Pith. sign in

REVIEW 3 major objections 8 minor 40 references

TNG-CLIP:Training-Time Negation Data Generation for Negation Awareness of CLIP

T0 review · 3 major / 8 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read Training-time negation caption generation makes CLIP negation-aware and sets state-of-the-art results across matching, retrieval, and generation.

desk verdict A clever training-time negation data-augmentation trick with strong empirical results, but the training labels are unverified and the abstract overclaims retrieval SOTA. read the letter →

arxiv 2505.18434 v1 pith:7TYY2XVB submitted 2025-05-24 cs.CV cs.AI

Yuliang Cai , Jesse Thomason , Mohammad Rostami This is my paper
classification cs.CVcs.AI
keywords negationunderstandingCLIPfine-tuningcontrastivelearningimage-textmatchingtext-to-imageretrievalgenerationtraining-timedataaugmentationbenchmark
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

TNG-CLIP is a fine-tuning procedure that makes CLIP negation-aware without any pre-built negation dataset or per-sample LLM calls. During each training batch, the method finds the visually most similar image-caption pair for every example, extracts a noun from that neighbor's caption, and uses template patterns to synthesize two kinds of negation captions: a compositional one (the original caption, but not the noun) and a full negation of some other caption in the batch. These dynamically generated captions are fed into an asymmetric contrastive objective, adding about 2.5% training time. The paper also contributes Neg-TtoI, a 2000-prompt benchmark for text-to-image generation under negation, and reports state-of-the-art results on image-to-text matching, text-to-image retrieval, and generation benchmarks. If correct, negation understanding can be improved cheaply and without reliance on LLM-generated training data.

What carries the argument

The central mechanism is batch-internal, template-based negation caption generation coupled with an asymmetric noise-augmented contrastive loss. The generator identifies a negation object from a visually similar neighbor's caption using cosine similarity, part-of-speech tagging, and semantic-distance filtering; then it applies one of 46 compositional or 18 full-negation patterns to build captions. Because the neighbor and the template are chosen per batch and per epoch, the same image receives different negation captions across training. The objective matters as much as the generation: the text-to-image term aligns original and negation captions with the image, while the image-to-text term uses random label noise, motivated by the claim that negation captions are out-of-distribution for pretrained CLIP, so that fine-tuning does not overfit. The Neg-TtoI benchmark, built from MS-COCO captions via LLM rewriting and question-answer verification, supplies the first text-to-image negation evaluation.

What would settle it

Run an object detector over query images during training and compute the fraction of generated compositional captions whose negated object is actually present in the image; if TNG-CLIP's gains depend on that fraction, the value of the method rests on label correctness. A direct test is to filter out all generated captions with a false negation and retrain: if accuracy does not drop, the method was learning from false labels.

Watch

Extended reading notes

Core claim

The paper's central discovery claim is that negation understanding in CLIP can be trained from negation captions synthesized on the fly from other captions in the same batch, instead of from a fixed, LLM-generated negation dataset. For each image-text pair, the pipeline embeds the batch images, picks the most similar image by cosine similarity, extracts noun phrases from that image's caption with a part-of-speech tagger, filters the noun through a lexical network to keep it semantically distant from the original caption, and inserts it into a randomly chosen negation template to form a compositional negation caption. A second full-negation caption is formed by negating an unrelated caption from another random pair. Each image is then paired with its original caption, its compositional negation, and a full negation from another sample. Using an asymmetric objective, the text-to-image direction aligns all three captions to the image while the image-to-text direction injects random label noise, and the visual encoder is frozen throughout. The paper reports that this procedure outperforms existing negation-aware CLIP variants on NegBench and Valse-Existence, and that plugging the fine-tuned text encoder into a latent diffusion model improves generation on the proposed Neg-TtoI benchmark.

Load-bearing premise

The load-bearing premise is that the noun selected from the most similar image's caption is genuinely absent from the original image, but the pipeline never verifies this absence, so false negation labels can be fed into training.

Editorial extensions

If this is right

  • Fine-tuning with dynamically generated negation captions costs only about 2.5% extra training time, so the method removes the main computational barrier to negation-aware CLIP.
  • A single text-encoder fine-tune transfers to image generation: replacing a diffusion model's text encoder with TNG-CLIP improves accuracy on Neg-TtoI over the original model and over CoN-CLIP.
  • Dynamic per-epoch negation captions outperform a fixed dataset of the same generated captions (51.61 vs 49.52 average accuracy on NegBench-MSCOCO matching), suggesting that diversity itself helps generalization.
  • The proposed Neg-TtoI benchmark gives a reusable 2000-prompt test for whether generative models both include the desired objects and omit the negated ones.
  • Zero-shot image classification remains near pretrained-CLIP levels, indicating the negation fine-tune does not cause catastrophic forgetting.

Reading between the lines

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

  • The unverified negation label is the soft spot: the noun borrowed from the most similar image's caption may already be present in the query image but unmentioned, so some training captions are false. A cheap detector-based or caption-consistency filter on the selected noun would test whether removing those false labels improves or changes the results.
  • Because part of the gain is attributed to random label noise in the image-to-text loss, the method suggests that negation fine-tuning succeeds by preventing overfitting to out-of-distribution text rather than only by adding negation examples; future work could separate these two effects by adding noise to an equal number of non-negation captions.
  • The same training-time construction could extend beyond image-text pairs to other contrastive or generative objectives, for instance visual question answering where a question about an absent object needs an aligned negative answer.
  • Neg-TtoI's LLM-based judge checks surface semantics; harder negatives that confuse the generator with visually similar distractor objects would reveal whether the model truly understands absence or only avoids the literal word.
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 / 8 minor

Summary. The paper proposes TNG-CLIP, a method for improving negation understanding in CLIP by generating negation captions dynamically during training. For each image-text pair in a batch, the pipeline finds the visually most similar pair via cosine similarity of frozen image features, extracts a noun from the similar pair's caption, and uses template-based rewriting to produce a compositional negation caption (e.g., "A dog playing a ball, but not a boy around") and a full negation caption sampled from another random pair. Each image is then associated with three captions (original, compositional negation, full negation) and trained with an asymmetric objective: the text-to-image loss aligns all three captions with the correct image, while the image-to-text loss uses randomly assigned labels as noise. The paper also introduces NEG-TTOI, a benchmark of 2000 text-to-image generation prompts with positive and negative questions. Experiments on NegBench, VALSE-Existence, and image generation with Stable Diffusion report strong gains over prior negation-aware CLIP baselines, with a reported 2.5% training-time overhead.

Significance. If the reported results hold, the paper's contribution is practically valuable: it offers a way to improve negation understanding with no per-sample LLM calls, no pre-generated negation dataset, and low training overhead, while extending evaluation to text-to-image generation. The paper also ships useful ablation evidence for the dynamic-data design and for the asymmetric noise objective, and it provides a new benchmark that could support future work. The central idea is plausible, but the correctness of the automatically generated training labels is not verified, which is a load-bearing gap; the abstract also overclaims retrieval-state-of-the-art. These issues are fixable, but they need to be addressed before the claims can be accepted.

major comments (3)
  1. [Sections 3.1.1-3.1.4, Eq. (2), and Section 3.2] The training-time negation labels are generated without verifying that the negated object is actually absent from the paired image. For the compositional negation caption, the negation object O_n is extracted from the caption of the most similar image-text pair P_s (Eq. 2) and then asserted to be absent from I_i, but neither the pipeline nor any reported analysis checks this. For the full negation caption, a random other sample's caption is paired with I_i without checking that the negated scene is absent. Since the text-to-image loss in Section 3.2 directly aligns these captions with I_i, false labels such as "not a boy around" on an image that contains a boy are trained as correct supervision. Similar images often share common objects, and image captions routinely omit visible objects, so this is a concrete risk rather than a hypothetical one. The paper should provide a quantitative audit of label correctness on a sample of generated captions, or add a filtering/verification step, before claiming that TNG-CLIP learns correct negation semantics.
  2. [Abstract and Table 1] The abstract claims state-of-the-art performance on "image-to-text matching, text-to-image retrieval, and image generation," but Table 1 shows that TNG-CLIP's Neg-R@5 on text-to-image retrieval is 61.11, lower than NegCLIP's 64.41. The main text itself states that TNG-CLIP "slightly underperforms the NegCLIP model on retrieval tasks." The abstract should be corrected to avoid claiming retrieval state-of-the-art, or the claim should be restricted to the matching and generation tasks.
  3. [Section 3.2, Eq. (4)] The random-label image-to-text objective is central to the proposed training recipe, but it is justified only by the informal argument that negation captions are out-of-distribution and that noise prevents overfitting. Table 6 provides an ablation showing that removing the noise degrades retrieval, which is useful, but the design is not compared with a more principled alternative such as using the original captions for image-to-text alignment with a weighting scheme or a temperature-scaled soft label. As written, the reader cannot tell whether the benefit comes from the specific noise mechanism or from simply reducing the strength of the image-to-text alignment. A targeted comparison would strengthen the paper's central training contribution.
minor comments (8)
  1. [Section 3.1.4] Typo: "denoated" should be "denoted".
  2. [Section 6] Typo: "mdoel's" should be "model's".
  3. [Section 5.1, Valse-Existence bullet] Typo: "imaget-to-text" should be "image-to-text".
  4. [Table 8 caption] Typo: "Generaiton" should be "Generation".
  5. [Appendix A.5] The sentence "we replace thecapwith the provided original caption" should read "we replace {cap} with the provided original caption" for clarity.
  6. [References] Related work cites "Truong et al." and "Zhang et al." without years or reference list entries; please add the full citations. Also, Alhamoud et al. 2025a and 2025b point to the same arXiv preprint; please disambiguate.
  7. [Footnote 1] The GitHub repository link is a placeholder text "here"; provide the actual URL or remove the footnote.
  8. [Tables 1-3, 5] Main results are reported without variance or significance tests; given the small margins in some comparisons, adding error bars or repeated-run statistics would improve confidence.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: TNG-CLIP's training-time negation generation is self-contained and evaluated on external benchmarks.

full rationale

The paper's derivation chain does not reduce to its inputs. The negation captions are generated from in-batch image-text pairs using cosine similarity of frozen visual embeddings and template-based composition (Sections 3.1.1-3.1.4), but the target evaluations (NegBench, Valse-Existence, NEG-TTOI) are external benchmarks not constructed from the training-time templates or the model's own outputs. The training objective (Section 3.2) is a modified contrastive loss with asymmetric noise; it is not fitted to test labels. The proposed NEG-TTOI benchmark is generated independently via GPT-4o from MS-COCO captions with explicit LLM verification, and evaluation uses GPT-4o judgments; while LLM-based verification introduces possible label noise, this is an empirical quality concern rather than a circular reduction. The only self-citations (Cai et al., 2023, in Related Works) are cited as general context for dynamic dataset training and are not load-bearing for the claimed results. No equation reduces to another by construction, and no fitted parameter is renamed as a prediction. The unverified absence of negation objects in training captions is a validity limitation noted by the skeptical review, but it does not make the paper's central claim equivalent to its inputs. The Limitations section acknowledges scope limits but does not assert a circular dependency.

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

The method introduces no new physical entities. It relies on a set of domain assumptions about the quality of automatically generated negation captions and about the transferability of the fine-tuned text encoder. The random-label objective is an ad hoc design choice, supported only by ablations.

free parameters (3)
  • Batch size = 128
    Chosen for GPU memory; affects contrastive batch negatives and the distribution of similar-image pairs.
  • Learning rate = 5e-6
    Chosen for stable fine-tuning; no schedule described.
  • Number of negation templates = 46 compositional, 18 full
    Fixed set generated by GPT-4o; diversity of generated captions depends on this set.
assumptions (5)
  • domain assumption Pre-trained CLIP visual features identify the most similar image whose caption contains a plausible absent object for the query image.
    Used in Section 3.1.1 to select the source pair; if visual similarity does not imply co-occurrence of objects, the negation object is arbitrary.
  • domain assumption Nouns extracted by NLTK and the WordNet 'furthest semantics' scoring yield a negation object that is absent in the query image.
    Section 3.1.2; no verification step. False captions can be generated if the object appears in the query image.
  • domain assumption GPT-4o-generated negation templates are semantically correct and diverse.
    Section 3.1.3; templates are fixed, but if some are malformed, the training signal degrades.
  • domain assumption Freezing the visual encoder preserves the image embedding space so the text encoder can be directly swapped into Stable Diffusion.
    Section 3.2 and 5.2; assumes the fine-tuned text encoder still aligns with the frozen image features.
  • ad hoc to paper Random-label image-to-text loss serves as benign noise that prevents overfitting to out-of-distribution negation captions.
    Introduced in Section 3.2; ablation supports it, but the mechanism is not proven.

how reviews work

0 comments
Cite this review

Pith. "Pith review of TNG-CLIP:Training-Time Negation Data Generation for Negation Awareness of CLIP." pith.science (2026). https://pith.science/paper/7TYY2XVB

@misc{pith2026250518434,
  author       = {Pith},
  title        = {Pith review of: TNG-CLIP:Training-Time Negation Data Generation for Negation Awareness of CLIP},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7TYY2XVB}},
  note         = {Machine review of arXiv:2505.18434}
}
read the original abstract

Vision-language models (VLMs), such as CLIP, have demonstrated strong performance across a range of downstream tasks. However, CLIP is still limited in negation understanding: the ability to recognize the absence or exclusion of a concept. Existing methods address the problem by using a large language model (LLM) to generate large-scale data of image captions containing negation for further fine-tuning CLIP. However, these methods are both time- and compute-intensive, and their evaluations are typically restricted to image-text matching tasks. To expand the horizon, we (1) introduce a training-time negation data generation pipeline such that negation captions are generated during the training stage, which only increases 2.5% extra training time, and (2) we propose the first benchmark, Neg-TtoI, for evaluating text-to-image generation models on prompts containing negation, assessing model's ability to produce semantically accurate images. We show that our proposed method, TNG-CLIP, achieves SOTA performance on diverse negation benchmarks of image-to-text matching, text-to-image retrieval, and image generation.

Figures

Figures reproduced from arXiv: 2505.18434 by the authors.

Figure 1
Figure 1. We present TNG-CLIP, a negation-aware CLIP that achieves outstanding negation understanding in image-to-text matching, text-to-image retrieval and proposed image generation NEG-TTOI benchmarks. presence or absence of specific elements critically alters meanings. To tackle this problem, current methods (Al￾hamoud et al., 2025a; Singh et al., 2024; Park et al., 2025; Yuksekgonul et al., 2023) focus on gener￾ating well… view at source ↗
Figure 2
Figure 2. Training Procedure of TNG-CLIP. The diagram shows the data generation pipeline during the training for one sample in the batch. For an image-text pair, Po, the most similar image pair, Ps is selected by the cosine similarity of their embedded image features. The captions from Po and Ps are used to find the negation object and generate two types of negation captions. The final image-text set, Si , for i th image-text… view at source ↗
Figure 3
Figure 3. The zero shot image classification accuracy [PITH_FULL_IMAGE:figures/full_fig_p013_3.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

40 extracted references · 7 canonical work pages

  1. [1]

    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 INTEGERS output.state before.all mid.sentence after.sentence after.block STRING...

  2. [2]

    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 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [4]

    Kumail Alhamoud, Shaden Alshammari, Yonglong Tian, Guohao Li, Philip Torr, Yoon Kim, and Marzyeh Ghassemi. 2025 b . https://arxiv.org/abs/2501.09425 Vision-language models do not understand negation . Preprint, arXiv:2501.09425

    arXiv 2025

  4. [5]

    O'Reilly Media, Inc

    Steven Bird, Ewan Klein, and Edward Loper. 2009. Natural language processing with Python: analyzing text with the natural language toolkit. " O'Reilly Media, Inc."

    2009

  5. [6]

    Tom B. Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, Sandhini Agarwal, Ariel Herbert-Voss, Gretchen Krueger, Tom Henighan, Rewon Child, Aditya Ramesh, Daniel M. Ziegler, Jeffrey Wu, Clemens Winter, and 12 others. 2020. https://arxiv.org/abs/2005.14165 Lan...

    arXiv 2020

  6. [7]

    Maximilian Böther, Ties Robroek, Viktor Gsteiger, Robin Holzinger, Xianzhe Ma, Pınar Tözün, and Ana Klimovic. 2025. https://doi.org/10.1145/3709705 Modyn: Data-centric machine learning pipeline orchestration . Proceedings of the ACM on Management of Data, 3(1):1–30

    doi:10.1145/3709705 2025

  7. [8]

    Yuliang Cai, Jesse Thomason, and Mohammad Rostami. 2023. Task-attentive transformer architecture for continual learning of vision-and-language tasks using knowledge distillation. In Findings of the Association for Computational Linguistics: EMNLP 2023, pages 6986--7000

    work page 2023

  8. [9]

    Soravit Changpinyo, Piyush Sharma, Nan Ding, and Radu Soricut. 2021. https://arxiv.org/abs/2102.08981 Conceptual 12m: Pushing web-scale image-text pre-training to recognize long-tail visual concepts . Preprint, arXiv:2102.08981

    arXiv 2021

Show all 40 references
  1. [10]

    Hao Chen, Zihan Wang, Ran Tao, Hongxin Wei, Xing Xie, Masashi Sugiyama, Bhiksha Raj, and Jindong Wang. 2025. https://arxiv.org/abs/2403.06869 Impact of noisy supervision in foundation model learning . Preprint, arXiv:2403.06869

    2025 arXiv

  2. [11]

    Lawrence Zitnick

    Xinlei Chen, Hao Fang, Tsung-Yi Lin, Ramakrishna Vedantam, Saurabh Gupta, Piotr Dollar, and C. Lawrence Zitnick. 2015. https://arxiv.org/abs/1504.00325 Microsoft coco captions: Data collection and evaluation server . Preprint, arXiv:1504.00325

    2015 arXiv

  3. [12]

    Ziheng Cheng, Zhong Li, and Jiang Bian. 2025. https://openreview.net/forum?id=2CflgSMLoK Data-efficient training by evolved sampling

    2025

  4. [13]

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

    2022 arXiv

  5. [14]

    Dumitru, Ian Goodfellow, Will Cukierski, and Yoshua Bengio. 2013. Challenges in representation learning: Facial expression recognition challenge. https://kaggle.com/competitions/challenges-in-representation-learning-facial-expression-recognition-challenge. Kaggle

    2013

  6. [15]

    Everingham, L

    M. Everingham, L. Van Gool, C. K. I. Williams, J. Winn, and A. Zisserman. The PASCAL V isual O bject C lasses C hallenge 2007 (VOC2007) R esults. http://www.pascal-network.org/challenges/VOC/voc2007/workshop/index.html

    2007

  7. [16]

    Atticus Geiger, Kyle Richardson, and Christopher Potts. 2020. https://doi.org/10.18653/v1/2020.blackboxnlp-1.16 Neural natural language inference models partially embed theories of lexical entailment and negation . In Proceedings of the Third BlackboxNLP Workshop on Analyzing ...

    2020 doi

  8. [17]

    Micah Hodosh, Peter Young, and Julia Hockenmaier. 2013. Framing image description as a ranking task: data, models and evaluation metrics. J. Artif. Int. Res., 47(1):853–899

    2013

  9. [18]

    Md Mosharaf Hossain, Venelin Kovatchev, Pranoy Dutta, Tiffany Kao, Elizabeth Wei, and Eduardo Blanco. 2020. https://doi.org/10.18653/v1/2020.emnlp-main.732 An analysis of natural language inference benchmarks through the lens of negation . In Proceedings of the 2020 Conference...

    2020 doi

  10. [19]

    Yushi Hu, Benlin Liu, Jungo Kasai, Yizhong Wang, Mari Ostendorf, Ranjay Krishna, and Noah A Smith. 2023. https://arxiv.org/abs/2303.11897 Tifa: Accurate and interpretable text-to-image faithfulness evaluation with question answering . Preprint, arXiv:2303.11897

    2023 arXiv

  11. [20]

    Yiding Jiang, Allan Zhou, Zhili Feng, Sadhika Malladi, and J Zico Kolter. 2024. Adaptive data optimization: Dynamic sample selection with scaling laws. arXiv preprint arXiv:2410.11820

    2024 arXiv

  12. [21]

    Alex Krizhevsky. 2009. https://api.semanticscholar.org/CorpusID:18268744 Learning multiple layers of features from tiny images

    2009

  13. [22]

    Ananya Kumar, Aditi Raghunathan, Robbie Jones, Tengyu Ma, and Percy Liang. 2022. https://arxiv.org/abs/2202.10054 Fine-tuning can distort pretrained features and underperform out-of-distribution . Preprint, arXiv:2202.10054

    2022 arXiv

  14. [23]

    George A. Miller. 1995. https://doi.org/10.1145/219717.219748 Wordnet: a lexical database for english . Commun. ACM, 38(11):39–41

    1995

  15. [24]

    OpenAI, :, Aaron Hurst, Adam Lerer, Adam P. Goucher, Adam Perelman, Aditya Ramesh, Aidan Clark, AJ Ostrow, Akila Welihinda, Alan Hayes, Alec Radford, Aleksander Mądry, Alex Baker-Whitcomb, Alex Beutel, Alex Borzunov, Alex Carney, Alex Chow, Alex Kirillov, and 401 others. 2024....

    2024 arXiv

  16. [25]

    Long Ouyang, Jeff Wu, Xu Jiang, Diogo Almeida, Carroll L. Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, John Schulman, Jacob Hilton, Fraser Kelton, Luke Miller, Maddie Simens, Amanda Askell, Peter Welinder, Paul Christiano, Jan Leike, and...

    2022 arXiv

  17. [26]

    Letitia Parcalabescu, Michele Cafagna, Lilitta Muradjan, Anette Frank, Iacer Calixto, and Albert Gatt. 2022. https://doi.org/10.18653/v1/2022.acl-long.567 Valse: A task-independent benchmark for vision and language models centered on linguistic phenomena . In Proceedings of th...

    2022 doi

  18. [27]

    Junsung Park, Jungbeom Lee, Jongyoon Song, Sangwon Yu, Dahuin Jung, and Sungroh Yoon. 2025. https://arxiv.org/abs/2501.10913 Know "no'' better: A data-driven approach for enhancing negation awareness in clip . Preprint, arXiv:2501.10913

    2025

  19. [28]

    Plummer, Liwei Wang, Chris M

    Bryan A. Plummer, Liwei Wang, Chris M. Cervantes, Juan C. Caicedo, Julia Hockenmaier, and Svetlana Lazebnik. 2016. https://arxiv.org/abs/1505.04870 Flickr30k entities: Collecting region-to-phrase correspondences for richer image-to-sentence models . Preprint, arXiv:1505.04870

    2016 arXiv

  20. [29]

    Vincent Quantmeyer, Pablo Mosteiro, and Albert Gatt. 2024. https://arxiv.org/abs/2407.10488 How and where does clip process negation? Preprint, arXiv:2407.10488

    2024 arXiv

  21. [30]

    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. 2021. https://arxiv.org/abs/2103.00020 Learning transferable visual models from natural lan...

    2021 arXiv

  22. [31]

    David Rolnick, Andreas Veit, Serge Belongie, and Nir Shavit. 2018. https://arxiv.org/abs/1705.10694 Deep learning is robust to massive label noise . Preprint, arXiv:1705.10694

    2018 arXiv

  23. [32]

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Björn Ommer. 2022. https://arxiv.org/abs/2112.10752 High-resolution image synthesis with latent diffusion models . Preprint, arXiv:2112.10752

    2022 arXiv

  24. [33]

    no" to say

    Jaisidh Singh, Ishaan Shrivastava, Mayank Vatsa, Richa Singh, and Aparna Bharati. 2024. https://arxiv.org/abs/2403.20312 Learn "no" to say "yes" better: Improving vision-language models via negations . Preprint, arXiv:2403.20312

    2024 arXiv

  25. [34]

    Thinh Hung Truong, Timothy Baldwin, Karin Verspoor, and Trevor Cohn. 2023. https://arxiv.org/abs/2306.08189 Language models are not naysayers: An analysis of language models on negation benchmarks . Preprint, arXiv:2306.08189

    2023 arXiv

  26. [35]

    Thinh Hung Truong, Yulia Otmakhova, Timothy Baldwin, Trevor Cohn, Jey Han Lau, and Karin Verspoor. 2022. https://doi.org/10.18653/v1/2022.aacl-main.65 Not another negation benchmark: The N a N - NLI test suite for sub-clausal negation . In Proceedings of the 2nd Conference of ...

    2022 doi

  27. [36]

    Neeraj Varshney, Satyam Raj, Venkatesh Mishra, Agneet Chatterjee, Ritika Sarkar, Amir Saeidi, and Chitta Baral. 2024. https://arxiv.org/abs/2406.05494 Investigating and addressing hallucinations of llms in tasks involving negation . Preprint, arXiv:2406.05494

    2024 arXiv

  28. [37]

    Rui Wang, Masao Utiyama, and Eiichiro Sumita. 2019. https://arxiv.org/abs/1805.00178 Dynamic sentence sampling for efficient training of neural machine translation . Preprint, arXiv:1805.00178

    2019 arXiv

  29. [38]

    Ehinger, Aude Oliva, and Antonio Torralba

    Jianxiong Xiao, James Hays, Krista A. Ehinger, Aude Oliva, and Antonio Torralba. 2010. https://doi.org/10.1109/CVPR.2010.5539970 Sun database: Large-scale scene recognition from abbey to zoo . In 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition,...

    2010

  30. [39]

    Qizhe Xie, Minh-Thang Luong, Eduard Hovy, and Quoc V. Le. 2020. https://arxiv.org/abs/1911.04252 Self-training with noisy student improves imagenet classification . Preprint, arXiv:1911.04252

    2020 arXiv

  31. [40]

    Mert Yuksekgonul, Federico Bianchi, Pratyusha Kalluri, Dan Jurafsky, and James Zou. 2023. https://arxiv.org/abs/2210.01936 When and why vision-language models behave like bags-of-words, and what to do about it? Preprint, arXiv:2210.01936

    2023 arXiv

  32. [41]

    HaoChen, James Zou, Percy Liang, and Serena Yeung

    Yuhui Zhang, Michihiro Yasunaga, Zhengping Zhou, Jeff Z. HaoChen, James Zou, Percy Liang, and Serena Yeung. 2023. https://arxiv.org/abs/2305.17311 Beyond positive scaling: How negation impacts scaling trends of language models . Preprint, arXiv:2305.17311

    2023 arXiv

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.