Pith. sign in

REVIEW 4 major objections 4 minor 36 references

RMPL: Relation-aware Multi-task Progressive Learning with Stage-wise Training for Multimedia Event Extraction

T0 review · 4 major / 4 minor · reviewed 2026-08-02 · deepseek-v4-flash

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

Pith's one-line read A vision-language model can learn multimedia event extraction from text-only, image-only, and relation supervision, without any annotated multimedia events.

desk verdict A sensible low-resource training recipe for multimedia event extraction, with consistent gains across three backbones; the main caveat is that the label mapping and evaluation protocol are not documented enough to verify. read the letter →

arxiv 2602.13748 v2 pith:L2WYBRJN submitted 2026-02-14 cs.CL cs.CV

Yongkang Jin , Jianwen Luo , Jingjing Wang , Jianmin Yao , Yu Hong This is my paper · ORCID
classification cs.CLcs.CV
keywords multimediaeventextractionvision-languagemodelsmulti-tasklearningprogressivesupervisedfine-tuningrelationlow-resourceargument
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

The paper claims that multimedia event extraction can be learned without any annotated multimedia event data by training in two stages. In the first stage, a vision-language model is warmed up on a unified schema covering textual event extraction, visual event extraction, and multimedia relation extraction. In the second stage, it is fine-tuned separately for event mention identification and argument role extraction on mixed text and image supervision. On the M2E2 benchmark, this raises event-mention and argument-role F1 across three different VLM backbones and in text-only, image-only, and multimedia evaluation settings. The significance is that it offers a route around the field's core bottleneck: M2E2 is the only benchmark and has no training split.

What carries the argument

The key mechanism is stage-wise progressive learning with a unified schema. A schema control variable selects among three supervision families (textual event extraction, visual event extraction, multimedia relation extraction), all serialized into the same autoregressive generation format. Stage I trains the model to predict these schemas from the document; Stage II fine-tunes the warmed-up model, without relation supervision, on the two MEE subtasks using mixed textual and visual data. The auxiliary relation extraction is what makes the initialization 'relation-aware', and the ablation attributes part of the final gain to it.

What would settle it

Re-annotate a held-out subsample of M2E2 as a training split and train the same VLM directly on it; if direct supervision does not beat RMPL's external-supervision pipeline, the transfer claim is weakened. Alternatively, replace the ACE/SWiG annotations with shuffled or coarse labels that preserve the task format but destroy event semantics; if RMPL still shows most of its gain, the improvement comes from format adaptation, not from event knowledge.

Watch

Extended reading notes

Core claim

The central discovery is that structured, event-centric representations for multimedia event extraction can be transferred from heterogeneous unimodal and relational supervision sources rather than from in-domain annotations. RMPL first trains the VLM to generate linearized event schemas for text, images, and relation triples under a shared interface, then specializes it with task-specific supervised fine-tuning. On M2E2, the full recipe outperforms both prompt-only inference and warm-up-only training for all three backbones tested; for example multimedia event-mention F1 reaches 92.44 on one backbone and argument-role F1 rises by up to 22.87 absolute points over the prompt-only baseline.

Load-bearing premise

Everything rests on the assumption that ACE 2005 textual events, SWiG visual activities, and MNRE relations can be projected onto M2E2's 8 event types and 15 roles without systematic loss; the paper asserts this mapping in Section 5.1 but provides no mapping details or coverage statistics.

Editorial extensions

If this is right

  • If the claim holds, MEE can be tackled in low-resource settings using existing unimodal event datasets plus a relation-extraction corpus, removing the need to annotate multimedia events.
  • The same two-stage recipe could be applied to other multimodal information-extraction tasks whose benchmarks lack training splits.
  • Relation-aware warm-up is presented as a generally useful initialization: even when Stage II does most of the work, the warm-up stabilises and improves the final model.
  • The gains are consistent across three VLM backbones, suggesting the training strategy transfers rather than being tuned to one model.
  • Direct comparison against prior MEE methods shows best or near-best F1 across most modality settings, so the paradigm is competitive with dedicated cross-modal alignment and prompting methods.

Reading between the lines

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

  • The paper's mapping from ACE 2005 event types and SWiG visual activities to M2E2's 8 event types and 15 roles is stated but not detailed; if that mapping is lossy, part of the reported gain could be an artefact of the projection rather than of the progressive-learning paradigm.
  • A testable extension the paper does not run: swap the relation-extraction corpus for a different multimedia relation dataset to see whether the relation-aware warm-up benefit is tied to that particular corpus's schema.
  • One implication left implicit is that the same staged recipe may transfer to other low-resource multimodal tasks such as multimodal named entity recognition or multimodal sentiment analysis, where heterogeneous unimodal supervision is available.
  • The paper does not isolate whether Stage II's gains come from supervised task format exposure or from the warm-up's representation; a variant that skips warm-up but uses the same Stage II data shows warm-up matters, but the boundary between format learning and semantic learning is not drawn.
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

4 major / 4 minor

Summary. The paper proposes RMPL, a two-stage training method for multimedia event extraction (MEE) under low-resource conditions where M2E2 has no training split. Stage I warm-ups a VLM under a unified generation schema using heterogeneous external supervision: ACE 2005 textual event extraction, SWiG/imSitu visual event extraction, and MNRE multimedia relation extraction. Stage II specializes the warmed-up model through supervised fine-tuning for event mention identification and argument role extraction, using the same external sources projected onto M2E2's 8 event types and 15 roles. Experiments on M2E2 with Qwen2-VL-7B, InternVL3_5-8B, and Qwen3-VL-8B report consistent F1 gains over prompt-only and warm-up-only baselines across text-only, image-only, and multimedia settings, together with ablations for relation-aware warm-up and mixing proportions.

Significance. If the reported gains are genuine, the paper makes a useful contribution: it shows that heterogeneous unimodal supervision plus a relation-aware warm-up and task-specific specialization can improve MEE without using M2E2 training labels. The use of three different VLM backbones under a common evaluation protocol and the component-wise ablations are strengths. However, the central transfer claim rests on an undocumented label mapping from ACE/SWiG/imSitu to M2E2, and a key hyperparameter (the 1:1:1 mixing ratio) is selected by evaluating on M2E2 itself. Until these issues are resolved, the source and magnitude of the reported gains remain uncertain.

major comments (4)
  1. [Section 5.1, Table 1] The load-bearing claim is that supervision from ACE 2005 (33 event types, 36 roles), SWiG/imSitu (504 activities, 1,788 roles), and MNRE (23 relations) transfers to M2E2's 8 event types and 15 roles. The paper only says, 'we map both textual event types and visual activity semantics to the same 8 event types defined in M2E2.' No mapping table, coverage statistics, or treatment of unmapped labels is given. If the projection is lossy or manually encodes M2E2's ontology into the training targets, the consistent gains in Table 2 could be produced by the label projection rather than by the proposed progressive-learning paradigm. The authors should provide the full mapping, counts of how many training examples fall in each M2E2 type/role, and a discussion of discarded or ambiguous labels.
  2. [Section 5.4.2, Figure 3] The supervision mixing ratio (1:1:1) is selected by comparing variants on M2E2 itself. Because M2E2 is the only evaluation benchmark, choosing the ratio on it means the reported 'consistently best' configuration is partly fitted to the test set, undermining the out-of-sample claim. The authors should either use a held-out portion of M2E2 for validation or pre-specify the ratio and report sensitivity. The same concern applies to the Stage-I/Stage-II training schedules, which are also free parameters selected for the reported configuration.
  3. [Tables 2-4, Experimental Settings] All results appear to be single runs with no error bars, multiple seeds, or significance tests. Given the relatively small M2E2 evaluation set and the variability of VLM fine-tuning, 'consistent improvements' could be within run-to-run noise. The authors should report means and standard deviations over at least three seeds per condition, and ideally a paired significance test across the three backbones. Without this, the magnitude of the claimed gains is not statistically grounded.
  4. [Section 4.2, Figure 2, Reproducibility] The method is defined by its prompt templates and schema serialization, but neither the exact templates nor the code is released or included in the appendix. Eq. (1) depends on the schema control variable τ and the serialization protocol, yet the actual prompts shown in Figure 2 are illustrative. For the central claim to be independently checkable, the authors should release the prompt templates, the mapping tables, and the training/inference code, or at least provide complete template details in a supplementary appendix.
minor comments (4)
  1. [Tables 2 and 3] Several table cells appear to have formatting errors with missing spaces or concatenated numbers, e.g., '88.5686.63' in the InternVL3_5-8B Baseline row of Table 2, and '75.02.6', '49.959.1', '42.449.945.8' in Table 3. These need to be corrected for readability.
  2. [Section 4.2, Eq. (2)] The notation m=τ is confusing: τ is introduced as a schema control variable in Eq. (1), and m is used for modality in Eq. (2). Please use distinct symbols or define the relation between m and τ explicitly.
  3. [Section 5.4.2] The mapping between α:β:λ in Eq. (4) and the 'sampling ratios' ACE:SWiG:MNRE in Figure 3 is not made explicit. Please clarify whether α, β, λ are loss weights, sampling probabilities, or both.
  4. [Section 5.3, Table 3] The comparison with prior work uses different backbone architectures and training setups. The paper correctly frames this as a 'high-level reference,' but the text says RMPL 'achieves the best overall performance'; please soften this to avoid overclaiming given the uncontrolled comparison.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: RMPL's training signal is external (ACE/SWiG/imSitu/MNRE); M2E2 is used only for evaluation and as a label ontology, not as a fitted training target.

full rationale

The derivation chain is not circular. RMPL's supervision comes from ACE 2005, SWiG/imSitu, and MNRE, while M2E2 is explicitly used only for evaluation: the paper states 'M2E2 is the only established benchmark... it provides annotations exclusively for evaluation.' The only point where M2E2 enters training is as a label ontology, via the sentence 'we map both textual event types and visual activity semantics to the same 8 event types defined in M2E2,' and this mapping is attributed to prior work by Li et al., not to the present authors. Using a target benchmark's label schema to project external supervision is standard transfer learning and is not equivalent to fitting on M2E2's annotations. The losses in Eqs. (2)-(6) are defined over external data distributions S_t, S_i, S_r and their labels; M2E2 never appears as a training distribution. The self-citation to REAR [21] appears only in related work and is not load-bearing. Two caveats are worth noting but they are not circularity: the ACE/SWiG-to-M2E2 mapping in Sec 5.1 is undocumented (a reproducibility risk), and the mixing ratio 1:1:1 in Sec 5.4.2 is selected by evaluating on M2E2 itself (test-set hyperparameter selection). These could inflate the reported numbers or make the transfer claim hard to audit, but they do not make the reported gains forced by construction, and no equation or fitted parameter reduces M2E2 output to M2E2 input.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The framework is empirical: no new physical entities, and no derivation. The main ledger items are the manual label-projection from ACE/SWiG to M2E2 types/roles, the assumption that MNRE relations help event-role grounding, and the mixing-ratio hyperparameter that was tuned on the evaluation benchmark.

free parameters (2)
  • Stage-I supervision mixing ratio (α:β:λ) = 1:1:1 (selected from 1:1:0.5, 1:1:1, 1:1:2)
    Eq. (4) weights the three Stage-I losses; Section 5.4.2 compares ratios on M2E2 and chooses 1:1:1, so 'balanced' is fitted to the evaluation benchmark.
  • Stage-I/Stage-II training schedule = 400 steps @ LR 1e-5; 800 steps @ LR 3e-5; batch size 16
    Hand-chosen hyperparameters reported in Section 5.1; no sensitivity analysis beyond mixing ratio, so the reported configuration is one arbitrary point.
assumptions (4)
  • domain assumption M2E2 has no usable training split; its annotations are for evaluation only.
    Stated in the abstract and Section 1; motivates all external supervision. If M2E2 train labels existed, the low-resource premise would weaken.
  • ad hoc to paper ACE 2005 types, SWiG/imSitu activities, and MNRE relations can be projected to M2E2's 8 event types and 15 roles.
    Section 5.1: 'we map both textual event types and visual activity semantics to the same 8 event types defined in M2E2'; mapping details absent, so the transfer assumption is load-bearing and paper-specific.
  • domain assumption Multimedia relation extraction provides structured supervision that improves event-centric representation learning.
    Motivates the auxiliary loss in Section 4.2; the ablation supports it on M2E2, but there is no external evidence that relation triples transfer to event roles.
  • domain assumption The VLM can generate linearized schemas and bounding boxes under the unspecified serialization protocol.
    Stage I loss (Eq. 1) and visual argument evaluation (IoU>0.5) require the model to output structured sequences/boxes; exact decoding/protocol is not specified.

how reviews work

0 comments
Cite this review

Pith. "Pith review of RMPL: Relation-aware Multi-task Progressive Learning with Stage-wise Training for Multimedia Event Extraction." pith.science (2026). https://pith.science/paper/L2WYBRJN

@misc{pith2026260213748,
  author       = {Pith},
  title        = {Pith review of: RMPL: Relation-aware Multi-task Progressive Learning with Stage-wise Training for Multimedia Event Extraction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/L2WYBRJN}},
  note         = {Machine review of arXiv:2602.13748}
}
read the original abstract

Multimedia Event Extraction (MEE) aims to identify events and their arguments from documents that contain both text and images. It requires grounding event semantics across different modalities. Progress in MEE is limited by the lack of annotated training data. M2E2 is the only established benchmark, but it provides annotations only for evaluation. This makes direct supervised training impractical. Existing methods mainly rely on cross-modal alignment or inference-time prompting with Vision--Language Models (VLMs). These approaches do not explicitly learn structured event representations and often produce weak argument grounding in multimodal settings. To address these limitations, we propose RMPL, a Relation-aware Multi-task Progressive Learning framework for MEE under low-resource conditions. RMPL incorporates heterogeneous supervision from unimodal event extraction and multimedia relation extraction with stage-wise training. The model is first trained with a unified schema to learn shared event-centric representations across modalities. It is then fine-tuned for event mention identification and argument role extraction using mixed textual and visual data. Experiments on the M2E2 benchmark with multiple VLMs show consistent improvements across different modality settings.

Figures

Figures reproduced from arXiv: 2602.13748 by the authors.

Figure 1
Figure 1. An example of MEE from M2E2 [1]. The Justice: Arrest-Jail event is recognized from multimedia event con￾tent with paired text and image inputs, triggered by “de￾tained”, and event arguments are classified across modalities. Most existing approaches to MEE treat this task as a cross-modal alignment or inference-based formulation. Early methods associate textual event mentions with visual evidence through weakly su￾pe… view at source ↗
Figure 2
Figure 2. Overview of RMPL. Stage I conducts unified schema warm-up training with heterogeneous event-centric supervision to [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Impact of different supervision mixing proportions. [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

36 extracted references · 1 linked inside Pith

  1. [1]

    Cross-media structured common space for multimedia event extraction

    Manling Li, Alireza Zareian, Qi Zeng, Spencer Whitehead, Di Lu, Heng Ji, and Shih-Fu Chang. Cross-media structured common space for multimedia event extraction. In Dan Jurafsky, Joyce Chai, Natalie Schluter, and Joel Tetreault, edi- tors,Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 2557–2568, Online, July ...

    2020

  2. [2]

    Scented-EAE: Stage- customized entity type embedding for event argument extraction

    Yu Yang, Jinyu Guo, Kai Shuang, and Chenrui Mao. Scented-EAE: Stage- customized entity type embedding for event argument extraction. In Lun-Wei Ku, Andre Martins, and Vivek Srikumar, editors,Findings of the Association for Computational Linguistics: ACL 2024, pages 5222–5235, Bangkok, Thailand, August 2024. Association for Computational Linguistics

    2024

  3. [3]

    A graph convolution network with a pos-aware filter and context enhancement mechanism for event detection

    Xintao Jiao, Jiansheng Chen, and Jiale Liu. A graph convolution network with a pos-aware filter and context enhancement mechanism for event detection. In Proceedings of the 2024 International Conference on Multimedia Retrieval, ICMR ’24, page 285–292, New York, NY, USA, 2024. Association for Computing Machinery

    2024

  4. [4]

    Instruction-tuning LLMs for event extraction with annotation guidelines

    Saurabh Srivastava, Sweta Pati, and Ziyu Yao. Instruction-tuning LLMs for event extraction with annotation guidelines. In Wanxiang Che, Joyce Nabende, Ekaterina Shutova, and Mohammad Taher Pilehvar, editors,Findings of the Association for Computational Linguistics: ACL 2025, pages 13055–13071, Vienna, Austria, July 2025. Association for Computational Linguistics

    2025

  5. [5]

    Adaptive schema-aware event extraction with retrieval-augmented generation

    Sheng Liang, Hang Lv, Zhihao Wen, Yaxiong Wu, Yongyue Zhang, Hao Wang, and Yong Liu. Adaptive schema-aware event extraction with retrieval-augmented generation. In Christos Christodoulopoulos, Tanmoy Chakraborty, Carolyn Rose, and Violet Peng, editors,Findings of the Association for Computational Linguistics: EMNLP 2025, pages 7927–7946, Suzhou, China, No...

    2025

  6. [6]

    Multimedia event extraction from news with a unified contrastive learning framework

    Jian Liu, Yufeng Chen, and Jinan Xu. Multimedia event extraction from news with a unified contrastive learning framework. InProceedings of the 30th ACM International Conference on Multimedia, MM ’22, page 1945–1953, New York, NY, USA, 2022. Association for Computing Machinery

    1945

  7. [7]

    Clip-event: Connecting text and images with event structures

    Manling Li, Ruochen Xu, Shuohang Wang, Luowei Zhou, Xudong Lin, Chen- guang Zhu, Michael Zeng, Heng Ji, and Shih-Fu Chang. Clip-event: Connecting text and images with event structures. In2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 16399–16408, 2022

    2022

  8. [8]

    Training multimedia event extraction with generated images and captions

    Zilin Du, Yunxin Li, Xu Guo, Yidan Sun, and Boyang Li. Training multimedia event extraction with generated images and captions. InProceedings of the 31st ACM International Conference on Multimedia, MM ’23, page 5504–5513, New York, NY, USA, 2023. Association for Computing Machinery

    2023

Show all 36 references
  1. [9]

    Cross-modal multi-task learning for multimedia event extraction

    Jianwei Cao, Yanli Hu, Zhen Tan, and Xiang Zhao. Cross-modal multi-task learning for multimedia event extraction. InProceedings of the AAAI Conference on Artificial Intelligence, volume 39, pages 11454–11462, 2025

    2025

  2. [10]

    MMUTF: Multimodal multimedia event argument extraction with unified template filling

    Philipp Seeberger, Dominik Wagner, and Korbinian Riedhammer. MMUTF: Multimodal multimedia event argument extraction with unified template filling. In Yaser Al-Onaizan, Mohit Bansal, and Yun-Nung Chen, editors,Findings of the Association for Computational Linguistics: EMNLP 202...

    2024

  3. [11]

    Multi-grained gradual inference model for multimedia event extraction.IEEE Transactions on Circuits and Systems for Video Technology, 34(10):10507–10520, 2024

    Yang Liu, Fang Liu, Licheng Jiao, Qianyue Bao, Long Sun, Shuo Li, Lingling Li, and Xu Liu. Multi-grained gradual inference model for multimedia event extraction.IEEE Transactions on Circuits and Systems for Video Technology, 34(10):10507–10520, 2024

    2024

  4. [12]

    Stepwise schema-guided prompting framework with param- eter efficient instruction tuning for multimedia event extraction

    Xiang Yuan, Xinrong Chen, Haochen Li, Hang Yang, Guanyu Wang, Weiping Li, and Tong Mo. Stepwise schema-guided prompting framework with param- eter efficient instruction tuning for multimedia event extraction. In2025 IEEE International Conference on Multimedia and Expo (ICME), ...

    2025

  5. [13]

    Multimedia event extraction with LLM knowledge editing

    Jiaao Yu, Yijing Lin, Zhipeng Gao, Xuesong Qiu, and Lanlan Rui. Multimedia event extraction with LLM knowledge editing. In Christos Christodoulopoulos, Tanmoy Chakraborty, Carolyn Rose, and Violet Peng, editors,Proceedings of the 2025 Conference on Empirical Methods in Natural...

    2025

  6. [14]

    Benchmarking and im- proving LVLMs on event extraction from multimedia documents

    Fuyu Xing, Zimu Wang, Wei Wang, and Haiyang Zhang. Benchmarking and im- proving LVLMs on event extraction from multimedia documents. InProceedings of the 18th International Natural Language Generation Conference, pages 734–742, Hanoi, Vietnam, October 2025. Association for Com...

    2025

  7. [15]

    Umie: Unified multimodal information extraction with instruction tuning

    Lin Sun, Kai Zhang, Qingyuan Li, and Renze Lou. Umie: Unified multimodal information extraction with instruction tuning. InProceedings of the AAAI Conference on Artificial Intelligence, volume 38, pages 19062–19070, 2024

    2024

  8. [16]

    Unified structure generation for universal information extraction

    Yaojie Lu, Qing Liu, Dai Dai, Xinyan Xiao, Hongyu Lin, Xianpei Han, Le Sun, and Hua Wu. Unified structure generation for universal information extraction. In Smaranda Muresan, Preslav Nakov, and Aline Villavicencio, editors,Proceed- ings of the 60th Annual Meeting of the Assoc...

    2022

  9. [17]

    Exploring multimodal relation extraction of hierarchical tabular data with multi-task learning

    Xinyu Zhang, Aibo Song, Jingyi Qiu, Jiahui Jin, Tianbo Zhang, and Xiaolin Fang. Exploring multimodal relation extraction of hierarchical tabular data with multi-task learning. In Wanxiang Che, Joyce Nabende, Ekaterina Shutova, and Mohammad Taher Pilehvar, editors,Proceedings o...

    2025

  10. [18]

    Mr-uie: Multi-perspective reasoning with reinforcement learning for universal information extraction, 2025

    Zhongqiu Li, Shiquan Wang, Ruiyu Fang, Mengjiao Bao, Zhenhe Wu, Shuangy- ong Song, Yongxiang Li, and Zhongjiang He. Mr-uie: Multi-perspective reasoning with reinforcement learning for universal information extraction, 2025

    2025

  11. [19]

    Training-free LLM merging for multi-task learning

    Zichuan Fu, Xian Wu, Yejing Wang, Wanyu Wang, Shanshan Ye, Hongzhi Yin, Yi Chang, Yefeng Zheng, and Xiangyu Zhao. Training-free LLM merging for multi-task learning. In Wanxiang Che, Joyce Nabende, Ekaterina Shutova, and Mohammad Taher Pilehvar, editors,Proceedings of the 63rd ...

    2025

  12. [20]

    Mul- timodal relation extraction with efficient graph alignment.Proceedings of the 29th ACM International Conference on Multimedia, 2021

    Changmeng Zheng, Junhao Feng, Ze Fu, Yiru Cai, Qing Li, and Tao Wang. Mul- timodal relation extraction with efficient graph alignment.Proceedings of the 29th ACM International Conference on Multimedia, 2021

    2021

  13. [21]

    REAR: Reinforced reason- ing optimization for event argument extraction with relation-aware support

    Jianwen Luo, Yu Hong, Shuai Yang, and Jianmin Yao. REAR: Reinforced reason- ing optimization for event argument extraction with relation-aware support. In Christos Christodoulopoulos, Tanmoy Chakraborty, Carolyn Rose, and Violet Peng, editors,Findings of the Association for Co...

    2025

  14. [22]

    Yang, Siyuan Li, and Chunsai Du

    Xiao Wang, Wei Zhou, Can Zu, Han Xia, Tianze Chen, Yuan Zhang, Rui Zheng, Junjie Ye, Qi Zhang, Tao Gui, Jihua Kang, J. Yang, Siyuan Li, and Chunsai Du. Instructuie: Multi-task instruction tuning for unified information extraction. ArXiv, abs/2304.08085, 2023

    2023 arXiv

  15. [23]

    The automatic content extraction (ACE) program – tasks, data, and evaluation

    George Doddington, Alexis Mitchell, Mark Przybocki, Lance Ramshaw, Stephanie Strassel, and Ralph Weischedel. The automatic content extraction (ACE) program – tasks, data, and evaluation. In Maria Teresa Lino, Maria Francisca Xavier, Fátima Ferreira, Rute Costa, and Raquel Silv...

    2004

  16. [24]

    Situation recognition: Visual semantic role labeling for image understanding

    Mark Yatskar, Luke Zettlemoyer, and Ali Farhadi. Situation recognition: Visual semantic role labeling for image understanding. In2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 5534–5542, 2016

    2016

  17. [25]

    Multi- modal relation extraction with efficient graph alignment

    Changmeng Zheng, Junhao Feng, Ze Fu, Yi Cai, Qing Li, and Tao Wang. Multi- modal relation extraction with efficient graph alignment. InProceedings of the 29th ACM International Conference on Multimedia, MM ’21, page 5298–5306, New York, NY, USA, 2021. Association for Computing...

    2021

  18. [26]

    Grounded situation recognition

    Sarah Pratt, Mark Yatskar, Luca Weihs, Ali Farhadi, and Aniruddha Kembhavi. Grounded situation recognition. In Andrea Vedaldi, Horst Bischof, Thomas Brox, and Jan-Michael Frahm, editors,Computer Vision – ECCV 2020, pages 314–332, Cham, 2020. Springer International Publishing

    2020

  19. [27]

    Qwen2-vl: Enhancing vision-language model’s perception of the world at any resolution, 2024

    Peng Wang, Shuai Bai, Sinan Tan, Shijie Wang, Zhihao Fan, Jinze Bai, Keqin Chen, Xuejing Liu, Jialin Wang, Wenbin Ge, Yang Fan, Kai Dang, Mengfei Du, Xuancheng Ren, Rui Men, Dayiheng Liu, Chang Zhou, Jingren Zhou, and Junyang Lin. Qwen2-vl: Enhancing vision-language model’s pe...

    2024

  20. [28]

    Qwen3-vl technical report, 2025

    Shuai Bai, Yuxuan Cai, Ruizhe Chen, Keqin Chen, Xionghui Chen, Zesen Cheng, Lianghao Deng, Wei Ding, Chang Gao, Chunjiang Ge, Wenbin Ge, Zhifang Guo, Qidong Huang, Jie Huang, Fei Huang, Binyuan Hui, Shutong Jiang, Zhaohai Li, Mingsheng Li, Mei Li, Kaixin Li, Zicheng Lin, Junya...

    2025

  21. [29]

    Internvl3.5: Advancing open-source multimodal models in versatility, reasoning, and efficiency, 2025

    Weiyun Wang, Zhangwei Gao, Lixin Gu, Hengjun Pu, Long Cui, Xingguang Wei, Zhaoyang Liu, Linglin Jing, Shenglong Ye, Jie Shao, Zhaokai Wang, Zhe Chen, Hongjie Zhang, Ganlin Yang, Haomin Wang, Qi Wei, Jinhui Yin, Wenhao Li, Erfei Cui, Guanzhou Chen, Zichen Ding, Changyao Tian, Z...

    2025

  22. [30]

    Learning transferable visual models from natural language supervision, 2021

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, and Ilya Sutskever. Learning transferable visual models from natural language supervision, 2021

    2021

  23. [31]

    BERT: Pre-training of deep bidirectional transformers for language understanding

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. BERT: Pre-training of deep bidirectional transformers for language understanding. In Jill Burstein, Christy Doran, and Thamar Solorio, editors,Proceedings of the Yongkang Jin, Jianwen Luo, Jingjing Wang, Jianmin...

    2019

  24. [32]

    An image is worth 16x16 words: Transformers for image recognition at scale

    Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xi- aohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, and Neil Houlsby. An image is worth 16x16 words: Transformers for image recognition ...

    2021

  25. [33]

    Yolov8: A novel object detection algorithm with enhanced performance and robustness

    Rejin Varghese and Sambath M. Yolov8: A novel object detection algorithm with enhanced performance and robustness. In2024 International Conference on Advances in Data Engineering and Intelligent Computing Systems (ADICS), pages 1–6, 2024

    2024

  26. [34]

    Chi, Jeff Dean, Jacob Devlin, Adam Roberts, Denny Zhou, Quoc V

    Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Yunxuan Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, Al- bert Webson, Shixiang Shane Gu, Zhuyun Dai, Mirac Suzgun, Xinyun Chen, Aakanksha Chowdhery, Alex Castro-Ros, Marie Pellat, Kevin Robin...

    2022

  27. [35]

    Colin Raffel, Noam Shazeer, Adam Roberts, Katherine Lee, Sharan Narang, Michael Matena, Yanqi Zhou, Wei Li, and Peter J. Liu. Exploring the limits of transfer learning with a unified text-to-text transformer.Journal of Machine Learning Research, 21(140):1–67, 2020

    2020

  28. [36]

    Improved baselines with visual instruction tuning

    Haotian Liu, Chunyuan Li, Yuheng Li, and Yong Jae Lee. Improved baselines with visual instruction tuning. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 26296–26306, June 2024

    2024

Pith tools

Reviewed August 2, 2026 · model on record in the stance chip above.