Papers › Visual Prompt Tuning

Visual Prompt Tuning

23 Mar 2022arXiv:2203.12119archive 2025-07-28

Menglin Jia, Luming Tang, Bor-Chun Chen, Claire Cardie, Serge Belongie, Bharath Hariharan, Ser-Nam Lim

The current modus operandi in adapting pre-trained models involves updating all the backbone parameters, ie, full fine-tuning. This paper introduces Visual Prompt Tuning (VPT) as an efficient and effective alternative to full fine-tuning for large-scale Transformer models in vision. Taking inspiration from recent advances in efficiently tuning large language models, VPT introduces only a small amount (less than 1% of model parameters) of trainable parameters in the input space while keeping the model backbone frozen. Via extensive experiments on a wide variety of downstream recognition tasks, we show that VPT achieves significant performance gains compared to other parameter efficient tuning protocols. Most importantly, VPT even outperforms full fine-tuning in many cases across model capacities and training data scales, while reducing per-task storage cost.

PaperPDFCodeCode Syntology ranMethod as code

In Syntology View this paper on Syntology, its page in Syntology's graph. That page adds which of its functions are also present in other papers' code Syntology holds, with dates, and its dated run record as JSON and a README badge. It also lists the repositories linked to the paper, the abstract and the calls for agents.

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

For agents, on Syntology's MCP service (how to connect):

Code

The archive's official repository (KMnP/vpt) was not harvested; the 26 samples below come from 5 community repositories, and 1 more is identical to code Syntology harvested from another paper's repository.

Syntology 27 code samples: 26 from 5 repositories linked to this paper, 1 identical to code harvested from another paper's repository. Across all 27: 18 ran (of which 12 constructed an object rather than computing a result; 16 with no instrument failure: 1 honoured, 1 violated, 14 with no contract checked; 2 where Syntology's instrument failed) · 9 unverified. Where Syntology's instrument failed: 1 where our contract draft was wrong; 1 where the contract check could not run; failures of Syntology's instrument, not of the code.

By repository: community (archive-listed): 26 samples from 5 repositories, 17 ran (15 with no instrument failure; 2 where Syntology's instrument failed); 1 identical to code first harvested elsewhere. The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.

Levels written between 2026-09-22 and 2026-09-24 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

KMnP/vpt officialmentioned in papermentioned on GitHubpytorchNOASSERTIONSyntology: not harvested report
TooTouch/VPT mentioned on GitHubpytorchSyntology: 1 sample, 1 ran (0 instrument-failed) · 0 unverified report
Yiming-M/CLIP-EBC mentioned on GitHubpytorchSyntology: 10 samples, 6 ran (0 instrument-failed) · 4 unverified report
heekhero/DTL mentioned on GitHubpytorchSyntology: 11 samples, 7 ran (2 instrument-failed) · 4 unverified report
wgcban/apt mentioned on GitHubpytorchSyntology: 3 samples, 2 ran (0 instrument-failed) · 1 unverified report
unites-lab/vpns pytorchSyntology: 1 sample, 1 ran (0 instrument-failed) · 0 unverified 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

27 samples harvested: 18 ran (of which 12 constructed an object rather than computing a result; 16 with no instrument failure: 1 honoured, 1 violated, 14 with no contract checked; 2 where Syntology's instrument failed) · 9 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Syntology's record for this page has not changed since , the first build that kept a record date for it; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

1ran · honoured contract
1ran · violated contract
1ran · our draft was wrong
1ran; the contract check could not run
14ran
9unverified

Licence: 15 of the 27 samples are 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 5 repositories linked to this paper, official or community; the list below is grouped by repository, each group headed by its name and what it is to the paper. 1 sample is listed last, in a group without a repository. “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. A line beginning “fixture:” under a sample that ran gives the arguments Syntology's fixture called the sample with, each as name:type/rank/element kind (? where none was recorded), and the type of what the fixture's run returned, exactly as the harness recorded them; the arguments and the returned type were recorded by two separate runs of the same fixture. No output value is shown, and a returned type says nothing about whether the output was right. “Level recorded” gives the UTC date when Syntology's graph wrote the sample's verification level, not when the code executed; where the level was written before Syntology's graph kept that date, the row says “level recorded, date not recorded”. For 7 samples that ran, the graph records a signature this page does not show: for a class the fixture built without calling, the harness called the object with inputs it made up itself, so the recorded signature is the harness's probe, not the fixture's arguments. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself. “Our draft was wrong” and “the contract check could not run” are failures of Syntology's instrument, not of the code: the sample ran on Syntology's synthesized input, and then the drafted contract did not describe it, or the check itself raised an error (RAISES), lacked a dependency (DEP_MISSING) or ran out of time (TIMEOUT); hover the label for which. A metamorphic tier, where shown, is evidence of the output's shape, not that the sample implements the method.

Each sample ends with its code_sha256. A sample's code_sha256 is Syntology's identity for that exact code. It is printed in full (16 hexadecimal characters for every sample in this build), and get_code(code_sha256="…") takes it exactly as printed. An agent calls it on Syntology's MCP service (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 are grouped by repository: a repository marked official first, then by how directly the repository is tied to the paper, then in the archive's order; never by how many ran. 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 six 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; the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar); this paper's copy was not located, so its licence is not recorded; or this paper's copy is cleared, but the stored copy of the same code that Syntology serves is not. 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.

A function or class name links to its page when the name occurs in the harvested code of at least 5 papers; the name is a string, not an identity. A census tag appears when Syntology's census ran implementations with that name on one shared input per bucket: it gives how many were compared, from how many papers, and how many distinct outputs they produced, and names this sample's bucket when this sample was one of them.

TooTouch/VPT community (archive-listed) 1 sample, 1 ran (0 instrument-failed) · 0 unverified
VPT models/vpt.py ran MIT (permissive) · 3c7693b669414501 · report fixture: signature x:Tensor/4/float · returned Tensor · level recorded 2026-09-24
Yiming-M/CLIP-EBC community (archive-listed) 10 samples, 6 ran (0 instrument-failed) · 4 unverified
ConvNormActivation in 5 papers models/encoder/vit.py ran · metamorphic tier: deterministic fingerprinted MIT (permissive) · 3208de787e3ca751 · report fixture: returned ConvNormActivation · level recorded 2026-09-22
ConvStemConfig models/encoder/vit.py ran · metamorphic tier: deterministic MIT (permissive) · a1aa4bfcabd828f1 · report fixture: returned ConvStemConfig · level recorded 2026-09-22
EncoderBlock in 23 papers models/encoder/vit.py ran · metamorphic tier: deterministic MIT (permissive) · c6bf5366cd01303c · report fixture: returned EncoderBlock · level recorded 2026-09-22
MLP in 375 papers models/encoder/vit.py ran · metamorphic tier: deterministic fingerprinted MIT (permissive) · 4eb8553e435c726a · report fixture: returned MLP · level recorded 2026-09-22
MLPBlock in 17 papers models/encoder/vit.py ran · metamorphic tier: deterministic MIT (permissive) · 8d3ecda0f4496893 · report fixture: returned MLPBlock · level recorded 2026-09-22
_make_ntuple models/encoder/vit.py ran · violated contract MIT (permissive) · b7edf42d365259f4 · report fixture: signature x:list/1/int, n:int/0/int · returned tuple · level recorded 2026-09-22
Conv2dNormActivation in 6 papers models/encoder/vit.py unverified MIT (permissive) · c5e62920f023fc96 · report
Encoder in 278 papers models/encoder/vit.py unverified MIT (permissive) · 0fe366661e030443 · report
VisionTransformer in 53 papers models/encoder/vit.py unverified MIT (permissive) · 232688cf9e59bf54 · report
_log_api_usage_once in 8 papers models/encoder/vit.py unverified MIT (permissive) · 10e31764ca79cdcd · report
heekhero/DTL community (archive-listed) 11 samples, 7 ran (2 instrument-failed) · 4 unverified
LoRALayer in 10 papers classification/models/swin_transformer.py ran · metamorphic tier: deterministic no licence file found · pointer only · 5b0e9bae29a51df8 · report fixture: returned LoRALayer · level recorded 2026-09-22
Mlp in 102 papers classification/models/swin_transformer.py ran · metamorphic tier: deterministic no licence file found · pointer only · 22e33235319249df · report fixture: returned Mlp · level recorded 2026-09-22
PatchEmbed in 123 papers
census: 9 implementations from 9 papers, 6 distinct outputs on one shared input
classification/models/swin_transformer.py ran · metamorphic tier: deterministic no licence file found · pointer only · 641bf19387f9e143 · report fixture: returned PatchEmbed · level recorded 2026-09-22
PatchMerging in 31 papers classification/models/swin_transformer.py ran · metamorphic tier: deterministic no licence file found · pointer only · dce20c2c623e0f35 · report fixture: returned PatchMerging · level recorded 2026-09-22
WindowAttention in 50 papers classification/models/swin_transformer.py ran · metamorphic tier: deterministic no licence file found · pointer only · 0762b0fd2462f471 · report fixture: returned WindowAttention · level recorded 2026-09-22
window_partition in 377 papers
census: 13 implementations from 17 papers, 5 distinct outputs across 3 buckets, one shared input per bucket
classification/models/swin_transformer.py ran; the contract check could not run (RAISES) · metamorphic tier: invariant fingerprinted no licence file found · pointer only · 993ae96666b00cb5 · report fixture: signature x:Tensor/4/float, window_size:int/0/int · returned Tensor · level recorded 2026-09-22
window_reverse in 296 papers classification/models/swin_transformer.py ran · our draft was wrong no licence file found · pointer only · 609922bd93c75117 · report fixture: signature windows:Tensor/4/float, window_size:int/0/int, H:int/0/int, W:int/0/int · returned Tensor · level recorded 2026-09-22
BasicLayer in 56 papers classification/models/swin_transformer.py unverified no licence file found · pointer only · 5d05f06f1874a08e · report
SwinTransformer in 15 papers classification/models/swin_transformer.py unverified no licence file found · pointer only · 8ffa2914d7fefefa · report
SwinTransformerBlock in 35 papers classification/models/swin_transformer.py unverified no licence file found · pointer only · a619548a5271517f · report
write in 5 papers classification/models/swin_transformer.py unverified no licence file found · pointer only · f38e7de3675b6095 · report
wgcban/apt community (archive-listed) 3 samples, 2 ran (0 instrument-failed) · 1 unverified
Attention in 344 papers
census: 13 implementations from 12 papers, 10 distinct outputs across 4 buckets, one shared input per bucket
modeling_VPT.py ran licence not identified · pointer only · bc0e89d4bae92c5f · report fixture: returned Attention · level recorded 2026-09-22
Block in 252 papers
census: 5 implementations from 5 papers, 3 distinct outputs across 2 buckets, one shared input per bucket
modeling_VPT.py ran licence not identified · pointer only · a6526ff13db3ed18 · report fixture: signature x:Tensor/3/float, p_k:Tensor/4/float, p_v:Tensor/4/float, scale:tuple/1/Tensor · returned Tensor · level recorded 2026-09-24
PromptedVisionTransformer modeling_VPT.py unverified licence not identified · pointer only · b2323593bea9d7f5 · report
unites-lab/vpns community (archive-listed) 1 sample, 1 ran (0 instrument-failed) · 0 unverified
ExpansiveVisualPrompt visual_prompt.py ran · metamorphic tier: deterministic MIT (permissive) · 05a71e461d5e0fe7 · report fixture: returned ExpansiveVisualPrompt · level recorded 2026-09-22
Repository not located 1 sample, 1 ran (0 instrument-failed) · 0 unverified
1 sample is identical to code Syntology harvested from another paper's repository; this paper's own copy was not located, so no repository is named and no licence is recorded.
swish in 258 papers
census: 14 implementations from 37 papers, 2 distinct outputs on one shared input this sample: bucket 1 of 1, arg 1: rank 2, kind float, dtype float32
repository not located ran · honoured contract fingerprinted licence of this copy not recorded · 5f094438dfb16a0d · report fixture: signature x:Tensor/2/float, beta:float/0/float · returned Tensor · level recorded 2026-09-22

Method as code Syntology

What Syntology's graph records as this paper's claimed method, lifted into code: each assembly names its bricks (functions and classes), each with its own role and its own verification level. An assembly is not a check that the code implements the method correctly or reproduces the paper's results. Levels are each brick's own verification level in Syntology's graph (level 0: no recorded run), listed brick by brick and never combined into one level; “ran” means the brick executed on a synthesized input, not that its output is correct, and the instrument figure counts failures of Syntology's instrument, not of the code. Where a model read the code against the paper's claim, each assembly's label verdict is one model's judgement, kept in this page's JSON twin; it is not calibrated, so no card prints it. Siblings are assemblies lifted from this paper in the same batch for the same method: alternative readings kept side by side; none replaces or supersedes another.

5 assemblies on this paper: 5 harvested, 0 generated, listed in 1 groups (siblings lifted in one batch form one group).

5 sibling assemblies for Visual Prompt Tuning (harvested), lifted in the same batch on 2026-09-09: alternative readings kept side by side; none replaces or supersedes another.

harvested Assembly for Visual Prompt Tuning (graph method: Visual Prompt Tuning). 10 bricks; individually on a fixture: 6 ran (of which 5 constructed an object rather than computing a result; 6 with no instrument failure: 0 honoured, 1 violated, 5 with no contract checked; 0 where Syntology's instrument failed) · 4 unverified. The assembly has not been run end-to-end.

one of 5 siblings

  1. VisionTransformer entry · level 0 · unverified · 232688cf9e59bf54
  2. Conv2dNormActivation helper · level 0 · unverified · c5e62920f023fc96
  3. ConvStemConfig helper · level 1 · ran · a1aa4bfcabd828f1
  4. Encoder model · level 0 · unverified · 0fe366661e030443
  5. _log_api_usage_once helper · level 0 · unverified · 10e31764ca79cdcd
  6. ConvNormActivation helper · level 1 · ran · 3208de787e3ca751
  7. EncoderBlock model · level 1 · ran · c6bf5366cd01303c
  8. _make_ntuple helper · level 1 · ran · violated contract · b7edf42d365259f4
  9. MLPBlock model · level 1 · ran · 8d3ecda0f4496893
  10. MLP helper · level 1 · ran · 4eb8553e435c726a

harvested Assembly for Visual Prompt Tuning (graph method: Visual Prompt Tuning). 1 brick; individually on a fixture: 1 ran (of which 0 constructed an object rather than computing a result; 1 with no instrument failure: 0 honoured, 0 violated, 1 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified. The assembly has not been run end-to-end.

one of 5 siblings

  1. VPT entry · level 1 · ran · 3c7693b669414501

harvested Assembly for Visual Prompt Tuning (graph method: Visual Prompt Tuning). 1 brick; individually on a fixture: 1 ran (of which 1 constructed an object rather than computing a result; 1 with no instrument failure: 0 honoured, 0 violated, 1 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified; the one sample that ran constructed an object rather than computing a result. The assembly has not been run end-to-end.

one of 5 siblings

  1. ExpansiveVisualPrompt entry · level 1 · ran · 05a71e461d5e0fe7

harvested Assembly for Visual Prompt Tuning (graph method: Visual Prompt Tuning). 12 bricks; individually on a fixture: 8 ran (of which 5 constructed an object rather than computing a result; 6 with no instrument failure: 1 honoured, 0 violated, 5 with no contract checked; 2 where Syntology's instrument failed) · 4 unverified. The assembly has not been run end-to-end.

one of 5 siblings

  1. SwinTransformer entry · level 0 · unverified · 8ffa2914d7fefefa
  2. BasicLayer model · level 0 · unverified · 5d05f06f1874a08e
  3. LoRALayer helper · level 1 · ran · 5b0e9bae29a51df8
  4. PatchEmbed model · level 1 · ran · 641bf19387f9e143
  5. PatchMerging model · level 1 · ran · dce20c2c623e0f35
  6. write helper · level 0 · unverified · f38e7de3675b6095
  7. SwinTransformerBlock model · level 0 · unverified · a619548a5271517f
  8. swish helper · level 1 · ran · honoured contract · 5f094438dfb16a0d
  9. Mlp model · level 1 · ran · 22e33235319249df
  10. WindowAttention model · level 1 · ran · 0762b0fd2462f471
  11. window_partition helper · level 1 · ran; the contract check could not run (RAISES) · 993ae96666b00cb5
  12. window_reverse helper · level 1 · ran · our draft was wrong · 609922bd93c75117

harvested Assembly for Visual Prompt Tuning (graph method: Visual Prompt Tuning). 7 bricks; individually on a fixture: 6 ran (of which 4 constructed an object rather than computing a result; 6 with no instrument failure: 1 honoured, 0 violated, 5 with no contract checked; 0 where Syntology's instrument failed) · 1 unverified. The assembly has not been run end-to-end.

one of 5 siblings

  1. PromptedVisionTransformer entry · level 0 · unverified · b2323593bea9d7f5
  2. Block model · level 1 · ran · a6526ff13db3ed18
  3. PatchEmbed model · level 1 · ran · 59f1eed28e030f58
  4. get_sinusoid_encoding_table helper · level 1 · ran · honoured contract · da651e3979a18f84
  5. Attention model · level 1 · ran · bc0e89d4bae92c5f
  6. DropPath model · level 1 · ran · ea7880551588ee9f
  7. Mlp model · level 1 · ran · e71525ebf968c4c3

Each brick ends with its code_sha256, Syntology's identity for that exact code; click one to copy the get_code call that returns it. For agents, on Syntology's MCP service: assembly(arxiv_id="2203.12119") returns these assemblies with each brick's role and wiring (how to connect).

Tasks

Image ClassificationLong-tail LearningPrompt EngineeringVisual Prompt Tuning

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Long-tail Learning CIFAR-100-LT (ρ=10) VPT Error Rate 10.4 #3 of 31 Archive leaderboard report
Long-tail Learning CIFAR-100-LT (ρ=100) VPT Error Rate 19 #4 of 66 Archive leaderboard report
Long-tail Learning CIFAR-100-LT (ρ=50) VPT Error Rate 15.2 #3 of 25 Archive leaderboard report
Prompt Engineering ImageNet-21k VPT Accuracy 24.8 #2 of 2 Archive leaderboard report
Visual Prompt Tuning FGVC VPT-Deep(ViT-B/16_MoCo_v3_pretrained_ImageNet-1K) Mean Accuracy 83.12 #4 of 10 Archive leaderboard report
Visual Prompt Tuning FGVC VPT-Shallow (ViT-B/16_MoCo_v3_pretrained_ImageNet-1K) Mean Accuracy 79.26 #6 of 10 Archive leaderboard report
Visual Prompt Tuning FGVC VPT-Deep (ViT-B/16_MAE_pretrained_ImageNet-1K) Mean Accuracy 72.02 #9 of 10 Archive leaderboard report
Visual Prompt Tuning FGVC VPT-Shallow (ViT-B/16_MAE_pretrained_ImageNet-1K) Mean Accuracy 57.84 #10 of 10 Archive leaderboard report
Visual Prompt Tuning VTAB-1k(Natural<7>) VPT-Deep(ViT-B/16_MoCo_v3_pretrained_ImageNet-1K) Mean Accuracy 70.27 #4 of 10 Archive leaderboard report
Visual Prompt Tuning VTAB-1k(Natural<7>) VPT-Shallow(ViT-B/16_MoCo_v3_pretrained_ImageNet-1K) Mean Accuracy 67.34 #5 of 10 Archive leaderboard report
Visual Prompt Tuning VTAB-1k(Natural<7>) VPT-Shallow(ViT-B/16_MAE_pretrained_ImageNet-1K) Mean Accuracy 39.96 #9 of 10 Archive leaderboard report
Visual Prompt Tuning VTAB-1k(Natural<7>) VPT-Deep(ViT-B/16_MAE_pretrained_ImageNet-1K) Mean Accuracy 36.02 #10 of 10 Archive leaderboard report
Visual Prompt Tuning VTAB-1k(Specialized<4>) VPT-Deep(ViT-B/16_MoCo_v3_pretrained_ImageNet-1K) Mean Accuracy 83.04 #5 of 10 Archive leaderboard report
Visual Prompt Tuning VTAB-1k(Specialized<4>) VPT-Shallow(ViT-B/16_MoCo_v3_pretrained_ImageNet-1K) Mean Accuracy 82.26 #6 of 10 Archive leaderboard report
Visual Prompt Tuning VTAB-1k(Specialized<4>) VPT-Shallow(ViT-B/16_MAE_pretrained_ImageNet-1K) Mean Accuracy 69.65 #9 of 10 Archive leaderboard report
Visual Prompt Tuning VTAB-1k(Specialized<4>) VPT-Deep(ViT-B/16_MAE_pretrained_ImageNet-1K) Mean Accuracy 60.61 #10 of 10 Archive leaderboard report
Visual Prompt Tuning VTAB-1k(Structured<8>) VPT-Deep(ViT-B/16_MoCo_v3_pretrained_ImageNet-1K) Mean Accuracy 42.38 #6 of 10 Archive leaderboard report
Visual Prompt Tuning VTAB-1k(Structured<8>) VPT-Shallow(ViT-B/16_MoCo_v3_pretrained_ImageNet-1K) Mean Accuracy 37.55 #7 of 10 Archive leaderboard report
Visual Prompt Tuning VTAB-1k(Structured<8>) VPT-Shallow(ViT-B/16_MAE_pretrained_ImageNet-1K) Mean Accuracy 27.50 #9 of 10 Archive leaderboard report
Visual Prompt Tuning VTAB-1k(Structured<8>) VPT-Deep(ViT-B/16_MAE_pretrained_ImageNet-1K) Mean Accuracy 26.57 #10 of 10 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

Absolute Position EncodingsAdamAttentionBPEDense ConnectionsDropoutLabel SmoothingLayer NormalizationLinear LayerMulti-Head AttentionPosition-Wise Feed-Forward LayerResidual ConnectionSoftmaxTransformer

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