Pith. sign in

REVIEW 4 major objections 5 minor 47 references

LogicAD: Explainable Anomaly Detection via VLM-based Text Feature Extraction

T0 review · 4 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read LogicAD claims that a one-shot pipeline converting a single normal image into guided text descriptions, embedding those descriptions, and checking them with an automated theorem prover achieves state-of-the-art logical anomaly detection…

desk verdict An interesting but overclaimed pipeline: the one-shot SOTA result rests on per-category hand-crafted prompts that may encode test-set knowledge, so the evidence does not yet support the headline. read the letter →

arxiv 2501.01767 v2 pith:VN6AX2V4 submitted 2025-01-03 cs.CV

Er Jin , Qihui Feng , Yongli Mou , Stefan Decker , Gerhard Lakemeyer , Oliver Simons , Johannes Stegmaier This is my paper · ORCID
classification cs.CV
keywords logicalanomalydetectionvision-languagemodelstextfeatureextractionone-shotlearningautomatedtheoremprovingexplainableMVTecLOCOADchain-of-thoughtprompting
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 tries to establish that the text output of modern vision-language models, rather than their visual features, can carry the full weight of logical anomaly detection in industrial images. With one normal image per category, LogicAD turns that image into structured text descriptions via guided chain-of-thought prompts, embeds the descriptions for a similarity score, and formalizes them into first-order logic for a theorem-prover check. On the MVTec LOCO AD benchmark it reports an AUROC of 86.0% and F1-max of 83.7%, exceeding the previous few-shot method by 18.1 and 4.6 points, and it produces a minimal logical explanation for every anomaly it flags. This matters because current logical anomaly detection methods tend to need full training sets, visual annotations, or both; a text-based, training-free route would remove those costs.

What carries the argument

The central machinery is the 'text feature' representation: instead of storing visual patch embeddings, LogicAD stores filtered, formatted, and logic-formalized text descriptions of a single normal image. Three components carry the argument: (1) fi2t text extraction, which uses Guided CoT prompts, GroundingDINO region cropping, and LOF filtering to make AVLM output stable; (2) format embedding, which normalizes the text into JSON and scores anomaly as 1 − cosine similarity; (3) the logic reasoner, which converts the JSON into a first-order logic theory Γ and invokes an automated theorem prover to check Γ |= ¬Σ0, extracting a minimal contradictory subset as explanation. The load-bearing identity is that textual descriptions can be compared both geometrically (embedding similarity) and symbolically (theorem proving) to decide normality.

What would settle it

Run a held-out one-shot evaluation on MVTec LOCO AD where the Guided CoT prompt and the normal logic specification for each category are written by someone who has seen only that category's single normal training image; if the AUROC does not stay near 86.0 (and above AnomalyMoE's 67.9), the SOTA claim is falsified as a statement about generalization.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that logical anomalies—missing objects, misplacements, wrong color combinations—are better detected from what a vision-language model says about an image than from what a vision encoder's memory bank stores about it. A single normal image is described by the AVLM using Guided CoT prompts over the whole image and over GroundingDINO-selected regions; outlier text is filtered by LOF; the surviving text is formatted into JSON. The format-embedding branch then scores a query by cosine similarity to the normal description, and the logic-reasoner branch translates both descriptions into a formal theory Γ (normal specification plus unique-name, functionality, and domain-closure axioms) and uses Prover9 to test whether Γ entails the negation of the query description. When it does, the reasoner returns a minimal subset of the query description that contradicts normality, which is the explanation. Reported results: 86.0 AUROC / 83.7 F1-max one-shot on MVTec LOCO AD, better than AnomalyMoE by 18.1 / 4.6 and competitive or better than full-shot methods like PatchCore and GCAD on several categories.

Load-bearing premise

Everything rests on the assumption that the manually crafted Guided CoT prompts and hand-written normal logic rules generalize to new test images, an assumption the paper does not fully defend: its own ablation drops AUROC from 86.0 to 23.4 without Guided CoT, and no protocol is documented for designing prompts without access to test anomalies.

Editorial extensions

If this is right

  • A new inspection task needs only one normal image, reusable text prompts, and a per-category normal logic rule, replacing full-shot training and visual segmentation annotations without fine-tuning.
  • Every anomaly verdict is backed by a machine-checkable explanation: the minimal subset of the query description that contradicts the normal specification, so a human can audit why an image was flagged.
  • The method's ceiling is tied to the vision-language backbone: GPT-4o gives 86.0 AUROC, LLaVA 1.6 gives 76.2, and LLaVA 1.5 gives 73.3, so future AVLMs should push the same pipeline higher without redesign.
  • The continuous format-embedding score and the binary logic-reasoner verdict rarely disagree, so the two branches can serve as independent cross-checks in deployment.

Reading between the lines

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

  • Inference: if the Guided CoT prompts and normal specifications could be auto-generated from the single normal image, the method would become fully annotation-free; this is the natural next experiment and the paper leaves it open.
  • Inference: the sharp AUROC drop from 86.0 to 23.4 when Guided CoT is removed implies the practical bottleneck is prompt design, not visual feature quality, so future gains likely come from better auto-prompting or logic-specification learning.
  • Inference: because the normal specification is just a first-order logic theory, the same pipeline could be lifted to other relational image checks, such as document layout consistency or scene-graph verification, without changing the reasoner.
  • Inference: the binary reasoner output could be blended with the embedding score to yield calibrated confidence, a combination the paper does not test but the architecture invites.
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 / 5 minor

Summary. LogicAD proposes a training-free, one-shot pipeline for logical anomaly detection on MVTec LOCO AD. It extracts text descriptions of images with an autoregressive vision-language model (GPT-4o, LLaVA 1.5/1.6), uses GroundingDINO to produce ROIs, stabilizes repeated descriptions with text-embedding filtering via LOF, and computes an anomaly score from cosine similarity between JSON-formatted embeddings of a single normal image and the query image. A separate logic-reasoner module converts the description into first-order logic and uses Prover9 to detect contradictions with a hand-written normal specification, providing explanations. The paper reports an AUROC of 86.0% and F1-max of 83.7% on MVTec LOCO AD, claiming an 18.1% AUROC improvement over AnomalyMoE and strong results on MVTec AD, and argues that text features from AVLMs are a viable alternative to visual memory banks for logical anomaly detection.

Significance. If the empirical claims are valid, LogicAD would be a useful contribution: it is one-shot, training-free, produces human-readable explanations, and directly targets logical anomalies that visual memory-bank methods handle poorly. The pipeline is clearly presented, the use of an automated theorem prover for explanation is a distinctive and interesting design choice, and the paper includes ablations showing the contribution of each module. The main weakness is that the reported state-of-the-art result depends on manually crafted, per-category prompts and logical specifications whose independence from the test set is not established; Figure 4 explicitly shows a Guided CoT prompt chosen using a ground-truth anomaly description. The baseline comparison to AnomalyMoE is also weak because that method is cited only as a challenge webpage without per-category results or error bars. These issues are load-bearing for the headline SOTA and one-shot generalization claims, so the paper needs substantial revision before the results can be relied upon.

major comments (4)
  1. [Text Feature Extraction, Figure 4] Figure 4 and its caption state that the Guided CoT prompt was designed using the ground-truth description of a test anomaly: 'The ground truth description specifies that two cables are not connected to the same slot position.' Table 3 shows that removing Guided CoT collapses AUROC from 86.0 to 23.4, so the entire reported margin depends on these hand-written prompts. The paper must demonstrate a blinded prompt-development protocol in which all category-specific prompts and normal specifications are derived only from a single normal training image and publicly known category semantics, with the prompts frozen before any test anomalies are inspected. Without such a protocol, the 'one-shot' and 'SOTA' claims are not substantiated; the current evidence is consistent with test-informed prompt engineering.
  2. [Logic Reasoner and normal specification] The logic reasoner relies on a per-category normal specification Σnorm and default values that the paper states 'need to be provided for each class of AD task' and that 'may not necessarily be the definition of normality in the original dataset.' The example specification explicitly encodes the kind of violation that appears in the test set (both apple and nectarine on the left, tangerine count, etc.). Since the logic reasoner's predictions are determined by this specification, the paper must show that these rules were authored without access to the MVTec LOCO AD test anomalies and must discuss how a practitioner would obtain such rules from the single normal training image alone. Otherwise the logic-reasoner results measure the quality of the hand-written rules, not the generality of the proposed method.
  3. [Table 1 and baseline comparison] The claimed 18.1% AUROC improvement over the 'existing SOTA' is based entirely on AnomalyMoE, which is cited only as a CVPR 2024 VAND challenge webpage and is not accompanied by per-category results, standard deviations, or a peer-reviewed description. The headline comparison is therefore not verifiable. The authors should provide a per-category comparison with error bars for AnomalyMoE if the numbers are available, and ideally compare against an additional published logical-anomaly baseline (e.g., a full-shot method with standard deviations) so that the claimed improvement is not an artifact of an unreviewed reference.
  4. [Table 3 and logic-reasoner evaluation] The logic-reasoner row in the ablation reports only F1-max and has 'N/A' for AUROC because the reasoner emits binary predictions. The text claims it is only 0.4% lower than the format-embedding score, but no threshold-free metric (e.g., balanced accuracy, Matthews correlation coefficient) or detailed case-level comparison is provided for the reasoner, and the statement that disagreements are rare is deferred to Appendix A.5 without quantification in the main text. The explainability claim would be much stronger with a direct evaluation of the reasoner's decisions against the embedding-based scores, including an error analysis.
minor comments (5)
  1. [Abstract and Introduction] There are typos: 'anomalys' should be 'anomalies' in the Introduction, and 'with with' appears in the Dataset and Metrics section.
  2. [Table 5 vs Table 1] Table 5 reports GPT-4o F1-max as 83.2, while Table 1 reports LogicAD average F1-max as 83.7; the discrepancy should be reconciled or explained.
  3. [Ablation table] The ablation in Table 3 would be clearer if it also included a row with only format embedding (no Guided CoT, no ROI) and a row with only the logic reasoner, so that the individual contributions of the modules can be attributed more precisely.
  4. [References] The AnomalyMoE reference [Gu et al. 2024b] should be updated to a peer-reviewed paper or technical report with full experimental details if one exists; the current citation to a challenge webpage is not sufficient for a SOTA comparison.
  5. [Dataset and Metrics] The phrase 'consistent with with SOTA and competing methods' contains a duplicated word and should be corrected.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: LogicAD's anomaly scores and logic-reasoner labels are computed from an external normal reference; manual prompt design raises a leakage risk, not an in-construction reduction.

full rationale

LogicAD is an empirical pipeline rather than a formal derivation. The anomaly score is defined as a_core = 1 - <ê_n, ê_q> in the Format Embedding section, where ê_n comes from a separate normal image and ê_q from the query; this is a standard one-shot distance, not a quantity that is fitted to the reported labels and then re-predicted. The logic reasoner labels an image abnormal iff Γ |= ¬Σ0, with Γ = Σnorm ∪ Σna ∪ Σfa ∪ Σdca and Σ0 obtained by parsing the query description; this is deductive entailment from an explicitly supplied normal specification, so 'anomaly' is defined relative to that specification rather than being smuggled out of the test labels. The paper does not invoke a uniqueness theorem from the authors, and its citations (MVTec LOCO, WinCLIP, Prover9, CoT) are external and independent. The one serious caveat is empirical rather than circular: per-category Guided CoT prompts and Σnorm/default rules are hand-written, and Figure 4 explicitly illustrates prompt design using a ground-truth anomaly description ('The ground truth description specifies that two cables are not connected to the same slot position'), while Table 3 shows AUROC collapses from 86.0 to 23.4 without Guided CoT. The Limitations section also concedes that curated prompts can be reused and need to be defined only once per AD task. This is a potential test-set-influence / overfitting concern that should be evaluated under an external-validity audit (e.g., a blinded prompt-development protocol), not a circularity reduction: no equation in the paper equals its own input, and no fitted parameter is renamed as a prediction. Under the required standard of quotable by-construction reduction, no significant circularity is present.

Assumptions & free parameters 3 free parameters · 6 assumptions · 0 invented entities

The method relies on pretrained AVLMs, a text embedding model, GroundingDINO, and manually supplied normal logic specifications. No new physical entities are proposed. The main free choices are sampling and filtering hyperparameters plus per-category prompt definitions, which are not fully specified.

free parameters (3)
  • K = 3
    Number of repeated text generations per ROI before LOF filtering, chosen without a reported sensitivity analysis.
  • AVLM sampling parameters (topp, temperature) = Not reported
    The paper mentions tuning these hyperparameters but gives no values, which affects reproducibility.
  • LOF configuration = Not reported
    The outlier filter parameters for text embedding filtering are unspecified.
assumptions (6)
  • domain assumption Domain closure: every object mentioned in the image must appear in the normal specification's constant set, otherwise it is anomalous (Sigma_dca).
    Introduced in the Logic Reasoner section, based on Reiter 1980, assuming a closed world over a fixed set of object constants.
  • domain assumption Unique name assumption: distinct object names refer to distinct objects, decided by LLM query (Sigma_na).
    In the Logic Reasoner section, used to determine synonymy and distinguishability of constants.
  • domain assumption Functionality of predicates: each predicate such as left(x, y) or right(x, y) assigns a unique number to each object (Sigma_fa).
    In the Logic Reasoner section, either manually specified or decided via LLM query.
  • domain assumption Default values: unmentioned predicates are completed with count 0, enabling missing-object detection.
    In the Logic Reasoner section, a closed-world completion of the image description.
  • domain assumption Text embedding cosine similarity reflects logical anomaly severity.
    In the Format Embedding section, the anomaly score is defined as 1 minus the cosine similarity of JSON-format embeddings, assuming semantic distance equals anomaly likelihood.
  • domain assumption AVLMs and GroundingDINO produce accurate visual descriptions and ROIs.
    In the Text Feature Extraction section, the whole pipeline depends on these pretrained models without error modeling.

how reviews work

0 comments
Cite this review

Pith. "Pith review of LogicAD: Explainable Anomaly Detection via VLM-based Text Feature Extraction." pith.science (2026). https://pith.science/paper/VN6AX2V4

@misc{pith2026250101767,
  author       = {Pith},
  title        = {Pith review of: LogicAD: Explainable Anomaly Detection via VLM-based Text Feature Extraction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VN6AX2V4}},
  note         = {Machine review of arXiv:2501.01767}
}
read the original abstract

Logical image understanding involves interpreting and reasoning about the relationships and consistency within an image's visual content. This capability is essential in applications such as industrial inspection, where logical anomaly detection is critical for maintaining high-quality standards and minimizing costly recalls. Previous research in anomaly detection (AD) has relied on prior knowledge for designing algorithms, which often requires extensive manual annotations, significant computing power, and large amounts of data for training. Autoregressive, multimodal Vision Language Models (AVLMs) offer a promising alternative due to their exceptional performance in visual reasoning across various domains. Despite this, their application to logical AD remains unexplored. In this work, we investigate using AVLMs for logical AD and demonstrate that they are well-suited to the task. Combining AVLMs with format embedding and a logic reasoner, we achieve SOTA performance on public benchmarks, MVTec LOCO AD, with an AUROC of 86.0% and F1-max of 83.7%, along with explanations of anomalies. This significantly outperforms the existing SOTA method by a large margin.

Figures

Figures reproduced from arXiv: 2501.01767 by the authors.

Figure 1
Figure 1. Overview of Anomaly Detection Approaches: (A) AD models trained from scratch require large-scale datasets and are capable of AD but lack reasoning capabili￾ties. (B) Memory-based AD methods leverage a pre-trained vision model to extract features from normal images, en￾abling few-shot AD. However, they often require additional visual annotations and lack reasoning. (C) Our method uses pre-trained AVLMs as a text feat… view at source ↗
Figure 2
Figure 2. Pipeline overview of LogicAD. The green box represents text feature extraction, fi2t, which extracts features via pre-trained AVLMs from the input image, the detailed process is depicted in [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Text feature extraction fi2t involves ROI extrac￾tion (blue box) and text embedding filtering (green box). Patches and the original image are processed by an AVLM to generate K text descriptions. The green box uses text￾embedding-3-large (Achiam et al. 2023) for output stabiliza￾tion. icant limitations, yielding markedly poorer performance on tasks involving logical inconsistencies compared to their ef￾fectiveness i… view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Illustration of a standard prompt versus a Guided Chain-of-Thought (Guided CoT) prompt. We use image Xq from [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

47 extracted references · 31 canonical work pages

  1. [1]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.a...

  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. [3]

    L.; Almeida, D.; Altenschmidt, J.; Altman, S.; Anadkat, S.; et al

    Achiam, J.; Adler, S.; Agarwal, S.; Ahmad, L.; Akkaya, I.; Aleman, F. L.; Almeida, D.; Altenschmidt, J.; Altman, S.; Anadkat, S.; et al. 2023. GPT-4 Technical Report. arXiv preprint arXiv:2303.08774

    arXiv 2023

  4. [4]

    Bergmann, P.; Batzner, K.; Fauser, M.; Sattlegger, D.; and Steger, C. 2022. Beyond Dents and Scratches: Logical Constraints in Unsupervised Anomaly Detection and Localization. International Journal of Computer Vision, 130(4): 947--969

    work page 2022

  5. [5]

    Bergmann, P.; Fauser, M.; Sattlegger, D.; and Steger, C. 2019. MVTec AD — A Comprehensive Real-World Dataset for Unsupervised Anomaly Detection. In Conference on Computer Vision and Pattern Recognition, 9592--9600

    work page 2019

  6. [6]

    M.; Kriegel, H.-P.; Ng, R

    Breunig, M. M.; Kriegel, H.-P.; Ng, R. T.; and Sander, J. 2000. LOF: Identifying Density-based Local Outliers. In Proceedings of the 2000 ACM SIGMOD International Conference on Management of Data, 93--104

    work page 2000

  7. [7]

    Caron, M.; Touvron, H.; Misra, I.; J \'e gou, H.; Mairal, J.; Bojanowski, P.; and Joulin, A. 2021. Emerging Properties in Self-Supervised Vision Transformers. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 9650--9660

    work page 2021

  8. [8]

    Chen, X.; Han, Y.; and Zhang, J. 2023. APRIL-GAN: A Zero-/Few-Shot Anomaly Classification and Segmentation Method for CVPR 2023 VAND Workshop Challenge Tracks 1&2: 1st Place on Zero-shot AD and 4th Place on Few-shot AD. arXiv preprint arXiv:2305.17382

    arXiv 2023

Show all 47 references
  1. [9]

    Chen, Z.; Zhou, Q.; Shen, Y.; Hong, Y.; Sun, Z.; Gutfreund, D.; and Gan, C. 2024. Visual Chain-of-Thought Prompting for Knowledge-Based Visual Reasoning. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, 1254--1262

    2024

  2. [10]

    Enderton, H. B. 2001. A Mathematical Introduction to Logic

    2001

  3. [11]

    V.; Joulin, A.; and Misra, I

    Girdhar, R.; El-Nouby, A.; Liu, Z.; Singh, M.; Alwala, K. V.; Joulin, A.; and Misra, I. 2023. ImageBind: One Embedding Space To Bind Them All. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 15180--15190

    2023

  4. [12]

    Goyal, Y.; Khot, T.; Summers-Stay, D.; Batra, D.; and Parikh, D. 2017. Making the V in VQA Matter: Elevating the Role of Image Understanding in Visual Question Answering. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 6904--6913

    2017

  5. [13]

    Gu, Z.; Zhu, B.; Zhu, G.; Chen, Y.; Tang, M.; and Wang, J. 2024 a . AnomalyGPT: Detecting industrial anomalies using large vision-language models. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, 1932--1940

    2024

  6. [14]

    Gu, Z.; Zhu, B.; Zhu, G.; Chen, Y.; and Wang, J. 2024 b . CVPR, Visual Anomaly and Novelty Detection 2.0 Winner 2024, https://www.hackster.io/contests/openvino2024, Accessed: 2024-08-01

    2024

  7. [15]

    Gunjal, A.; Yin, J.; and Bas, E. 2024. Detecting and Preventing Hallucinations in Large Vision Language Models. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, 18135--18143

    2024

  8. [16]

    He, K.; Zhang, X.; Ren, S.; and Sun, J. 2016. Deep Residual Learning for Image Recognition. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 770--778

    2016

  9. [17]

    Jeong, J.; Zou, Y.; Kim, T.; Zhang, D.; Ravichandran, A.; and Dabeer, O. 2023. WinCLIP: Zero-/Few-Shot Anomaly Classification and Segmentation. In Conference on Computer Vision and Pattern Recognition, 19606--19616

    2023

  10. [18]

    M.; and Park, S

    Kim, S.; An, S.; Chikontwe, P.; Kang, M.; Adeli, E.; Pohl, K. M.; and Park, S. H. 2024. Few Shot Part Segmentation Reveals Compositional Logic for Industrial Anomaly Detection. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, 8591--8599

    2024

  11. [19]

    Lee, M. 2023. A Mathematical Investigation of Hallucination and Creativity in GPT Models. Mathematics, 11(10): 2320

    2023

  12. [20]

    Li, J.; Li, D.; Savarese, S.; and Hoi, S. 2023. BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models. In International Conference on Machine Learning, 19730--19742

    2023

  13. [21]

    Liu, H.; Li, C.; Li, Y.; and Lee, Y. J. 2024 a . Improved Baselines with Visual Instruction Tuning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 26296--26306

    2024

  14. [22]

    Liu, H.; Li, C.; Li, Y.; Li, B.; Zhang, Y.; Shen, S.; and Lee, Y. J. 2024 b . LLaVA-NeXT: Improved reasoning, OCR, and world knowledge

    2024

  15. [23]

    Liu, S.; Zeng, Z.; Ren, T.; Li, F.; Zhang, H.; Yang, J.; Li, C.; Yang, J.; Su, H.; Zhu, J.; et al. 2023 a . Grounding dino: Marrying dino with grounded pre-training for open-set object detection. arXiv preprint arXiv:2303.05499

    2023 arXiv

  16. [24]

    Liu, T.; Li, B.; Du, X.; Jiang, B.; Jin, X.; Jin, L.; and Zhao, Z. 2023 b . Component-aware anomaly detection framework for adjustable and logical industrial visual inspection. Advanced Engineering Informatics, 58: 102161

    2023

  17. [25]

    Lu, P.; Mishra, S.; Xia, T.; Qiu, L.; Chang, K.-W.; Zhu, S.-C.; Tafjord, O.; Clark, P.; and Kalyan, A. 2022. Learn to Explain: Multimodal Reasoning via Thought Chains for Science Question Answering. Advances in Neural Information Processing Systems, 35: 2507--2521

    2022

  18. [26]

    X.; Gu, A.; Lipkin, B.; Zhang, C

    Olausson, T. X.; Gu, A.; Lipkin, B.; Zhang, C. E.; Solar-Lezama, A.; Tenenbaum, J. B.; and Levy, R. P. 2023. LINC : A Neurosymbolic Approach for Logical Reasoning by Combining Language Models with First-Order Logic Provers. In The 2023 Conference on Empirical Methods in Natura...

    2023

  19. [27]

    Paiss, R.; Ephrat, A.; Tov, O.; Zada, S.; Mosseri, I.; Irani, M.; and Dekel, T. 2023. Teaching CLIP to Count to Ten. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 3170--3180

    2023

  20. [28]

    Pan, L.; Albalak, A.; Wang, X.; and Wang, W. 2023. Logic- LM : Empowering Large Language Models with Symbolic Solvers for Faithful Logical Reasoning. In Bouamor, H.; Pino, J.; and Bali, K., eds., Findings of the Association for Computational Linguistics: EMNLP 2023, 3806--3824...

    2023

  21. [29]

    W.; Hallacy, C.; Ramesh, A.; Goh, G.; Agarwal, S.; Sastry, G.; Askell, A.; Mishkin, P.; Clark, J.; et al

    Radford, A.; Kim, J. W.; Hallacy, C.; Ramesh, A.; Goh, G.; Agarwal, S.; Sastry, G.; Askell, A.; Mishkin, P.; Clark, J.; et al. 2021. Learning Transferable Visual Models From Natural Language Supervision. In International Conference on Machine Learning, 8748--8763

    2021

  22. [30]

    Reiter, R. 1980. Equality and domain closure in first-order databases. Journal of the ACM (JACM), 27(2): 235--249

    1980

  23. [31]

    Rippel, O.; Mertens, P.; K \"o nig, E.; and Merhof, D. 2021. Modeling the Distribution of Normal Data in Pre-Trained Deep Features for Anomaly Detection. IEEE Transactions on Instrumentation and Measurement, 70: 1--13

    2021

  24. [32]

    Rippel, O.; Mertens, P.; and Merhof, D. 2021. Modeling the Distribution of Normal Data in Pre-Trained Deep Features for Anomaly Detection. In 2020 25th International Conference on Pattern Recognition (ICPR), 6726--6733

    2021

  25. [33]

    Roth, K.; Pemula, L.; Zepeda, J.; Sch \"o lkopf, B.; Brox, T.; and Gehler, P. 2022. Towards Total Recall in Industrial Anomaly Detection. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 14318--14328

    2022

  26. [34]

    Rudolph, M.; Wehrbein, T.; Rosenhahn, B.; and Wandt, B. 2023. Asymmetric Student-Teacher Networks for Industrial Anomaly Detection. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, 2592--2602

    2023

  27. [35]

    J.; and Yan, Y

    Shang, Y.; Cai, M.; Xu, B.; Lee, Y. J.; and Yan, Y. 2024. Llava-prumerge: Adaptive token reduction for efficient large multimodal models. arXiv preprint arXiv:2403.15388

    2024

  28. [36]

    Song, Y.; Wang, G.; Li, S.; and Lin, B. Y. 2024. The Good, The Bad, and The Greedy: Evaluation of LLMs Should Not Ignore Non-Determinism. arXiv preprint arXiv:2407.10457

    2024 arXiv

  29. [37]

    Sun, Q.; Fang, Y.; Wu, L.; Wang, X.; and Cao, Y. 2023. EVA-CLIP: Improved Training Techniques for CLIP at Scale. arXiv preprint arXiv:2303.15389

    2023 arXiv

  30. [38]

    Sur \' s, D.; Menon, S.; and Vondrick, C. 2023. Vipergpt: Visual inference via python execution for reasoning. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 11888--11898

    2023

  31. [39]

    Touvron, H.; Lavril, T.; Izacard, G.; Martinet, X.; Lachaux, M.-A.; Lacroix, T.; Rozi \`e re, B.; Goyal, N.; Hambro, E.; Azhar, F.; et al. 2023. LLaMA: Open and Efficient Foundation Language Models. arXiv preprint arXiv:2302.13971

    2023 arXiv

  32. [40]

    Wang, H.; Ma, S.; Dong, L.; Huang, S.; Wang, H.; Ma, L.; Yang, F.; Wang, R.; Wu, Y.; and Wei, F. 2023. Bitnet: Scaling 1-bit transformers for large language models. arXiv preprint arXiv:2310.11453

    2023 arXiv

  33. [41]

    Wang, W.; Chen, Z.; Chen, X.; Wu, J.; Zhu, X.; Zeng, G.; Luo, P.; Lu, T.; Zhou, J.; Qiao, Y.; et al. 2024. Visionllm: Large language model is also an open-ended decoder for vision-centric tasks. Advances in Neural Information Processing Systems, 36

    2024

  34. [42]

    V.; Zhou, D.; et al

    Wei, J.; Wang, X.; Schuurmans, D.; Bosma, M.; Xia, F.; Chi, E.; Le, Q. V.; Zhou, D.; et al. 2022. Chain-of-Thought Prompting Elicits Reasoning in Large Language Models. Advances in Neural Information Processing Systems, 35: 24824--24837

    2022

  35. [43]

    Ye, X.; Chen, Q.; Dillig, I.; and Durrett, G. 2024. SatLM: Satisfiability-Aided Language Models Using Declarative Prompting. Advances in Neural Information Processing Systems, 36

    2024

  36. [44]

    Zhang, C.; and Wang, S. 2024. Good at Captioning, Bad at Counting: Benchmarking GPT-4v on Earth Observation data. arXiv preprint arXiv:2401.17600

    2024 arXiv

  37. [45]

    Zhang, Y.; Cao, Y.; Xu, X.; and Shen, W. 2024. LogiCode: an LLM-Driven Framework for Logical Anomaly Detection. arXiv preprint arXiv:2406.04687

    2024 arXiv

  38. [46]

    Zhou, Q.; Pang, G.; Tian, Y.; He, S.; and Chen, J. 2024. AnomalyCLIP: Object-agnostic Prompt Learning for Zero-shot Anomaly Detection. In The Twelfth International Conference on Learning Representations

    2024

  39. [47]

    Zou, Y.; Jeong, J.; Pemula, L.; Zhang, D.; and Dabeer, O. 2022. SPot-the-Difference Self-Supervised Pre-training for Anomaly Detection and Segmentation. In European Conference on Computer Vision, 392--408

    2022

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.