Color in Visual-Language Models: CLIP deficiencies

Guillem Arias; Maria Vanrell; Ramon Baldrich

REVIEW 5 major objections 6 minor 1 cited by

Color in Visual-Language Models: CLIP deficiencies

T0 review · 5 major / 6 minor · reviewed 2026-08-08 · deepseek-v4-flash

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

Pith's one-line read CLIP understands chromatic colors but treats white, gray, and black as non-colors and prefers reading color words over perceiving font colors, a deficiency traceable to text-selective neurons.

desk verdict The paper's behavioral evidence for CLIP's achromatic bias and text-priority is solid and worth engaging with, while the neuron-level mechanistic claims are more speculative and need sharper quantification. read the letter →

arxiv 2502.04470 v1 pith:XQA3BWIM submitted 2025-02-06 cs.CV cs.AI

Guillem Arias , Ramon Baldrich , Maria Vanrell This is my paper · ORCID

classification cs.CVcs.AI

keywords CLIPcolornamingachromaticcolorsStroopeffectmulti-modalneuronsneuronselectivityvisual-languagemodels

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

This paper asks whether CLIP, the contrastive visual-language model used widely for zero-shot tasks, genuinely understands color as a visual attribute. Using synthetic images of colored shapes and a Stroop-style dataset of color words rendered in colored fonts, the authors show that CLIP names chromatic colors accurately but systematically refuses achromatic labels: white, gray, and black are almost never assigned when a chromatic color is present. They also demonstrate a strong reading bias: asked for the font color, CLIP outputs the written color name in 81.1% of white-background cases and only 2.35% of cases when a colored background is added. At the neuron level, the paper attributes these failures to a high number of text-selective neurons in deep layers and a smaller set of color multi-modal neurons in shallow layers. If the paper is right, any downstream system built on CLIP embeddings carries a built-in color deficiency: it cannot treat achromatic colors as colors, and it reads color words instead of seeing colors.

What carries the argument

The evaluation machinery is a zero-shot color-naming task: for each image, a set of color-label text prompts is ranked by cosine similarity between the CLIP text and image embeddings, and the highest-scoring label is the prediction. This turns every experiment into a direct comparison of how the text encoder and image encoder align. For the internal analysis, the paper introduces a Color-Label Selectivity Index, $f_c(n_{i,L}) = \frac{\sum_{j=1}^{N_c} w_{j,i,L}}{\sum_{l=1}^{N} w_{l,i,L}}$, which measures how often a specific color label appears among the top-activating images for a neuron. This index is used to classify neurons into five types—Color, Any Word, Color Word, Color Multimodal, and Not activated—and to map their distribution across layers.

What would settle it

Use the same colored-text images but present the answer options as color swatches (patches of color) instead of color-name words, and measure accuracy on achromatic font colors and font-color identification. If CLIP still fails on achromatic colors and still returns the written color name, the deficiency is visual; if accuracy rises sharply, the reported failures are an artifact of prompting with color names.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that CLIP's color ability is split: it correctly links chromatic color labels to visual stimuli, but it does not treat achromatic stimuli as belonging to the color concept. In the shape experiments, when the asked object is achromatic and the background chromatic, correct object-color prediction falls to 0.19%; when the background is achromatic and the object chromatic, correct background-color prediction falls to 5%. In the Stroop experiments, CLIP's font-color accuracy is 16.7% on a white background and 2.35% on a colored background, while the written color name is chosen 81.1% and 59.5% of the time respectively. The neuron analysis then shows why: deep layers are dominated by text-selective neurons, and the color multi-modal neurons that do exist are few and appear in shallow layers.

Load-bearing premise

The results assume that picking the top-scoring color-label prompt by cosine similarity faithfully measures what CLIP perceives as the image's color, so a bias in how the prompt text is phrased could masquerade as a visual deficiency.

Editorial extensions

If this is right

CLIP-based zero-shot color classifiers will systematically mislabel white, gray, and black whenever a chromatic color is in view, regardless of which image region the prompt asks about.
Any image that contains a color word will be labeled by reading rather than by perceived color, so downstream tasks such as font-color extraction or color-aware captioning will fail on such images.
The reading bias grows with visual distractions: adding a colored background drops font-color accuracy from 16.7% to 2.35%.
Since text-selective neurons concentrate in deep layers while color multi-modal neurons are few and shallow, the deficiency is structural rather than a simple threshold artifact, and will not be cured by prompt rewording alone.

Reading between the lines

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

This suggests a testable extension: if the same Stroop images are evaluated with color swatches instead of color-name prompts as answer options, and achromatic accuracy stays near zero, the failure is in the visual encoder; if accuracy jumps, the reported metric was partly a text artifact.
The shallow-layer color multi-modal neurons suggest that low-level attribute concepts are encoded earlier in CLIP than previously assumed; probing these neurons might predict a model's color robustness before full evaluation.
Models finetuned from CLIP for image generation or editing may inherit the achromatic deficiency, producing black, gray, or white objects only with difficulty; this could be checked by prompting such models for achromatic objects in colorful scenes.
The authors' progressive-training hypothesis could be tested directly by re-training a small CLIP variant on an ordered curriculum and measuring achromatic Stroop accuracy before and after.

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.

Desk Editor's Note

read the letter

The main news here is that CLIP really does have a reproducible achromatic blind spot: when an achromatic object or background sits next to a chromatic one, CLIP almost always labels the chromatic stimulus, and the text-over-vision Stroop effect is severe, especially with colored backgrounds. Those findings are supported by large synthetic datasets, multiple prompt templates, and consistent percentages across conditions. The neuron analysis is also a genuine addition—particularly the observation of color multimodal neurons in shallow layers, which extends the usual deep-layer story from prior work. The new Color-Label Selectivity Index is a reasonable tool, and the hue-selectivity correlation with ImageNet (R=0.965) is a nice sanity check.

That said, the paper has real soft spots. There are no error bars or significance tests despite the word “significant” being used; the thresholds for neuron categories are unquantified; and no code or data are released, which makes the claims harder to verify. The stress-test worry about the prompt-ranking operationalization is worth taking seriously, but it is not fatal: the same text prompts identify chromatic objects correctly, so a pure text-encoder prior against achromatic labels would not explain the asymmetry. Still, a visual-side control (e.g., a linear probe on image embeddings) would make the central claim much tighter. The neuron taxonomy is defined on the same Stroop data used to report the taxonomy, which is a mild circularity, though the behavioral results stand independently. The proposed explanation—that achromatic colors are prevalent as backgrounds in training data—is a plausible hypothesis but not tested.

Who should read this: anyone working on CLIP interpretability, color understanding in multimodal models, or evaluating CLIP for downstream tasks where color matters. It is not a methodological breakthrough, but it documents a real deficiency that practitioners should know about. I would send it to peer review with a request for revision: add error bars and data/code, quantify the neuron thresholds, and address the operationalization concern with at least one probe experiment. The core empirical findings deserve to be in the literature, and the neuron-level analysis is a useful starting point for follow-up work.

Referee Report

5 major / 6 minor

Summary. This paper presents a set of experiments on synthetic images and Stroop-style stimuli to evaluate CLIP's ability to label colors. The authors report that CLIP correctly labels chromatic colors but fails to assign achromatic labels (white, gray, black) when a chromatic color is present in the image, and that CLIP strongly prefers the written color word over the actual font color in Stroop tasks. To explain these behaviors, the authors introduce a Color-Label Selectivity Index and analyze neuron activations across CLIP's visual encoder, proposing a taxonomy of neuron types and reporting the presence of color multimodal neurons in shallow layers. The main contributions are empirical: a systematic behavioral characterization of CLIP's color biases and a descriptive neuron-level analysis.

Significance. If confirmed, the behavioral findings are valuable for the VLM community, as they highlight systematic color deficiencies in a widely used model. The use of large synthetic datasets with multiple prompt templates is a methodological strength, and the Stroop experiments provide a compelling demonstration of text-over-vision bias. The neuron analysis is exploratory and less central; its contribution would be strengthened by more rigorous definitions and quantitative validation. The paper does not provide code or model specifications, which limits its immediate reproducibility.

major comments (5)

[Section 2, Experiments 1-2 (Table 2)] The central claim that "CLIP does not attribute the color word to achromatic parts of the image in presence of a Chromatic color" is inferred exclusively from the ranking of 11 color-name text prompts by cosine similarity to the full image embedding. This operationalization conflates the image encoder's representation, the text encoder's embedding of color words, and the contrastive alignment. The paper varies the prompt template (Experiment 3) but never varies the output channel; all answers are text-label rankings. A visual-side control—for example, a linear probe trained on the image encoder's features to predict the actual object/background color, or a forced-choice task with achromatic labels in a matched context—is needed to determine whether the achromatic failure reflects a visual encoding deficiency or a text-encoder prior that achromatic words are less plausible color labels. Without such a control, the causal interpretation of the achromatic deficiency (and the hypothesis in the Conclusions about achromatic backgrounds in datasets) is underdetermined.
[Section 4, "Activation Analysis" and Figure 5] The neuron taxonomy is defined using unquantified thresholds. The criteria "high Color-Label Selectivity Index", "high activation for any word", "does not reach 50% of the maximum activation", and the conjunction required for "Color Multimodal" are not specified numerically (e.g., percentile cutoffs, minimum activation ratios, or how the three color-label conditions are combined). The reported distribution of neuron types and the finding of shallow-layer color multimodal neurons cannot be reproduced or evaluated without these thresholds.
[Throughout (all experiments)] The paper does not identify the exact CLIP model variant or implementation used for any experiment. The Introduction says the image encoder is "Resnet-like" and the neuron analysis refers to "blocks of convolutional layers", but the model name (e.g., RN50, RN101, ViT-B/32), the pretraining dataset, and the checkpoint or codebase are not given. All experimental results, and especially the neuron-level analysis, depend on this specificity.
[Abstract and Section 3 (Experiment 5)] The word "significant" is used without supporting statistics. The paper reports aggregate percentages over 500 samples per condition but provides no error bars, confidence intervals, or significance tests. This is particularly problematic for the claim that achromatic backgrounds have "significant lower errors" in the Stroop test, where the effect appears to be smaller than the main text-over-vision bias. At minimum, the authors should report per-condition variability and avoid "significant" without a test.
[Section 4, Equation (1)] The definition of the Color-Label Selectivity Index is unclear. The notation uses N_c as a count of images but then sums activation values w_{j,i,L} over j, and the denominator sums over N_l w_{l,i,L}. The text should clarify whether f_c is a count ratio or an activation-weighted frequency, and should specify the set of images ("N cropped top scoring images") over which the index is computed (e.g., which dataset, how crops are selected, how many). This is necessary to interpret the subsequent neuron classification.

minor comments (6)

[Section 2, Experiment 2] The text says "the performance decreases to 0.19%" when the object is achromatic, but Table 2 shows 0.15% for the corresponding cell; please align the text with the table.
[Section 3, Experiment 3] The text says "2.23%" for the font-color percentage of the selected prompt, but Table 3 shows 2.35%; please correct.
[Abstract] The abstract uses "VML" while the rest of the paper uses "VLM"; please standardize.
[Introduction] Reference [5] contains "Akash Ghosh1 et al." with a stray superscript; please clean up the citation.
[General] Please provide a reproducibility statement or release the code and data; the paper's value would be greatly increased by including the synthetic dataset generation code and the neuron analysis scripts.
[Section 4, neuron categories] In the description of the neuron categories, the paper should explicitly define what "high activation" means in terms of the activation distribution (e.g., relative to the top-scoring ImageNet patches).

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the behavioral results are external evaluations on synthetic data, the neuron taxonomy is descriptive, and self-citations are methodological.

full rationale

The paper's headline deficiencies (achromatic colors rarely selected as labels; text over font color in Stroop tests) are direct measurements on synthetic datasets in Sections 2 and 3. No parameter is fitted to the reported outcome and no equation forces the result by definition: the cosine-similarity ranking over 11 color prompts is an operationalization whose validity can be debated, but it is not a circular reduction. The neuron-level analysis in Section 4 is a descriptive taxonomy: the Color-Label Selectivity Index (Eq. 1) measures activation frequencies on the Stroop dataset and the neuron categories are defined by that index, so the taxonomy is not an independent prediction of the behavioral findings, nor is it used as the derivation of those findings. Self-citations [11,12] provide the base selectivity-index methodology and a comparison distribution; they are not load-bearing verifications of the central claims. Since the central behavioral results are self-contained against external benchmark-like synthetic datasets and the internal neuron analysis is explicitly descriptive, no exhibited circular step meets the quoted-reduction standard.

Assumptions & free parameters 4 free parameters · 4 assumptions · 1 invented entities

The central claims rest on hand-chosen color sets, prompt ranking as ground truth, and a neuron taxonomy whose thresholds are not quantified. The neuron-related explanation is the least externally anchored part of the paper.

free parameters (4)

Fixed RGB values for the 11 basic color labels = not stated in the paper
All synthetic images and Stroop stimuli are generated from these hand-chosen RGB values; achromatic results for white, gray, and black depend on their exact settings.
Top-100 patch count for the color selectivity index = 100
The number of top-scoring patches used to compute each neuron's color selectivity is inherited from Rafegas et al. [11] and is not varied or justified for CLIP.
Activation cutoff for the 'Not activated' neuron category = 50% of ImageNet maximum
Neurons whose maximum Stroop activation falls below 50% of their ImageNet activation are labeled 'Not activated'; this threshold is introduced in Section 4 without sensitivity analysis.
Color-Label selectivity 'high' threshold = not specified
Classification of neurons into Color, Any Word, Color Word, and Color Multimodal depends on a 'high' index that is described qualitatively and never quantified.

assumptions (4)

domain assumption Cosine similarity between CLIP image and text embeddings is a valid proxy for color understanding
Every experiment treats the label with the highest cosine similarity as CLIP's answer; this is standard zero-shot evaluation but is not validated for color semantics.
domain assumption Flat synthetic renders with homogeneous backgrounds and basic shapes represent color perception tasks
The synthetic datasets in Sections 2-3 remove texture, lighting, and context; conclusions about CLIP's general color deficiencies assume this simplification preserves the relevant visual attribute.
domain assumption The Berlin and Kay 11 basic color terms are the correct closed label set
The evaluation restricts answers to these 11 terms, and achromatic results are defined entirely by the presence of white, gray, and black.
domain assumption Top-scoring image patches reveal a neuron's semantic selectivity
The Color-Label Selectivity Index follows Rafegas et al. [12] in assuming that the images that most activate a neuron expose its preferred concept; this is the standard method but it is an interpretability assumption.

invented entities (1)

Neuron taxonomy of color-label selective types
purpose: Proposed explanatory categories for CLIP units that respond to colors, words, or both, including the Color Multimodal neuron type offered as the internal mechanism behind color concept binding.
The taxonomy is defined in Section 4 using the same Stroop dataset on which it is reported; no ablation, causal intervention, or out-of-sample replication is provided to show these neuron types are causally involved.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Color in Visual-Language Models: CLIP deficiencies." pith.science (2026). https://pith.science/paper/XQA3BWIM

@misc{pith2026250204470,
  author       = {Pith},
  title        = {Pith review of: Color in Visual-Language Models: CLIP deficiencies},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XQA3BWIM}},
  note         = {Machine review of arXiv:2502.04470}
}

read the original abstract

This work explores how color is encoded in CLIP (Contrastive Language-Image Pre-training) which is currently the most influential VML (Visual Language model) in Artificial Intelligence. After performing different experiments on synthetic datasets created for this task, we conclude that CLIP is able to attribute correct color labels to colored visual stimulus, but, we come across two main deficiencies: (a) a clear bias on achromatic stimuli that are poorly related to the color concept, thus white, gray and black are rarely assigned as color labels; and (b) the tendency to prioritize text over other visual information. Here we prove it is highly significant in color labelling through an exhaustive Stroop-effect test. With the aim to find the causes of these color deficiencies, we analyse the internal representation at the neuron level. We conclude that CLIP presents an important amount of neurons selective to text, specially in deepest layers of the network, and a smaller amount of multi-modal color neurons which could be the key of understanding the concept of color properly. Our investigation underscores the necessity of refining color representation mechanisms in neural networks to foster a more comprehensive comprehension of colors as humans understand them, thereby advancing the efficacy and versatility of multimodal models like CLIP in real-world scenarios.

Figures

Figures reproduced from arXiv: 2502.04470 by the authors.

**Figure 1.** CLIP architecture set for a color naming task. The input image in the visual encoder is contrasted with several color labels within the input text. The output is the label that maximizes the visual and text embedding (Example: Input Text is ”The background is { color }” and Input Image is a green triangle with a pink background). an initial pre-training of the visual encoder with ImageNet dataset [4], followed by a … view at source ↗

**Figure 2.** Distribution of Color Selective Neurons in CLIP Visual Encoder Layers. 4. Color in CLIP Visual Encoder Color Selectivity Once we concluded several deficiencies in color label assignment, in this section we explore possible causes for these drawbacks. Our exploration is done at the neuron level. Firstly, we analyse the generic Color Selectivity Index of individual neuron units as it was definded in [11], where the … view at source ↗

**Figure 4.** Example of 4 types of Neurons. Left side: Neuron feature (weighted averaged of the first 100 top-scoring cropped images). Right side: 9 top-scoring cropped images from the Stroop dataset [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗

Figures from the paper (1 more)

**Figure 5.** Figure 5: Distribution of Neuron Types per layers in CLIP. Stroop dataset does not reach 50% of the maximum activation that this neuron achieves with the ImageNet dataset. In figure 5 we show the distribution of the different type of neurons, according to the previous descript…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

Language-based Color ISP Tuning
eess.IV 2025-09 conditional novelty 6.0 of 10

Language-described color styles can be applied to photos by optimizing a small camera color matrix with gradient descent against a vision-language model's similarity score.

Reference graph

Works this paper leans on

15 extracted references · 12 canonical work pages · cited by 1 Pith paper

[1]

Exploring the categorical nature of colour perception: Insights from artificial networks

Arash Akbarinia. Exploring the categorical nature of colour perception: Insights from artificial networks. bioRxiv, 2024. Preprint available on bioRxiv

work page 2024
[2]

Basic Color Terms: Their Uni- versality and Evolution

Brent Berlin and Paul Kay. Basic Color Terms: Their Uni- versality and Evolution . University of California Press, Berkeley and Los Angeles, 1969

work page 1969
[3]

A simple framework for contrastive learning of visual representations

Ting Chen, Simon Kornblith, Mohammad Norouzi, and Ge- offrey Hinton. A simple framework for contrastive learning of visual representations. In International Conference on Machine Learning, pages 1597–1607, 2020

work page 2020
[4]

J. Deng, W. Dong, R. Socher, L.-J. Li, K. Li, and L. Fei-Fei. ImageNet: A Large-Scale Hierarchical Image Database. In Proc. CVPR, 2009

work page 2009
[5]

Exploring the frontier of vision- language models: A survey of current methodologies and future directions

Akash Ghosh1 et al. Exploring the frontier of vision- language models: A survey of current methodologies and future directions. arXiv preprint arXiv:2102.05918, 2021

arXiv 2021
[6]

Multimodal neurons in artificial neural net- works

Gabriel Goh, Nick Cammarata †, Chelsea V oss †, Shan Carter, Michael Petrov, Ludwig Schubert, Alec Radford, and Chris Olah. Multimodal neurons in artificial neural net- works. Distill, 2021. https://distill.pub/2021/multimodal- neurons

work page 2021
[7]

Deep residual learning for image recognition

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

2016
[8]

Multiviz: Towards visualizing and understanding multimodal models

Paul Pu Liang, Yiwei Lyu, Gunjan Chhablani, Nihal Jain, Zihao Deng, Xingbo Wang, Louis-Philippe Morency, and Ruslan Salakhutdinov. Multiviz: Towards visualizing and understanding multimodal models. In International Confer- ence on Learning Representations (ICLR), 2023

work page 2023

Show all 15 references

[9]

Invariant visual repre- sentation by single neurons in the human brain

Rodrigo Quian Quiroga, Leila Reddy, Gabriel Kreiman, Christof Koch, and Itzhak Fried. Invariant visual repre- sentation by single neurons in the human brain. Nature, 435(7045):1102–1107, 2005

2005
[10]

Learning transferable vi- sual models from natural language supervision

Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, and Ilya Sutskever. Learning transferable vi- sual models from natural language supervision. CoRR, abs/2103.00020, 2021

2021 arXiv
[11]

Color encoding in biologically-inspired convolutional neural networks

Ivet Rafegas and Maria Vanrell. Color encoding in biologically-inspired convolutional neural networks. Vision Research, 151:7–17, 2018. Color: cone opponency and be- yond

2018
[12]

Understanding trained cnns by indexing neuron se- lectivity

Ivet Rafegas, Maria Vanrell, Lu´ıs A Alexandre, and Guillem Arias. Understanding trained cnns by indexing neuron se- lectivity. Pattern Recognition Letters, 136:318–325, 2020

2020
[13]

Are vision-language transformers learn- ing multimodal representations? a probing perspective

Emmanuelle Salin, Badreddine Farah, St ´ephane Ayache, and Benoit Favre. Are vision-language transformers learn- ing multimodal representations? a probing perspective. In Proceedings of the AAAI Conference on Artificial Intelli- gence, pages 11248–11257, 2022

2022
[14]

Studies of interference in serial ver- bal reactions

John Ridley Stroop. Studies of interference in serial ver- bal reactions. Journal of Experimental Psychology, 18:643– 662, 1935

1935
[15]

Attention is all you need

Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszko- reit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. Attention is all you need. InAdvances in neural information processing systems, pages 5998–6008, 2017

2017

Pith tools

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

[1] [1]

Exploring the categorical nature of colour perception: Insights from artificial networks

Arash Akbarinia. Exploring the categorical nature of colour perception: Insights from artificial networks. bioRxiv, 2024. Preprint available on bioRxiv

work page 2024

[2] [2]

Basic Color Terms: Their Uni- versality and Evolution

Brent Berlin and Paul Kay. Basic Color Terms: Their Uni- versality and Evolution . University of California Press, Berkeley and Los Angeles, 1969

work page 1969

[3] [3]

A simple framework for contrastive learning of visual representations

Ting Chen, Simon Kornblith, Mohammad Norouzi, and Ge- offrey Hinton. A simple framework for contrastive learning of visual representations. In International Conference on Machine Learning, pages 1597–1607, 2020

work page 2020

[4] [4]

J. Deng, W. Dong, R. Socher, L.-J. Li, K. Li, and L. Fei-Fei. ImageNet: A Large-Scale Hierarchical Image Database. In Proc. CVPR, 2009

work page 2009

[5] [5]

Exploring the frontier of vision- language models: A survey of current methodologies and future directions

Akash Ghosh1 et al. Exploring the frontier of vision- language models: A survey of current methodologies and future directions. arXiv preprint arXiv:2102.05918, 2021

arXiv 2021

[6] [6]

Multimodal neurons in artificial neural net- works

Gabriel Goh, Nick Cammarata †, Chelsea V oss †, Shan Carter, Michael Petrov, Ludwig Schubert, Alec Radford, and Chris Olah. Multimodal neurons in artificial neural net- works. Distill, 2021. https://distill.pub/2021/multimodal- neurons

work page 2021

[7] [7]

Deep residual learning for image recognition

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

2016

[8] [8]

Multiviz: Towards visualizing and understanding multimodal models

Paul Pu Liang, Yiwei Lyu, Gunjan Chhablani, Nihal Jain, Zihao Deng, Xingbo Wang, Louis-Philippe Morency, and Ruslan Salakhutdinov. Multiviz: Towards visualizing and understanding multimodal models. In International Confer- ence on Learning Representations (ICLR), 2023

work page 2023