Papers › ViLT: Vision-and-Language Transformer Without Convolution or Region Supervision

ViLT: Vision-and-Language Transformer Without Convolution or Region Supervision

5 Feb 2021arXiv:2102.03334archive 2025-07-28

Wonjae Kim, Bokyung Son, Ildoo Kim

Vision-and-Language Pre-training (VLP) has improved performance on various joint vision-and-language downstream tasks. Current approaches to VLP heavily rely on image feature extraction processes, most of which involve region supervision (e.g., object detection) and the convolutional architecture (e.g., ResNet). Although disregarded in the literature, we find it problematic in terms of both (1) efficiency/speed, that simply extracting input features requires much more computation than the multimodal interaction steps; and (2) expressive power, as it is upper bounded to the expressive power of the visual embedder and its predefined visual vocabulary. In this paper, we present a minimal VLP model, Vision-and-Language Transformer (ViLT), monolithic in the sense that the processing of visual inputs is drastically simplified to just the same convolution-free manner that we process textual inputs. We show that ViLT is up to tens of times faster than previous VLP models, yet with competitive or better downstream task performance. Our code and pre-trained weights are available at https://github.com/dandelin/vilt.

PaperPDFCodeCode Syntology ran

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

For agents, Syntology's MCP tool lists every function and class Syntology harvested from this paper and whether it ran (how to connect): get_harvested_code_for_paper(arxiv_id="2102.03334")

Code

Syntology Ran 1 of 4 code samples harvested from 3 repositories linked to this paper; 3 have no recorded run. Of those that ran: 1 ran with no contract checked.

By repository: official repository: 1 sample from 1 repository, 0 ran; community (archive-listed): 3 samples from 2 repositories, 1 ran. The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.

dandelin/vilt officialmentioned in papermentioned on GitHubpytorch report
glamor-usc/climb mentioned on GitHubpytorch report
guilk/vlc mentioned on GitHubpytorch report
huggingface/transformers mentioned on GitHubpytorch report
wglab/gestaltmml mentioned on GitHubpytorch report
wglab/gestaltmml-gestaltgpt mentioned on GitHubpytorch report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

4 samples harvested; 1 ran; 0 honoured the contract we drafted; 3 have no recorded run. Read from Syntology's graph 2026-09-24; that is when this build read the record, not when the samples ran.

1ran
3unverified

Licence: 1 of the 4 samples is pointer only, meaning Syntology does not serve that copy's text. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from 3 repositories linked to this paper, official or community; each sample names its own and says which. “Ran” means the sample executed on a synthesized input. It does not mean the output is correct, and nothing here reproduces the paper's results. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself; “our draft was wrong” and “fixture could not drive it” are failures of Syntology's instrument, not of the code.

Each sample ends with its code_sha256, Syntology's identity for that exact code. An agent fetches the stored sample with Syntology's MCP tool get_code(code_sha256="…") (how to connect); click an identity to copy that call.

Repository labels, per sample. official repository: The archive marks this repository official for the paper. named in the paper: The archive records that the paper mentions this repository; it is not marked official. community (archive-listed): In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper. found in paper text by Syntology: Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted. community: Not in the archive's code links for this paper; a community repository Syntology harvested. Samples from a repository marked official are listed first. Licence labels name the repository's licence as recorded at harvest. “Pointer only” means Syntology does not serve that copy's text, for one of four reasons: no licence file was found; the licence was not identified; the licence is recorded as permissive but that copy's record is not marked cleared; or the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar). Some licences outside that list permit redistribution, such as WTFPL, and GPL-3.0 under its conditions; they are simply not on the list. Hover a licence label for the reason. File links open the file on GitHub at the default branch, which may have changed since the harvest.

ViLTransformerSS dandelin/vilt/vilt/modules/vilt_module.py official repository unverified Apache-2.0 (permissive) · dab2a98f28bf818f · report
EncoderWrapper glamor-usc/climb/src/modeling/vilt.py community (archive-listed) ran MIT (permissive) · f5e8d3006a5a1a6f · report
VLCTransformer guilk/vlc/vlc/modules/vlc_module.py community (archive-listed) unverified no licence file found · pointer only · 57128e00df804ac7 · report
ViltEncoderWrapper glamor-usc/climb/src/modeling/vilt.py community (archive-listed) unverified MIT (permissive) · 9e90629b7bccbd2a · report

Tasks

Cross-Modal RetrievalImage RetrievalMultimodal Intent RecognitionVisual Question Answering (VQA)Visual ReasoningZero-Shot Cross-Modal Retrievalmultimodal interactionobject-detection

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Cross-Modal Retrieval COCO 2014 ViLT-B/32 Image-to-text R@1 61.5 #27 of 36 Archive leaderboard report
Cross-Modal Retrieval COCO 2014 ViLT-B/32 Image-to-text R@10 92.7 #27 of 36 Archive leaderboard report
Cross-Modal Retrieval COCO 2014 ViLT-B/32 Image-to-text R@5 86.3 #27 of 36 Archive leaderboard report
Cross-Modal Retrieval COCO 2014 ViLT-B/32 Text-to-image R@1 42.7 #27 of 36 Archive leaderboard report
Cross-Modal Retrieval COCO 2014 ViLT-B/32 Text-to-image R@10 83.1 #27 of 36 Archive leaderboard report
Cross-Modal Retrieval COCO 2014 ViLT-B/32 Text-to-image R@5 72.9 #27 of 36 Archive leaderboard report
Cross-Modal Retrieval Flickr30k ViLT-B/32 Image-to-text R@1 83.5 #13 of 27 Archive leaderboard report
Cross-Modal Retrieval Flickr30k ViLT-B/32 Image-to-text R@10 98.6 #13 of 27 Archive leaderboard report
Cross-Modal Retrieval Flickr30k ViLT-B/32 Image-to-text R@5 96.7 #13 of 27 Archive leaderboard report
Cross-Modal Retrieval Flickr30k ViLT-B/32 Text-to-image R@1 64.4 #13 of 27 Archive leaderboard report
Cross-Modal Retrieval Flickr30k ViLT-B/32 Text-to-image R@10 93.8 #13 of 27 Archive leaderboard report
Cross-Modal Retrieval Flickr30k ViLT-B/32 Text-to-image R@5 88.7 #13 of 27 Archive leaderboard report
Image Retrieval PhotoChat ViLT R1 11.5 #3 of 5 Archive leaderboard report
Image Retrieval PhotoChat ViLT R@10 25.6 #3 of 5 Archive leaderboard report
Image Retrieval PhotoChat ViLT R@5 33.8 #3 of 5 Archive leaderboard report
Image Retrieval PhotoChat ViLT Sum(R@1,5,10) 71.0 #3 of 5 Archive leaderboard report
Multimodal Intent Recognition MMDialog ViLT F1 55.8 #4 of 4 Archive leaderboard report
Multimodal Intent Recognition PhotoChat ViLT F1 52.4 #5 of 6 Archive leaderboard report
Multimodal Intent Recognition PhotoChat ViLT Precision 55.4 #5 of 6 Archive leaderboard report
Multimodal Intent Recognition PhotoChat ViLT Recall 58.9 #5 of 6 Archive leaderboard report
Visual Question Answering (VQA) VQA v2 test-dev ViLT-B/32 Accuracy 71.26 #25 of 56 Archive leaderboard report
Visual Reasoning NLVR2 Dev ViLT-B/32 Accuracy 75.7 #13 of 15 Archive leaderboard report
Visual Reasoning NLVR2 Test ViLT-B/32 Accuracy 76.13 #14 of 14 Archive leaderboard report
Zero-Shot Cross-Modal Retrieval COCO 2014 ViLT-B/32 Image-to-text R@1 56.5 #16 of 18 Archive leaderboard report
Zero-Shot Cross-Modal Retrieval COCO 2014 ViLT-B/32 Image-to-text R@10 89.6 #16 of 18 Archive leaderboard report
Zero-Shot Cross-Modal Retrieval COCO 2014 ViLT-B/32 Image-to-text R@5 82.6 #16 of 18 Archive leaderboard report
Zero-Shot Cross-Modal Retrieval COCO 2014 ViLT-B/32 Text-to-image R@1 40.4 #16 of 18 Archive leaderboard report
Zero-Shot Cross-Modal Retrieval COCO 2014 ViLT-B/32 Text-to-image R@10 81.1 #16 of 18 Archive leaderboard report
Zero-Shot Cross-Modal Retrieval COCO 2014 ViLT-B/32 Text-to-image R@5 70 #16 of 18 Archive leaderboard report
Zero-Shot Cross-Modal Retrieval Flickr30k ViLT-B/32 Image-to-text R@1 73.2 #19 of 22 Archive leaderboard report
Zero-Shot Cross-Modal Retrieval Flickr30k ViLT-B/32 Image-to-text R@10 96.5 #19 of 22 Archive leaderboard report
Zero-Shot Cross-Modal Retrieval Flickr30k ViLT-B/32 Image-to-text R@5 93.6 #19 of 22 Archive leaderboard report
Zero-Shot Cross-Modal Retrieval Flickr30k ViLT-B/32 Text-to-image R@1 55 #19 of 22 Archive leaderboard report
Zero-Shot Cross-Modal Retrieval Flickr30k ViLT-B/32 Text-to-image R@10 89.8 #19 of 22 Archive leaderboard report
Zero-Shot Cross-Modal Retrieval Flickr30k ViLT-B/32 Text-to-image R@5 82.5 #19 of 22 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

Introduced by this paper: ViLT

Absolute Position EncodingsAttentionDense ConnectionsDropoutLayer NormalizationLinear LayerMulti-Head AttentionPosition-Wise Feed-Forward LayerResidual ConnectionSoftmaxTransformerViLT

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections