Papers › Large-Scale Adversarial Training for Vision-and-Language Representation Learning

Large-Scale Adversarial Training for Vision-and-Language Representation Learning

11 Jun 2020NeurIPS 2020 12arXiv:2006.06195archive 2025-07-28

Zhe Gan, Yen-Chun Chen, Linjie Li, Chen Zhu, Yu Cheng, Jingjing Liu

We present VILLA, the first known effort on large-scale adversarial training for vision-and-language (V+L) representation learning. VILLA consists of two training stages: (i) task-agnostic adversarial pre-training; followed by (ii) task-specific adversarial finetuning. Instead of adding adversarial perturbations on image pixels and textual tokens, we propose to perform adversarial training in the embedding space of each modality. To enable large-scale training, we adopt the "free" adversarial training strategy, and combine it with KL-divergence-based regularization to promote higher invariance in the embedding space. We apply VILLA to current best-performing V+L models, and achieve new state of the art on a wide range of tasks, including Visual Question Answering, Visual Commonsense Reasoning, Image-Text Retrieval, Referring Expression Comprehension, Visual Entailment, and NLVR2.

PaperPDFConference PDFCodeCode 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.

Syntology holds a reference implementation of VILLA, generated from the paper text (not the authors' code), at level 3 on Syntology's ladder: V2 “two implementations GENERATED from the paper's signature and spec agree numerically (1e-6) on fixture inputs” “neither of the two ever read the authors' own code, so V2 measures how unambiguously the paper specifies the method, not fidelity to the authors' implementation (measured 2026-09-11, commit 75eaa77: cross_checked_by is a second LLM on 2,628 of 2,831 served samples)”; V3 “paper-derived property tests pass on both of those same two generations.” (quoted from Syntology's server card). For this implementation, level 3 is recorded, but its report lists no property test that passed on both. No rung compares this code against the authors' implementation. For agents: get_reference_implementation(name='VILLA') (how to connect); the free check have('2006.06195') needs no token.

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

Code

Syntology 20 code samples harvested from 2 repositories linked to this paper: 12 ran (of which 5 constructed an object rather than computing a result; 9 with no instrument failure: 1 honoured, 0 violated, 8 with no contract checked; 3 where Syntology's instrument failed) · 8 unverified. Where Syntology's instrument failed: 2 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: official repository: 20 samples from 2 repositories, 12 ran (9 with no instrument failure; 3 where Syntology's instrument failed). 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 2026-09-25 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed. 1 of the 12 samples that ran has no recorded date.

zhegan27/LXMERT-AdvTrain officialmentioned in papermentioned on GitHubpytorchSyntology: 11 samples, 6 ran (0 instrument-failed) · 5 unverified report
zhegan27/VILLA officialmentioned in papermentioned on GitHubpytorchMITSyntology: 9 samples, 6 ran (3 instrument-failed) · 3 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

20 samples harvested: 12 ran (of which 5 constructed an object rather than computing a result; 9 with no instrument failure: 1 honoured, 0 violated, 8 with no contract checked; 3 where Syntology's instrument failed) · 8 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
2ran · our draft was wrong
1ran; the contract check could not run
8ran
8unverified

Licence: 8 of the 20 samples are pointer only, meaning Syntology does not serve that copy's text. For 2 of them, this paper's copy is under a permissive licence and cleared, but the stored copy of the same code that Syntology serves is not; the licence label's tooltip gives Syntology's recorded reason. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from 2 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. “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 3 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.

zhegan27/LXMERT-AdvTrain official repository 11 samples, 6 ran (0 instrument-failed) · 5 unverified
BertEmbeddings in 21 papers src/lxrt/modeling.py ran MIT (permissive) · 7cd6bab68e15976d · report fixture: signature input_ids:Tensor/2/int, token_type_ids:Tensor/2/int, adv_training:bool/0/bool, adv_modality:NoneType/0/?, adv_delta:NoneType/0/? · returned Tensor · level recorded 2026-09-24
BertPredictionHeadTransform in 10 papers src/lxrt/modeling.py ran · metamorphic tier: deterministic fingerprinted MIT (permissive) · 2ba3b1dfb07ba4b1 · report fixture: returned BertPredictionHeadTransform · level recorded 2026-09-22
BertVisualAnswerHead src/lxrt/modeling.py ran · metamorphic tier: deterministic fingerprinted MIT (permissive) · 2c8c022f71d9ae6e · report fixture: returned BertVisualAnswerHead · level recorded 2026-09-22
BertVisualObjHead src/lxrt/modeling.py ran · metamorphic tier: deterministic MIT (permissive) · 0e7be2781b71b508 · report fixture: returned BertVisualObjHead · level recorded 2026-09-22
GeLU src/lxrt/modeling.py ran · metamorphic tier: deterministic fingerprinted MIT (permissive) · ce655ad320784ce8 · report fixture: returned GeLU · level recorded 2026-09-22
VisualFeatEncoder src/lxrt/modeling.py ran · metamorphic tier: deterministic MIT (permissive) · 1bea2e32e2dc619d · report fixture: returned VisualFeatEncoder · level recorded 2026-09-22
BertLMPredictionHead in 10 papers src/lxrt/modeling.py unverified MIT (permissive) · 77ff902ef6858211 · report
BertPreTrainingHeads in 5 papers src/lxrt/modeling.py unverified MIT (permissive) · 781e8b75f1ca0f7c · report
LXRTEncoder src/lxrt/modeling.py unverified MIT (permissive) · 2c0fd477bfbddac7 · report
LXRTModel src/lxrt/modeling.py unverified MIT (permissive) · 95e61302335909fb · report
LXRTPretraining src/lxrt/modeling.py unverified MIT (permissive) · 6f76d36162ffef27 · report
zhegan27/VILLA official repository 9 samples, 6 ran (3 instrument-failed) · 3 unverified
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
model/layer.py ran · our draft was wrong fingerprinted MIT recorded; this copy not marked cleared · pointer only · 0f786c407fb1ee4c · report fixture: signature x:Tensor/2/float · returned Tensor · level recorded, date not recorded
compute_score_with_logits in 12 papers inf_vqa.py ran fingerprinted MIT recorded; this copy not marked cleared · pointer only · 2731fd9de6d31bc0 · report fixture: signature logits:Tensor/2/float, labels:Tensor/2/float · returned Tensor · level recorded 2026-09-25
cost_matrix_cosine model/ot.py ran; the contract check could not run (RAISES) · metamorphic tier: invariant fingerprinted MIT (permissive) here; the stored copy is pointer only · 9c9d2823696faa43 · report fixture: signature x:Tensor/3/float, y:Tensor/3/float, eps:float/0/float · returned Tensor · level recorded 2026-09-22
gelu in 308 papers
census: 43 implementations from 74 papers, 6 distinct outputs on one shared input this sample: bucket 1 of 1, arg 1: rank 2, kind float, dtype float32
model/layer.py ran · honoured contract fingerprinted MIT recorded; this copy not marked cleared · pointer only · fdc64f4c72036ae4 · report fixture: signature x:Tensor/2/float · returned Tensor · level recorded 2026-09-22
multi_head_attention_forward in 73 papers model/attention.py ran · our draft was wrong MIT (permissive) · fed42c63be2ceb6c · report fixture: signature query:Tensor/3/float, key:Tensor/3/float, value:Tensor/3/float, embed_dim_to_check:int/0/int, num_heads:int/0/int, in_proj_weight:Tensor/2/float, in_proj_bias:Tensor/1/float, bias_k:NoneType/0/?, bias_v:NoneType/0/?, add_zero_attn:bool/0/bool, dropout_p:float/0/float, out_proj_weight:Tensor/2/float, out_proj_bias:Tensor/1/float, training:bool/0/bool, key_padding_mask:NoneType/0/?, need_weights:bool/0/bool, attn_mask:NoneType/0/?, use_separate_proj_weight:bool/0/bool, q_proj_weight:NoneType/0/?, k_proj_weight:NoneType/0/?, v_proj_weight:NoneType/0/?, static_k:NoneType/0/?, static_v:NoneType/0/? · returned tuple · level recorded 2026-09-22
save_for_submission inf_vcr.py ran MIT recorded; this copy not marked cleared · pointer only · 70361102cdeaaf85 · report fixture: signature pred_file:str/0/str · returned DataFrame · level recorded 2026-09-25
evaluate in 514 papers inf_nlvr2.py unverified MIT recorded; this copy not marked cleared · pointer only · 28e194bd5c984ae7 · report
ipot model/ot.py unverified MIT (permissive) here; the stored copy is pointer only · cbf43203110456f0 · report
trace in 6 papers model/ot.py unverified MIT recorded; this copy not marked cleared · pointer only · d3f4c62a97861b9d · report

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. For a generated assembly, an independent peer is a second implementation generated separately from the same paper text; neither implementation read the authors' code. At level 2 the two agreed numerically on the fixture's inputs; level 3 is recorded when property tests derived from the paper also passed on both, and each level 3 card prints how many its report lists. 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. A generated assembly's label was set when its code was generated for the method, and no model judged it; its card prints it as "by construction" or "not judged by a model". Where an assembly's phrase has no constructed figure, a brick that ran is not among the harvested samples this build reads, and whether a brick's run only constructed an object is read from that brick's harvested sample, so the figure was not counted; it does not say there were none.

2 assemblies on this paper: 1 harvested, 1 generated.

generated Assembly for VILLA (graph method: VILLA), generated from the paper text, not the repository's code. 1 brick; individually on a fixture: 1 ran (1 with no instrument failure: 1 honoured, 0 violated, 0 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified. Ran end-to-end against an independent peer; level 3 is recorded, but its report lists no property test that passed on both.

label matches by construction: the code was generated for this method (not judged by a model)

  1. project_perturbation_to_frobenius_ball entry · level 3 · ran · honoured contract · d1b8bc71e189a603f704d0410ab3f6e55daa302389111610737fabcf4e0255d2

harvested Assembly for VILLA (graph method: VILLA). 34 bricks; individually on a fixture: 19 ran (of which 11 constructed an object rather than computing a result; 16 with no instrument failure: 1 honoured, 0 violated, 15 with no contract checked; 3 where Syntology's instrument failed) · 15 unverified. The assembly has not been run end-to-end.

  1. LXRTPretraining entry · level 0 · unverified · 6f76d36162ffef27
  2. BertPreTrainedModel helper · level 0 · unverified · 805f9fc004ffa11b
  3. BertPreTrainingHeads model · level 0 · unverified · 781e8b75f1ca0f7c
  4. BertVisualAnswerHead model · level 1 · ran · 2c8c022f71d9ae6e
  5. BertVisualObjHead model · level 1 · ran · 0e7be2781b71b508
  6. LXRTModel helper · level 0 · unverified · 95e61302335909fb
  7. BertConfig helper · level 1 · ran · 5324d31f8ea9ca59
  8. cached_path helper · level 0 · unverified · a4e8483353cec520
  9. load_tf_weights_in_bert helper · level 0 · unverified · 7099305544368003
  10. BertLMPredictionHead model · level 0 · unverified · 77ff902ef6858211
  11. GeLU helper · level 1 · ran · ce655ad320784ce8
  12. BertPredictionHeadTransform model · level 1 · ran · 2ba3b1dfb07ba4b1
  13. BertEmbeddings model · level 1 · ran · 7cd6bab68e15976d
  14. BertPooler model · level 1 · ran · 8f1a0b9437ccc562
  15. LXRTEncoder helper · level 0 · unverified · 2c0fd477bfbddac7
  16. get_from_cache helper · level 0 · unverified · 242c3c3d7e930e18
  17. gelu helper · level 1 · ran · honoured contract · fdc64f4c72036ae4
  18. BertLayer helper · level 0 · unverified · 2f7d2c26e185cb3f
  19. LXRTXLayer helper · level 0 · unverified · 0b1bea7c4d0afd60
  20. VisualFeatEncoder model · level 1 · ran · 1bea2e32e2dc619d
  21. http_get helper · level 0 · unverified · 18f49ada16168f38
  22. s3_etag helper · level 0 · unverified · 51dd06ec77c8a8d1
  23. s3_get helper · level 0 · unverified · 1079045e75efbf23
  24. url_to_filename helper · level 1 · ran · our draft was wrong · 5eff22fa0a651276
  25. BertIntermediate model · level 1 · ran · c7123d4962da3767
  26. BertOutput model · level 1 · ran · 61f383a856ac089e
  27. BertSelfattLayer model · level 1 · ran · 9c6012122feb6f81
  28. BertCrossattLayer model · level 1 · ran · 0e0a344be52d2fdc
  29. s3_request helper · level 1 · ran · our draft was wrong · 7661139be003d514
  30. split_s3_path helper · level 0 · unverified · d3852f416bfe86c3
  31. BertAttOutput helper · level 1 · ran · 5ccefc0113e7b26e
  32. BertAttention model · level 1 · ran · b8c495ff176f844a
  33. VisualConfig helper · level 1 · ran · 3644fb46790a5387
  34. swish helper · level 2 · ran · our draft was wrong · 0f786c407fb1ee4c

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="2006.06195") returns these assemblies with each brick's role and wiring (how to connect).

Tasks

Image-text RetrievalQuestion AnsweringReferring ExpressionReferring Expression ComprehensionRepresentation LearningRetrievalText RetrievalVisual Commonsense ReasoningVisual EntailmentVisual Question AnsweringVisual Question Answering (VQA)Visual Reasoning

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Visual Entailment SNLI-VE val VILLA-LARGE Accuracy 80.18 #7 of 9 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.

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