [77]

    Aishwarya Ramakrishnan and Chun-Kit Ngan. 2022. https://doi.org/10.1109/AIKE55402.2022.00012 A hybrid video-to-text summarization framework and algorithm on cascading advanced extractive- and abstractive-based approaches for supporting viewers' video navigation and understandi...

    2022

  70. [78]

    Craige Roberts. 2012. Information structure: Towards an integrated formal theory of pragmatics. Semantics and pragmatics, 5:6--1

    2012

  71. [79]

    Ramon Sanabria, Ozan Caglayan, Shruti Palaskar, Desmond Elliott, Lo \" c Barrault, Lucia Specia, and Florian Metze. 2018. How2: A large-scale dataset for multimodal language understanding. In NeurIPS

    2018

  72. [80]

    Xindi Shang, Zehuan Yuan, Anran Wang, and Changhu Wang. 2021. https://doi.org/10.1145/3474085.3475321 Multimodal video summarization via time-aware transformers . In Proceedings of the 29th ACM International Conference on Multimedia, MM '21, page 1756–1765, New York, NY, USA. ...

    2021

  73. [81]

    Aseem Srivastava, Smriti Joshi, Tanmoy Chakraborty, and Md Shad Akhtar. 2024. https://doi.org/10.18653/v1/2024.emnlp-main.984 Knowledge planning in large language models for domain-aligned counseling summarization . In Proceedings of the 2024 Conference on Empirical Methods in...

    2024 doi

  74. [82]

    Ashima Suvarna, Xiao Liu, Tanmay Parekh, Kai-Wei Chang, and Nanyun Peng. 2024. https://doi.org/10.18653/v1/2024.emnlp-main.76 QUDSELECT : Selective decoding for questions under discussion parsing . In Proceedings of the 2024 Conference on Empirical Methods in Natural Language ...

    2024 doi

  75. [83]

    Sotaro Takeshita, Tommaso Green, Ines Reinig, Kai Eckert, and Simone Ponzetto. 2024. https://doi.org/10.18653/v1/2024.naacl-long.371 ACLS um: A new dataset for aspect-based summarization of scientific publications . In Proceedings of the 2024 Conference of the North American C...

    2024 doi

  76. [84]

    Gemini Team, Rohan Anil, Sebastian Borgeaud, Jean-Baptiste Alayrac, Jiahui Yu, Radu Soricut, Johan Schalkwyk, Andrew M Dai, Anja Hauth, Katie Millican, et al. 2023. Gemini: a family of highly capable multimodal models. arXiv preprint arXiv:2312.11805

    2023 arXiv

  77. [85]

    Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, Marie-Anne Lachaux, Timoth \'e e Lacroix, Baptiste Rozi \`e re, Naman Goyal, Eric Hambro, Faisal Azhar, et al. 2023. Llama: Open and efficient foundation language models. arXiv preprint arXiv:2302.13971

    2023 arXiv

  78. [86]

    Ramakrishna Vedantam, C Lawrence Zitnick, and Devi Parikh. 2015. Cider: Consensus-based image description evaluation. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 4566--4575

    2015

  79. [87]

    Ye Wang, Xiaojun Wan, and Zhiping Cai. 2022. https://doi.org/10.18653/v1/2022.findings-emnlp.248 Guiding abstractive dialogue summarization with content planning . In Findings of the Association for Computational Linguistics: EMNLP 2022, pages 3408--3413, Abu Dhabi, United Ara...

    2022 doi

  80. [88]

    Tianhe Wu, Kede Ma, Jie Liang, Yujiu Yang, and Lei Zhang. 2025. A comprehensive study of multimodal large language models for image quality assessment. In European Conference on Computer Vision, pages 143--160. Springer

    2025

  81. [89]

    Yating Wu, Ritika Mangla, Greg Durrett, and Junyi Jessy Li. 2023 a . https://doi.org/10.18653/v1/2023.emnlp-main.325 QUD eval: The evaluation of questions under discussion discourse parsing . In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Proces...

    2023 doi

  82. [90]

    Yating Wu, William Sheffield, Kyle Mahowald, and Junyi Jessy Li. 2023 b . https://doi.org/10.18653/v1/2023.emnlp-main.336 Elaborative simplification as implicit questions under discussion . In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processi...

    2023 doi

  83. [91]

    Jiabo Ye, Haiyang Xu, Haowei Liu, Anwen Hu, Ming Yan, Qi Qian, Ji Zhang, Fei Huang, and Jingren Zhou. 2025. https://openreview.net/forum?id=pr37sbuhVa m PLUG -owl3: Towards long image-sequence understanding in multi-modal large language models . In The Thirteenth International...

    2025

  84. [92]

    Xiang Yue, Yuansheng Ni, Kai Zhang, Tianyu Zheng, Ruoqi Liu, Ge Zhang, Samuel Stevens, Dongfu Jiang, Weiming Ren, Yuxuan Sun, et al. 2024. Mmmu: A massive multi-discipline multimodal understanding and reasoning benchmark for expert agi. In Proceedings of the IEEE/CVF Conferenc...

    2024

  85. [93]

    Hang Zhang, Xin Li, and Lidong Bing. 2023. https://doi.org/10.18653/v1/2023.emnlp-demo.49 Video- LL a MA : An instruction-tuned audio-visual language model for video understanding . In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing: Syst...

    2023 doi

  86. [94]

    Litian Zhang, Xiaoming Zhang, Linfeng Han, Zelong Yu, Yun Liu, and Zhoujun Li. 2024 a . Multi-task hierarchical heterogeneous fusion framework for multimodal summarization. Information Processing & Management, 61(4):103693

    2024

  87. [95]

    Weinberger, and Yoav Artzi

    Tianyi Zhang, Varsha Kishore, Felix Wu, Kilian Q. Weinberger, and Yoav Artzi. 2020. https://openreview.net/forum?id=SkeHuCVFDr Bertscore: Evaluating text generation with bert . In International Conference on Learning Representations

    2020

  88. [96]

    Yu Zhang, Xiusi Chen, Bowen Jin, Sheng Wang, Shuiwang Ji, Wei Wang, and Jiawei Han. 2024 b . https://doi.org/10.18653/v1/2024.emnlp-main.498 A comprehensive survey of scientific large language models and their applications in scientific discovery . In Proceedings of the 2024 C...

    2024 doi

  89. [97]

    Xing, Hao Zhang, Joseph E

    Lianmin Zheng, Wei-Lin Chiang, Ying Sheng, Siyuan Zhuang, Zhanghao Wu, Yonghao Zhuang, Zi Lin, Zhuohan Li, Dacheng Li, Eric P. Xing, Hao Zhang, Joseph E. Gonzalez, and Ion Stoica. 2024. Judging llm-as-a-judge with mt-bench and chatbot arena. In Proceedings of the 37th Internat...

    2024

  90. [98]

    Luowei Zhou, Chenliang Xu, and Jason Corso. 2018. Towards automatic learning of procedures from web instructional videos. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 32

    2018

  91. [99]

    online" 'onlinestring :=

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint eprinttype howpublished institution journal key month note number organization pages publisher school series title type volume year doi pubmed url lastchecked label extra.label sort.label short.list...

  92. [100]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.