Pith. sign in

REVIEW 4 major objections 6 minor 42 references

Creative4U: MLLMs-based Advertising Creative Image Selector with Comparative Reasoning

T0 review · 4 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read The paper claims that a vision-language model trained to compare two ad creatives can predict which one users are more likely to click and explain why, beating bandit and ranking baselines offline and lifting CTR by up to 7.07% in…

desk verdict Real online gains and a new dataset make this worth reading, but reporting sloppiness and the unvalidated Q10 label keep it from being a clean accept. read the letter →

arxiv 2508.12628 v2 pith:G4W4FCVT submitted 2025-08-18 cs.CV

Yukang Lin , Xiang Zhang , Shichang Jia , Bowen Wan , Chenghan Fu , Xudong Ren , Yueran Liu , Wanxian Guan
show 4 more authors
Pengji Wang Jian Xu Bo Zheng Baolin Liu
This is my paper · ORCID
classification cs.CV
keywords advertisingcreativeselectionimagequalityassessmentmultimodallargelanguagemodelcomparativereasoningchain-of-thoughtreinforcementlearninge-commerce
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

Creative4U claims that the problem of choosing which advertising image will perform best can be turned into a comparative reasoning task: given two creatives for the same product, the product title, and the user's search terms, a multimodal language model can decide which image users are more likely to click on and produce a step-by-step explanation of the decision. To support this, the paper introduces a dataset of 8,879 paired creatives annotated under a structured Creative Evaluation Protocol, each pair supplied with a chain-of-thought comparison generated by a large vision-language model and audited for consistency. A 7-billion-parameter multimodal model is then trained with supervised fine-tuning followed by group-relative policy optimization with recall and precision rewards, so the model is rewarded both for picking the right image and for covering the comparison dimensions that actually matter. The paper reports offline selection accuracy of 79.1% on held-out pairs, beating four baselines, and an online A/B test in which replacing the production offline selector lifts click-through rate by up to 7.07% relative to random selection. If the claim holds, advertisers facing a flood of AI-generated creatives could select offline, with explainable feedback, instead of spending traffic to learn which image wins.

What carries the argument

The load-bearing mechanism is the reward function in group-relative policy optimization: $R(y_i) = R_{\mathrm{acc}}(y_i) + \alpha_1 R_{\mathrm{format}}(y_i) + \alpha_2 R_{\mathrm{rec}}(y_i) + \alpha_3 R_{\mathrm{prec}}(y_i)$. The two reasoning-level rewards are defined over the sets of protocol questions: $R_{\mathrm{rec}} = |Q_{\mathrm{correct}}|/|Q_{\mathrm{gt}}|$ and $R_{\mathrm{prec}} = |Q_{\mathrm{correct}}|/|Q_{\mathrm{pred}}|$, where $Q_{\mathrm{correct}}$ counts dimensions on which the model's judgement matches the ground-truth annotation, $Q_{\mathrm{gt}}$ is the set of dimensions where the two images actually differ, and $Q_{\mathrm{pred}}$ is the set the model chose to discuss. Recall pushes the chain-of-thought to cover all decision-relevant differences; precision penalizes irrelevant or hallucinated comparisons. The Creative Evaluation Protocol supplies those dimensions in the first place: annotators compare image-text consistency, text in image, models and props, layout, product subject, and background before giving a final 'more likely to click' verdict. GRPO-RPR is this reward design plus a group-relative advantage update, and the paper's ablations show that the recall and precision terms are complementary: removing either one degrades accuracy or F1.

What would settle it

Take a random sample of CreativePair pairs, have two independent annotation teams re-label them, and measure agreement on the final Q10 answer: near-chance agreement would mean the dataset's target is mostly annotation noise. Stronger: log the actual click outcome when a served product shows one creative and a replacement candidate, and test whether the human label predicts which image gets the higher CTR significantly above 0.5.

Watch

Extended reading notes

Core claim

The central claim is that explainable creative selection can be learned end-to-end from paired comparisons, and that the explanation quality is not a by-product but part of what makes the selection accurate. The model's output has two parts: a <think> chain in which the model compares the two creatives along seven protocol dimensions (query hit rate, title hit rate, text in image, models and props, layout, product subject, background), and an <answer> giving the final choice. Training uses a composite reward: an accuracy reward for the final answer, a format reward, and two reasoning-level rewards, where recall is the fraction of ground-truth differing dimensions that the model mentions correctly and precision is the fraction of mentioned dimensions that are correct. The policy is updated by normalizing these rewards within a sampled group. On the CreativePair test set, this reaches 0.791 accuracy, 0.799 recall, and 0.514 F1, ahead of the best baseline at 0.751 accuracy; in production, replacing the existing offline selector with the model's rankings lifts CTR by 7.07% without the downstream ranker and by 3.75% with it, and also lifts CVR and RPM.

Load-bearing premise

The load-bearing premise is that the human annotator's final answer — 'which image the user is more likely to click on' — matches what real users actually click; the paper does not report inter-annotator agreement or label-to-CTR validation, so if that intuition is wrong, the dataset target, the reward, and the offline accuracy all inherit the error.

Editorial extensions

If this is right

  • An e-commerce platform can rank AI-generated creatives offline from pairwise comparisons, avoiding small-traffic online exploration and still improving CTR, CVR, and RPM over both random selection and a bandit selector.
  • Because the selector emits dimension-level rationales, designers and AIGC pipelines receive actionable feedback; in the paper's deployment, iterating on that feedback raised the win rate of AI-generated creatives against human-made ones from 12.2% to 44.5%.
  • The coarse-to-fine deployment strategy keeps the cost of pairwise selection manageable: with up to 100 candidates, one pass against the current creative keeps roughly 12%, so the subsequent all-pairs ranking runs on a small survivor set.
  • The reward decomposition implies that reasoning quality is not decorative: the recall and precision rewards jointly drive decision accuracy, since the ablation without recall drops F1 from 0.514 to 0.376 and the ablation without precision drops accuracy to 0.729.
  • At 0.71 seconds per pair on the hardware used, the selector fits an offline batch pipeline; moving it to real-time serving would require lower latency.

Reading between the lines

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

  • The same recall–precision reward design should transfer to any MLLM task whose chain-of-thought can be checked against a fixed set of dimensions, such as document relevance or diagnostic image triage; the paper only demonstrates it on e-commerce creatives.
  • A natural validation the paper does not report is to compare CreativePair's human labels against the actual click-through rate of the same pairs served in production; if labels predict clicks, the offline metric is a faithful proxy, and if not, the model has been trained to match annotator taste rather than user behaviour.
  • The reported jump in AIGC win rate suggests the rationales could be fed back into the generation model itself as prompts or edit instructions, which the paper did not test; this would close the loop between selection and generation.
  • The two-stage coarse-to-fine selection recipe is a general pattern for converting a pairwise preference model into a ranked list under a candidate budget; it could be reused for any content domain where only pairwise judgments are available.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper proposes an explainable advertising creative image selection paradigm. It introduces CreativePair, a dataset of 8,879 image pairs annotated with the Creative Evaluation Protocol (similarity, image-text consistency, five quality dimensions, and a final Q10 click-likelihood label) and with chain-of-thought reasoning generated by Qwen2.5-VL-72B-Instruct. The proposed Creative4U model is Qwen2.5-VL-7B trained with CoT-SFT followed by GRPO with rewards for format, answer accuracy, and CoT recall/precision. Experiments report offline accuracy 0.791, outperforming several IQA and ranking baselines, and an online A/B test on Taobao reporting CTR/RPM/CVR improvements over random and bandit baselines. The paper also describes a two-stage deployment pipeline and an explainability use case in AIGC feedback.

Significance. If valid, the work is significant: it is a first systematic, explainable approach to creative selection, with a structured protocol and dataset that the community can reuse; the GRPO-RPR reward design is a concrete, reproducible contribution; and the online deployment demonstrates practical value. The central offline claim, however, rests on human Q10 labels that are not validated against real click behavior, and the online evidence lacks statistical grounding. With those gaps addressed, the contribution would be solid.

major comments (4)
  1. [Section 3.2, Q10] The ground-truth labels for the selection task ('which image is the user more likely to click on') are never validated against actual user behavior. The paper reports no inter-annotator agreement, no comparison of Q10 labels with logged click-through rates for the same creative pairs, and no estimate of annotation noise. Since R_acc in Eq. (4) and every offline accuracy in Tables 1-3 depend on these labels, a systematic divergence between annotators' click intuition and real user behavior would directly invalidate the headline 0.791 and the offline comparisons. The online A/B test validates the full pipeline, not the dataset labels, so it cannot by itself rescue the offline claim. Please add annotation-quality metrics (e.g., Cohen's kappa or Krippendorff's alpha) and, if possible, a held-out subset where Q10 labels are compared with observed CTR for the same creatives.
  2. [Section 4.2, Eqs. (5)-(7) and Section 3.2] The CoT ground truth used for the recall and precision rewards, and for the reasoning-level metrics in Tables 1-3, is generated by Qwen2.5-VL-72B-Instruct and then audited by humans with a claimed 95.3% pass rate. This means the reasoning-quality metrics measure agreement with a model-generated protocol, not with independent human reasoning. The paper does not report how many CoT texts were rejected or what the refinements changed, nor whether the audit was performed by multiple annotators with agreement measurement. Please provide a human-written or fully human-verified reasoning subset and report agreement between the model-generated CoT and human reasoning on that subset, so the reasoning-level claims are not circular.
  3. [Section 5.6, Table 4] The online A/B results are reported without confidence intervals or significance tests, so the reader cannot assess whether the CTR/RPM/CVR gaps are meaningful or within noise. Furthermore, the prose contradicts the table: without the online ranker the text reports '2.4% in CTR, 2.08% in RPM, and 0.96% in CVR', while Table 4 lists +7.07%, +6.92%, +2.48% for the same setting; with the online ranker the text says 1.32%/1.34%/0.35% while the table lists +3.75%/+3.86%/+1.69%. Please reconcile these numbers, report traffic allocation and the number of ad sets, and add statistical evaluation (e.g., bootstrap confidence intervals or p-values) for the reported improvements.
  4. [Section 5.1, Table 1, VAM row] The VAM baseline is described as 'a visual-aware creative ranking method', but the original VAM (reference [22]) is a hybrid bandit model that relies on online feedback. The paper states that 'all baselines are trained on CreativePair with their default settings', yet gives no details on how a bandit algorithm is adapted to a static pairwise selection task. Without this adaptation protocol, the reported 0.608 accuracy for VAM is not interpretable, and the claim that Creative4U outperforms ranking-based approaches is weakened. Please specify how VAM was adapted, what its training data and loss were, and how the pairwise comparison was performed at inference.
minor comments (6)
  1. [Abstract and Section 3.2] The abstract says '8k annotated image pairs' while the body reports 8,879 pairs (7,088 training + 1,791 test). Please make the numbers consistent and clarify that filtering reduced the collected 10,000 pairs to 8,879.
  2. [References [32] and [33]] References [32] and [33] appear to be the same work (DepictQA-Wild, arXiv:2405.18842). Please remove the duplicate or cite distinct versions if intended.
  3. [Section 5.4, Table 3] The statement that 'a balanced weighting yields the best trade-off' and that the default setting attains 'the best F1 with near-best accuracy' is slightly misleading: ID 6 in Table 3 has accuracy 0.796, higher than the default 0.791, with comparable precision, at the cost of a small F1 decrease (0.510 vs 0.514). Please rephrase to avoid implying the default setting is strictly dominant.
  4. [Section 5.7] The claim that the win rate of AIGC creatives against human-made ones improved from 12.2% to 44.5% lacks methodology: no description of how the comparison was set up, how many creatives were evaluated, or whether the judges were blind to condition. Please add this information or clearly label the result as an anecdotal report.
  5. [Figure 6 and general presentation] The caption of Figure 6 says green highlights incorrect reasoning steps and red emphasizes correct ones, but the surrounding text discusses models being 'misled' without a consistent color-key. Please check the color assignments for consistency with the figure. Also, 'Image-Text Consistence' in Figure 1 and Algorithm 1 should read 'Consistency'.
  6. [Section 1 and Section 3] The dataset and code are announced as 'will be made public', but no release URL or planned access mechanism is provided. Please include a link or a clear statement about how reviewers can access the data for verification.

Circularity Check

1 steps flagged · score 3.0 of 10

Partial, secondary circularity: reasoning-quality metrics are defined as agreement with a teacher-generated CoT, but the headline selection accuracy rests on human labels and an online A/B test.

  1. self definitional [Sec. 3.2 (Chain-of-Thought generation); Sec. 4.2, Eq. (5)-(7); Sec. 5.1 (reasoning metrics)]
    "we utilize Qwen2.5-VL-72B-Instruct to generate CoT-style reasoning data for each question in the Creative Evaluation Protocol. Importantly, we enforce strict consistency between the generated reasoning and the answer in each question... For a given sample, we denote the ground-truth CoT annotation and the model-generated CoT as CoTgt = ... CoTpred = ..."

    The recall/precision rewards (Eq. 7) and the reported recall/precision/F1 metrics all compare the student's CoT (CoTpred) against CoTgt, which is generated by Qwen2.5-VL-72B-Instruct and only audited, not independently authored, by humans. Because the trained student is also a Qwen2.5-VL model (7B), high reasoning-level scores are, by construction, a measure of agreement with a teacher model of the same family rather than with an externally grounded standard of explanation quality. The paper's claim that Creative4U 'covers a much larger portion of the ground-truth comparison dimensions' therefore reduces to student-teacher agreement for those metrics.

full rationale

The paper's central selection result is not circular: R_acc in Eq. (4) and the accuracy numbers in Tables 1-3 are anchored to human Q10 annotations from the Creative Evaluation Protocol (Algorithm 1), and Table 4's online A/B test supplies independent CTR/RPM/CVR evidence for the deployed pipeline. No load-bearing self-citation or imported uniqueness theorem appears; the self-citations in the reference list are unrelated AIGC works. The only circularity-like step is the reasoning-quality evaluation: CoTgt in Eq. (5) is called 'ground-truth' but was generated by Qwen2.5-VL-72B-Instruct, and the recall/precision rewards and metrics measure how closely the 7B student matches that teacher. This makes the explanation-quality claims self-referential, but it does not by construction force the headline accuracy or the online gains, which are measured against human labels and real clicks respectively. The missing inter-annotator agreement and the prose/table mismatch in online CTR gains are validity and correctness concerns, not circularity.

Assumptions & free parameters 5 free parameters · 3 assumptions · 0 invented entities

The central claim rests on a human-designed evaluation protocol, on teacher-generated CoT used as ground truth for the reasoning rewards, and on an unverified link between human Q10 judgments and real click behavior. These are domain assumptions rather than standard mathematical axioms; the paper provides no external validation for them.

free parameters (5)
  • alpha_1 format reward weight = 0.1
    Weight on format reward; chosen by hand and shown non-critical in Table 3.
  • alpha_2 recall reward weight = 0.5
    Weight on CoT recall reward; selected via sensitivity sweep in Table 3.
  • alpha_3 precision reward weight = 0.5
    Weight on CoT precision reward; selected via sensitivity sweep in Table 3.
  • beta KL regularization weight = 0.001
    KL weight in the GRPO objective (Eq. 9); chosen by hand, not ablated.
  • group size G = 8
    Number of sampled responses per input in GRPO; set to 8, no sensitivity reported.
assumptions (3)
  • ad hoc to paper The Creative Evaluation Protocol dimensions (image-text consistency, text in image, model/props, layout, product subject, background) are a sufficient and faithful abstraction of the factors that determine whether users click an ad creative.
    Introduced in Sec 3.1 and 3.2 from a 10% pre-annotation by designers; no empirical validation shows these dimensions cover the causal drivers of click behavior.
  • domain assumption The Chain-of-Thought rationales generated by Qwen2.5-VL-72B and audited at a 95.3% pass rate are a valid ground truth for reasoning quality.
    The reasoning-level rewards (Eq. 7) and evaluation metrics compare the student model against these teacher-generated rationales, not against independently elicited human reasoning.
  • domain assumption Human annotator Q10 responses, which image the user is more likely to click on, accurately reflect real user click behavior.
    No inter-annotator agreement, annotator quality controls, or calibration of labels against actual CTR data are reported in Sec 3.2; the online A/B test evaluates the full pipeline, not the labels directly.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Creative4U: MLLMs-based Advertising Creative Image Selector with Comparative Reasoning." pith.science (2026). https://pith.science/paper/G4W4FCVT

@misc{pith2026250812628,
  author       = {Pith},
  title        = {Pith review of: Creative4U: MLLMs-based Advertising Creative Image Selector with Comparative Reasoning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/G4W4FCVT}},
  note         = {Machine review of arXiv:2508.12628}
}
read the original abstract

Creative image in advertising is the heart and soul of e-commerce platform. An eye-catching creative image can enhance the shopping experience for users, boosting income for advertisers and advertising revenue for platforms. With the advent of AIGC technology, advertisers can produce large quantities of creative images at minimal cost. However, they struggle to assess the creative quality to select. Existing methods primarily focus on creative ranking, which fails to address the need for explainable creative selection. In this work, we propose the first paradigm for explainable creative assessment and selection. Powered by multimodal large language models (MLLMs), our approach integrates the assessment and selection of creative images into a natural language generation task. To facilitate this research, we construct CreativePair, the first comparative reasoning-induced creative dataset featuring 8k annotated image pairs, with each sample including a label indicating which image is superior. Additionally, we introduce Creative4U (pronounced Creative for You), a MLLMs-based creative selector that takes into account users' interests. Through Reason-to-Select RFT, which includes supervised fine-tuning with Chain-of-Thought (CoT-SFT) and Group Relative Policy Optimization (GRPO) based reinforcement learning, Creative4U is able to evaluate and select creative images accurately. Both offline and online experiments demonstrate the effectiveness of our approach. Our code and dataset will be made public to advance research and industrial applications.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

42 extracted references · 15 canonical work pages

  1. [22]

    Shiyao Wang, Qi Liu, Tiezheng Ge, Defu Lian, and Zhiqiang Zhang. 2021. A hybrid bandit model with visual priors for creative ranking in display advertising. InProceedings of the web conference 2021. 2324–2334

    work page 2021

  2. [1]

    Javad Azimi, Ruofei Zhang, Yang Zhou, Vidhya Navalpakkam, Jianchang Mao, and Xiaoli Fern. 2012. The Impact of Visual Appearance on User Response in Online Display Advertising. arXiv:1202.2158 [cs.HC] https://arxiv.org/abs/1202. 2158

    arXiv 2012

  3. [2]

    Shuai Bai, Keqin Chen, Xuejing Liu, Jialin Wang, Wenbin Ge, Sibo Song, Kai Dang, Peng Wang, Shijie Wang, Jun Tang, et al . 2025. Qwen2. 5-vl technical report.arXiv preprint arXiv:2502.13923(2025)

    arXiv 2025

  4. [3]

    ByteDance Seed. 2026. Seedance 2.0: Advancing Video Generation for World Complexity.arXiv preprint arXiv:2604.14148(2026)

    arXiv 2026

  5. [4]

    Jin Chen, Tiezheng Ge, Gangwei Jiang, Zhiqiang Zhang, Defu Lian, and Kai Zheng. 2021. Efficient Optimal Selection for Composited Advertising Creatives with Tree Structure. arXiv:2103.01453 [cs.IR] https://arxiv.org/abs/2103.01453

    work page Pith review arXiv 2021

  6. [5]

    Patrick Esser, Sumith Kulal, Andreas Blattmann, Rahim Entezari, Jonas Müller, Harry Saini, Yam Levi, Dominik Lorenz, Axel Sauer, Frederic Boesel, Dustin Podell, Tim Dockhorn, Zion English, Kyle Lacey, Alex Goodwin, Yannik Marek, and Robin Rombach. 2024. Scaling Rectified Flow Transformers for High- Resolution Image Synthesis.arXiv preprint arXiv:2403.03206(2024)

    arXiv 2024

  7. [6]

    Shuhao Han, Haotian Fan, Fangyuan Kong, Wenjie Liao, Chunle Guo, Chongyi Li, Radu Timofte, Liang Li, Tao Li, Junhui Cui, Yunqiu Wang, Yang Tai, Jingwei Sun, Yukang Lin, Yan Hong, Chuanbiao Song, Siyuan Li, Jun Lan, Zhichao Zhang, Xinyue Li, Wei Sun, Zicheng Zhang, Yunhao Li, Xiaohong Liu, Guangtao Zhai, Zitong Xu, Huiyu Duan, Jiarui Wang, Guangji Ma, Liu ...

    work page 2025

  8. [7]

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. 2016. Deep residual learning for image recognition. InProceedings of the IEEE conference on computer vision and pattern recognition. 770–778

    2016

Show all 42 references
  1. [8]

    Alex Krizhevsky, Ilya Sutskever, and Geoffrey E Hinton. 2012. Imagenet classifi- cation with deep convolutional neural networks.Advances in neural information processing systems25 (2012)

    2012

  2. [9]

    Kuaishou Technology. 2024. Kling AI: Video Generation Model. https://klingai. kuaishou.com/

    2024

  3. [10]

    Yann LeCun, Yoshua Bengio, and Geoffrey Hinton. 2015. Deep learning.nature 521, 7553 (2015), 436–444

    2015

  4. [11]

    Weiqi Li, Xuanyu Zhang, Shijie Zhao, Yabin Zhang, Junlin Li, Li Zhang, and Jian Zhang. 2025. Q-insight: Understanding image quality via visual reinforcement learning.arXiv preprint arXiv:2503.22679(2025)

    2025 arXiv

  5. [12]

    Yukang Lin, Hokit Fung, Jianjin Xu, Zeping Ren, Adela S. M. Lau, Guosheng Yin, and Xiu Li. 2025. MVPortrait: Text-Guided Motion and Emotion Control for Multi-view Vivid Portrait Animation. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 26242–26252

    2025

  6. [13]

    Yukang Lin, Haonan Han, Chaoqun Gong, Zunnan Xu, Yachao Zhang, and Xiu Li. 2024. Consistent123: One Image to Highly Consistent 3D Asset Using Case- Aware Diffusion Priors. InProceedings of the 32nd ACM International Conference on Multimedia

    2024

  7. [14]

    Yukang Lin, Yan Hong, Zunnan Xu, Xindi Li, Chao Xu, Chuanbiao Song, Ronghui Li, Haoxing Chen, Jun Lan, Huijia Zhu, Weiqiang Wang, Jianfu Zhang, and Xiu Li

  8. [15]

    Daisuke Moriwaki, Komei Fujita, Shota Yasui, and Takahiro Hoshino. 2020. Fatigue-Aware Ad Creative Selection. arXiv:1908.08936 [cs.CY] https://arxiv. org/abs/1908.08936

    2020 arXiv

  9. [16]

    Long Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida, Carroll Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, et al. 2022. Training language models to follow instructions with human feedback.Advances in neural information processing systems35 (...

    2022

  10. [17]

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Björn Ommer. 2022. High-Resolution Image Synthesis with Latent Diffusion Mod- els. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 10684–10695

    2022

  11. [18]

    David E Rumelhart, Geoffrey E Hinton, and Ronald J Williams. 1986. Learning representations by back-propagating errors.nature323, 6088 (1986), 533–536

    1986

  12. [19]

    Zhihong Shao, Peiyi Wang, Qihao Zhu, Runxin Xu, Junxiao Song, Xiao Bi, Haowei Zhang, Mingchuan Zhang, YK Li, Yang Wu, et al. 2024. Deepseekmath: Pushing the limits of mathematical reasoning in open language models.arXiv preprint arXiv:2402.03300(2024)

    2024 arXiv

  13. [20]

    Hamid R Sheikh and Alan C Bovik. 2006. Image information and visual quality. IEEE Transactions on Image Processing15, 2 (2006), 430–444

    2006

  14. [21]

    Karen Simonyan and Andrew Zisserman. 2014. Very deep convolutional net- works for large-scale image recognition.arXiv preprint arXiv:1409.1556(2014)

    2014 arXiv

  15. [23]

    Zhou Wang, Alan C Bovik, Hamid R Sheikh, and Eero P Simoncelli. 2004. Image quality assessment: from error visibility to structural similarity.IEEE Transactions on Image Processing13, 4 (2004), 600–612

    2004

  16. [24]

    Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Fei Xia, Ed Chi, Quoc V Le, Denny Zhou, et al. 2022. Chain-of-thought prompting elicits reason- ing in large language models.Advances in neural information processing systems 35 (2022), 24824–24837

    2022

  17. [25]

    Haoning Wu, Zicheng Zhang, Erli Zhang, Chaofeng Chen, Liang Liao, Annan Wang, Kaixin Xu, Chunyi Li, Jingwen Hou, Guangtao Zhai, Geng Xue, Wenxiu Sun, Qiong Yan, and Weisi Lin. 2023. Q-Instruct: Improving Low-level Visual Abilities for Multi-modality Foundation Models. arXiv:23...

    2023 arXiv

  18. [26]

    Haoning Wu, Zicheng Zhang, Weixia Zhang, Chaofeng Chen, Liang Liao, Chunyi Li, Yixuan Gao, Annan Wang, Erli Zhang, Wenxiu Sun, Qiong Yan, Xiongkuo Min, Guangtao Zhai, and Weisi Lin. 2023. Q-Align: Teaching LMMs for Visual Scoring via Discrete Text-Defined Levels. arXiv:2312.17...

    2023 arXiv

  19. [27]

    Haoning Wu, Hanwei Zhu, Zicheng Zhang, Erli Zhang, Chaofeng Chen, Liang Liao, Chunyi Li, Annan Wang, Wenxiu Sun, Qiong Yan, et al. 2024. Towards open-ended visual quality comparison. InEuropean Conference on Computer Vision. Springer, 360–377

    2024

  20. [28]

    Tianhe Wu, Kede Ma, Jie Liang, Yujiu Yang, and Lei Zhang. 2024. A Comprehen- sive Study of Multimodal Large Language Models for Image Quality Assessment. arXiv:2403.10854 [cs.CV] https://arxiv.org/abs/2403.10854

    2024 arXiv

  21. [29]

    Xianghui Yang, Huiwen Shi, Bowen Zhang, Fan Yang, Jiacheng Wang, Hongxu Zhao, Xinhai Liu, Xinzhou Wang, Qingxiang Lin, Jiaao Yu, Lifu Wang, Zhuo Chen, Sicong Liu, Yuhong Liu, Yong Yang, Di Wang, Jie Jiang, and Chunchao Guo

  22. [30]

    Zhiguang Yang, Lu Wang, Chun Gan, Liufang Sang, Haoran Wang, Wen- long Chen, Jie He, Changping Peng, Zhangang Lin, and Jingping Shao. 2023. Parallel Ranking of Ads and Creatives in Real-Time Advertising Systems. arXiv:2312.12750 [cs.IR] https://arxiv.org/abs/2312.12750

    2023 arXiv

  23. [31]

    Zhiyuan You, Xin Cai, Jinjin Gu, Tianfan Xue, and Chao Dong. 2025. Teaching Large Language Models to Regress Accurate Image Quality Scores using Score Distribution. arXiv:2501.11561 [cs.CV] https://arxiv.org/abs/2501.11561

    2025

  24. [33]

    Zhiyuan You, Jinjin Gu, Zheyuan Li, Xin Cai, Kaiwen Zhu, Chao Dong, and Tianfan Xue. 2024. Descriptive image quality assessment in the wild.arXiv preprint arXiv:2405.18842(2024)

    2024

  25. [34]

    Zhiyuan You, Zheyuan Li, Jinjin Gu, Zhenfei Yin, Tianfan Xue, and Chao Dong

  26. [35]

    Qiying Yu, Zheng Zhang, Ruofei Zhu, Yufeng Yuan, Xiaochen Zuo, Yu Yue, Weinan Dai, Tiantian Fan, Gaohong Liu, Lingjun Liu, et al. 2025. Dapo: An open- source llm reinforcement learning system at scale.arXiv preprint arXiv:2503.14476 (2025)

    2025 arXiv

  27. [36]

    Wei Zhang, Ping Zhang, Jian Dong, Yongkang Wang, Pengye Zhang, Bo Zhang, Xingxing Wang, and Dong Wang. 2023. Cross-Element Combinatorial Selection for Multi-Element Creative in Display Advertising. arXiv:2307.01593 [cs.IR] https://arxiv.org/abs/2307.01593

    2023 arXiv

  28. [37]

    arXiv:2312.08962 [cs.CV] https://arxiv.org/abs/ 2312.08962

    Depicting Beyond Scores: Advancing Image Quality Assessment through Multi-modal Language Models. arXiv:2312.08962 [cs.CV] https://arxiv.org/abs/ 2312.08962

    arXiv

  29. [38]

    Zhichen Zhao, Lei Li, Bowen Zhang, Meng Wang, Yuning Jiang, Li Xu, Fengkun Wang, and Weiying Ma. 2019. What you look matters? offline evaluation of advertising creatives for cold-start problem. InProceedings of the 28th ACM international conference on information and knowledge...

    2019

  30. [39]

    Chujie Zheng, Shixuan Liu, Mingze Li, Xiong-Hui Chen, Bowen Yu, Chang Gao, Kai Dang, Yuqiong Liu, Rui Men, An Yang, et al. 2025. Group sequence policy optimization.arXiv preprint arXiv:2507.18071(2025)

    2025 arXiv

  31. [40]

    Yuze Zhao, Jintao Huang, Jinghan Hu, Xingjun Wang, Yunlin Mao, Daoze Zhang, Zeyinzi Jiang, Zhikai Wu, Baole Ai, Ang Wang, Wenmeng Zhou, and Yingda Chen. 2024. SWIFT:A Scalable lightWeight Infrastructure for Fine-Tuning. arXiv:2408.05517 [cs.CL] https://arxiv.org/abs/2408.05517

    2024 arXiv

  32. [43]

    Hanwei Zhu, Haoning Wu, Yixuan Li, Zicheng Zhang, Baoliang Chen, Lingyu Zhu, Yuming Fang, Guangtao Zhai, Weisi Lin, and Shiqi Wang. 2024. Adaptive Image Quality Assessment via Teaching Large Multimodal Model to Compare. arXiv:2405.19298 [cs.CV] https://arxiv.org/abs/2405.19298

    2024 arXiv

  33. [2024]

    Hunyuan3D 1.0: A Unified Framework for Text-to-3D and Image-to-3D Generation.arXiv preprint arXiv:2411.02293(2024)

    2024 arXiv

  34. [2025]

    InProceedings of the 33rd ACM International Conference on Multimedia

    InterAnimate: Taming Region-aware Diffusion Model for Realistic Human Interaction Animation. InProceedings of the 33rd ACM International Conference on Multimedia

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.