Papers › Zero-Shot Video Question Answering via Frozen Bidirectional Language Models

Zero-Shot Video Question Answering via Frozen Bidirectional Language Models

16 Jun 2022arXiv:2206.08155archive 2025-07-28

Antoine Yang, Antoine Miech, Josef Sivic, Ivan Laptev, Cordelia Schmid

Video question answering (VideoQA) is a complex task that requires diverse multi-modal data for training. Manual annotation of question and answers for videos, however, is tedious and prohibits scalability. To tackle this problem, recent methods consider zero-shot settings with no manual annotation of visual question-answer. In particular, a promising approach adapts frozen autoregressive language models pretrained on Web-scale text-only data to multi-modal inputs. In contrast, we here build on frozen bidirectional language models (BiLM) and show that such an approach provides a stronger and cheaper alternative for zero-shot VideoQA. In particular, (i) we combine visual inputs with the frozen BiLM using light trainable modules, (ii) we train such modules using Web-scraped multi-modal data, and finally (iii) we perform zero-shot VideoQA inference through masked language modeling, where the masked text is the answer to a given question. Our proposed approach, FrozenBiLM, outperforms the state of the art in zero-shot VideoQA by a significant margin on a variety of datasets, including LSMDC-FiB, iVQA, MSRVTT-QA, MSVD-QA, ActivityNet-QA, TGIF-FrameQA, How2QA and TVQA. It also demonstrates competitive performance in the few-shot and fully-supervised setting. Our code and models are publicly available at https://github.com/antoyang/FrozenBiLM.

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

Syntology 34 code samples: 33 from 3 repositories linked to this paper, 1 identical to code harvested from another paper's repository. Across all 34: 14 ran (of which 9 constructed an object rather than computing a result; 14 with no instrument failure: 1 honoured, 1 violated, 12 with no contract checked; 0 where Syntology's instrument failed) · 20 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code; here there are none.

By repository: official repository: 7 samples from 1 repository, 0 ran (0 with no instrument failure; 0 where Syntology's instrument failed); community (archive-listed): 26 samples from 2 repositories, 14 ran (14 with no instrument failure; 0 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.

antoyang/FrozenBiLM officialmentioned in papermentioned on GitHubpytorchSyntology: 7 samples, 0 ran · 7 unverified report
klauscc/dam mentioned on GitHubpytorchSyntology: 4 samples, 0 ran · 4 unverified report
sts-vlcc/sts-vlcc mentioned on GitHubpytorchSyntology: 22 samples, 14 ran (0 instrument-failed) · 8 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

34 samples harvested: 14 ran (of which 9 constructed an object rather than computing a result; 14 with no instrument failure: 1 honoured, 1 violated, 12 with no contract checked; 0 where Syntology's instrument failed) · 20 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
12ran
20unverified

Licence: 1 of the 34 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; 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 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.

antoyang/FrozenBiLM official repository 7 samples, 0 ran · 7 unverified
DebertaV2Attention model/deberta.py unverified Apache-2.0 (permissive) · ed339d8069718237 · report
DebertaV2Encoder model/deberta.py unverified Apache-2.0 (permissive) · deded63a5e7d8699 · report
DebertaV2Layer model/deberta.py unverified Apache-2.0 (permissive) · 32cb0a816174decd · report
DebertaV2Model model/deberta.py unverified Apache-2.0 (permissive) · f576b70e9c2bc9c4 · report
DisentangledSelfAttention model/deberta.py unverified Apache-2.0 (permissive) · b2ac2b6c1c0b64dc · report
build_relative_position in 17 papers model/deberta.py unverified Apache-2.0 (permissive) · b1852809457a4b1f · report
make_log_bucket_position in 15 papers model/deberta.py unverified Apache-2.0 (permissive) · 29371b9c270911e1 · report
klauscc/dam community (archive-listed) 4 samples, 0 ran · 4 unverified
BaseModelOutput model/deberta.py unverified Apache-2.0 (permissive) · 5c632e836a01b0f1 · report
DebertaV2Embeddings model/deberta.py unverified Apache-2.0 (permissive) · d402090dd56a6f46 · report
DebertaV2Encoder model/deberta.py unverified Apache-2.0 (permissive) · cc3b22c7c43000bf · report
DebertaV2Model model/deberta.py unverified Apache-2.0 (permissive) · 63c2fffe385cf5e9 · report
sts-vlcc/sts-vlcc community (archive-listed) 22 samples, 14 ran (0 instrument-failed) · 8 unverified
Adapter in 28 papers model/deberta.py ran · metamorphic tier: deterministic Apache-2.0 (permissive) · 020a7d784fec3e8b · report fixture: returned Adapter · level recorded 2026-09-22
ConvLayer in 26 papers model/deberta.py ran · metamorphic tier: deterministic Apache-2.0 (permissive) · be2fdff3546c4ee9 · report fixture: returned ConvLayer · level recorded 2026-09-22
DebertaV2Embeddings model/deberta.py ran · metamorphic tier: deterministic Apache-2.0 (permissive) · fb069d6e466f4cf8 · report fixture: returned DebertaV2Embeddings · level recorded 2026-09-22
DebertaV2Intermediate model/deberta.py ran fingerprinted Apache-2.0 (permissive) · 713015c84afd123d · report fixture: signature hidden_states:Tensor/2/float · returned Tensor · level recorded 2026-09-24
DebertaV2LMPredictionHead model/deberta.py ran fingerprinted Apache-2.0 (permissive) · 4f4202be5c178fde · report fixture: signature hidden_states:Tensor/3/float, embedding_weight:Tensor/2/float, bias:NoneType/0/? · returned Tensor · level recorded 2026-09-24
DebertaV2OnlyMLMHead model/deberta.py ran fingerprinted Apache-2.0 (permissive) · 491522c8a24566a6 · report fixture: signature sequence_output:Tensor/3/float, embedding_weight:Tensor/2/float, bias:NoneType/0/? · returned Tensor · level recorded 2026-09-24
DebertaV2Output model/deberta.py ran · metamorphic tier: deterministic Apache-2.0 (permissive) · d3b79844fb5a1268 · report fixture: returned DebertaV2Output · level recorded 2026-09-22
DebertaV2PreTrainedModel model/deberta.py ran · metamorphic tier: deterministic Apache-2.0 (permissive) · 48742f01e32d514b · report fixture: returned DebertaV2PreTrainedModel · level recorded 2026-09-22
DebertaV2SelfOutput model/deberta.py ran · metamorphic tier: deterministic Apache-2.0 (permissive) · a14cfd37805f8d1a · report fixture: returned DebertaV2SelfOutput · level recorded 2026-09-22
StableDropout model/deberta.py ran · metamorphic tier: deterministic fingerprinted Apache-2.0 (permissive) · 104d374e92036a6f · report fixture: returned StableDropout · level recorded 2026-09-22
XDropout model/deberta.py ran Apache-2.0 (permissive) · 1f587485f1c0247d · report fixture: returned XDropout · level recorded 2026-09-22
XSoftmax model/deberta.py ran Apache-2.0 (permissive) · d4b3d76edcfa3131 · report fixture: returned XSoftmax · level recorded 2026-09-22
build_relative_position in 17 papers model/deberta.py ran · honoured contract fingerprinted Apache-2.0 (permissive) · 0464dc8467b456dc · report fixture: signature query_size:int/0/int, key_size:int/0/int, bucket_size:int/0/int, max_position:int/0/int · returned Tensor · level recorded 2026-09-22
make_log_bucket_position in 15 papers model/deberta.py ran · violated contract Apache-2.0 (permissive) · 567c8b010f8e20d2 · report fixture: signature relative_pos:ndarray/1/int, bucket_size:int/0/int, max_position:int/0/int · returned ndarray · level recorded 2026-09-22
BaseModelOutput model/deberta.py unverified Apache-2.0 (permissive) · 06f89381bbcaa62a · report
DebertaV2Attention model/deberta.py unverified Apache-2.0 (permissive) · 19240ea1e9c179bc · report
DebertaV2Encoder model/deberta.py unverified Apache-2.0 (permissive) · 0905316df8bf0d3e · report
DebertaV2ForMaskedLM model/deberta.py unverified Apache-2.0 (permissive) · 24dd2616a44d83c7 · report
DebertaV2Layer model/deberta.py unverified Apache-2.0 (permissive) · b76e671720991509 · report
DebertaV2Model model/deberta.py unverified Apache-2.0 (permissive) · 6a90151a5129dc8e · report
DisentangledSelfAttention model/deberta.py unverified Apache-2.0 (permissive) · f5ca472802a4cfb0 · report
DropoutContext model/deberta.py unverified Apache-2.0 (permissive) · ae0fc5a9e16f1c5d · report
Repository not located 1 sample, 0 ran · 1 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.
get_mask in 119 papers
census: 9 implementations from 10 papers, 9 distinct outputs across 7 buckets, one shared input per bucket
repository not located unverified licence of this copy not recorded · 05732f1a5126672d · 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. 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.

3 assemblies on this paper: 3 harvested, 0 generated, listed in 1 groups (siblings lifted in one batch form one group).

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

harvested Assembly for FrozenBiLM (graph method: FrozenBiLM). 23 bricks; individually on a fixture: 14 ran (of which 9 constructed an object rather than computing a result; 14 with no instrument failure: 1 honoured, 1 violated, 12 with no contract checked; 0 where Syntology's instrument failed) · 9 unverified. The assembly has not been run end-to-end.

one of 3 siblings

  1. DebertaV2ForMaskedLM entry · level 0 · unverified · 24dd2616a44d83c7
  2. DebertaV2Model model · level 0 · unverified · 6a90151a5129dc8e
  3. DebertaV2OnlyMLMHead model · level 1 · ran · 491522c8a24566a6
  4. DebertaV2PreTrainedModel helper · level 1 · ran · 48742f01e32d514b
  5. BaseModelOutput helper · level 0 · unverified · 06f89381bbcaa62a
  6. DebertaV2Embeddings model · level 1 · ran · fb069d6e466f4cf8
  7. DebertaV2Encoder model · level 0 · unverified · 0905316df8bf0d3e
  8. DebertaV2LMPredictionHead helper · level 1 · ran · 4f4202be5c178fde
  9. StableDropout helper · level 1 · ran · 104d374e92036a6f
  10. ConvLayer helper · level 1 · ran · be2fdff3546c4ee9
  11. DebertaV2Layer helper · level 0 · unverified · b76e671720991509
  12. build_relative_position helper · level 1 · ran · honoured contract · 0464dc8467b456dc
  13. DropoutContext helper · level 0 · unverified · ae0fc5a9e16f1c5d
  14. XDropout helper · level 1 · ran · 1f587485f1c0247d
  15. DebertaV2Attention helper · level 0 · unverified · 19240ea1e9c179bc
  16. DebertaV2Intermediate helper · level 1 · ran · 713015c84afd123d
  17. DebertaV2Output helper · level 1 · ran · d3b79844fb5a1268
  18. make_log_bucket_position helper · level 1 · ran · violated contract · 567c8b010f8e20d2
  19. get_mask helper · level 0 · unverified · 05732f1a5126672d
  20. DebertaV2SelfOutput helper · level 1 · ran · a14cfd37805f8d1a
  21. DisentangledSelfAttention helper · level 0 · unverified · f5ca472802a4cfb0
  22. Adapter helper · level 1 · ran · 020a7d784fec3e8b
  23. XSoftmax helper · level 1 · ran · d4b3d76edcfa3131

harvested Assembly for FrozenBiLM (graph method: FrozenBiLM). 20 bricks; individually on a fixture: 10 ran (of which 9 constructed an object rather than computing a result; 10 with no instrument failure: 0 honoured, 0 violated, 10 with no contract checked; 0 where Syntology's instrument failed) · 10 unverified. The assembly has not been run end-to-end.

one of 3 siblings

  1. DebertaV2Model entry · level 0 · unverified · f576b70e9c2bc9c4
  2. BaseModelOutput helper · level 0 · unverified · 06f89381bbcaa62a
  3. DebertaV2Embeddings model · level 1 · ran · fb069d6e466f4cf8
  4. DebertaV2Encoder model · level 0 · unverified · deded63a5e7d8699
  5. DebertaV2PreTrainedModel helper · level 1 · ran · 48742f01e32d514b
  6. StableDropout helper · level 1 · ran · 104d374e92036a6f
  7. ConvLayer helper · level 1 · ran · be2fdff3546c4ee9
  8. DebertaV2Layer model · level 0 · unverified · 32cb0a816174decd
  9. build_relative_position helper · level 0 · unverified · b1852809457a4b1f
  10. DropoutContext helper · level 0 · unverified · ae0fc5a9e16f1c5d
  11. XDropout helper · level 1 · ran · 1f587485f1c0247d
  12. DebertaV2Attention helper · level 0 · unverified · ed339d8069718237
  13. DebertaV2Intermediate helper · level 1 · ran · 713015c84afd123d
  14. DebertaV2Output helper · level 1 · ran · d3b79844fb5a1268
  15. make_log_bucket_position helper · level 0 · unverified · 29371b9c270911e1
  16. get_mask helper · level 0 · unverified · 05732f1a5126672d
  17. DebertaV2SelfOutput helper · level 1 · ran · a14cfd37805f8d1a
  18. DisentangledSelfAttention model · level 0 · unverified · b2ac2b6c1c0b64dc
  19. Adapter helper · level 1 · ran · 020a7d784fec3e8b
  20. XSoftmax helper · level 1 · ran · d4b3d76edcfa3131

harvested Assembly for FrozenBiLM (graph method: FrozenBiLM). 20 bricks; individually on a fixture: 9 ran (of which 8 constructed an object rather than computing a result; 9 with no instrument failure: 0 honoured, 0 violated, 9 with no contract checked; 0 where Syntology's instrument failed) · 11 unverified. The assembly has not been run end-to-end.

one of 3 siblings

  1. DebertaV2Model entry · level 0 · unverified · 63c2fffe385cf5e9
  2. BaseModelOutput helper · level 0 · unverified · 5c632e836a01b0f1
  3. DebertaV2Embeddings model · level 0 · unverified · d402090dd56a6f46
  4. DebertaV2Encoder model · level 0 · unverified · cc3b22c7c43000bf
  5. DebertaV2PreTrainedModel helper · level 1 · ran · 48742f01e32d514b
  6. StableDropout helper · level 1 · ran · 104d374e92036a6f
  7. ConvLayer helper · level 1 · ran · be2fdff3546c4ee9
  8. DebertaV2Layer model · level 0 · unverified · 32cb0a816174decd
  9. build_relative_position helper · level 0 · unverified · b1852809457a4b1f
  10. DropoutContext helper · level 0 · unverified · ae0fc5a9e16f1c5d
  11. XDropout helper · level 1 · ran · 1f587485f1c0247d
  12. DebertaV2Attention model · level 0 · unverified · ed339d8069718237
  13. DebertaV2Intermediate model · level 1 · ran · 713015c84afd123d
  14. DebertaV2Output model · level 1 · ran · d3b79844fb5a1268
  15. make_log_bucket_position helper · level 0 · unverified · 29371b9c270911e1
  16. get_mask helper · level 0 · unverified · 05732f1a5126672d
  17. DebertaV2SelfOutput helper · level 1 · ran · a14cfd37805f8d1a
  18. DisentangledSelfAttention model · level 0 · unverified · b2ac2b6c1c0b64dc
  19. Adapter helper · level 1 · ran · 020a7d784fec3e8b
  20. XSoftmax helper · level 1 · ran · d4b3d76edcfa3131

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="2206.08155") returns these assemblies with each brick's role and wiring (how to connect).

Tasks

Fill MaskLanguage ModelingLanguage ModellingMasked Language ModelingQuestion AnsweringTGIF-FrameVideo Question AnsweringVisual Question AnsweringVisual Question Answering (VQA)Zero-Shot LearningZero-Shot Video Question AnswerZeroshot Video Question Answer

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Video Question Answering ActivityNet-QA FrozenBiLM Accuracy 43.2 #21 of 36 Archive leaderboard report
Video Question Answering ActivityNet-QA FrozenBiLM (0-shot) Accuracy 25.9 #34 of 36 Archive leaderboard report
Video Question Answering How2QA FrozenBiLM Accuracy 86.7 #2 of 8 Archive leaderboard report
Video Question Answering How2QA FrozenBiLM (0-shot) Accuracy 58.4 #7 of 8 Archive leaderboard report
Video Question Answering MSRVTT-QA FrozenBiLM Accuracy 47.0 #7 of 14 Archive leaderboard report
Video Question Answering MSRVTT-QA FrozenBiLM (0-shot) Accuracy 16.7 #14 of 14 Archive leaderboard report
Video Question Answering TVQA FrozenBiLM Accuracy 82 #2 of 6 Archive leaderboard report
Video Question Answering iVQA FrozenBiLM Accuracy 39.6 #2 of 7 Archive leaderboard report
Video Question Answering iVQA FrozenBiLM (0-shot) Accuracy 26.8 #6 of 7 Archive leaderboard report
Visual Question Answering MSRVTT-QA FrozenBiLM Accuracy 0.470 #2 of 4 Archive leaderboard report
Visual Question Answering MSVD-QA FrozenBiLM Accuracy 0.548 #1 of 2 Archive leaderboard report
Zero-Shot Learning LSMDC FrozenBiLM Accuracy 51.5 #1 of 1 Archive leaderboard report
Zero-Shot Learning iVQA FrozenBiLM Accuracy 0.268 #1 of 1 Archive leaderboard report
Zero-Shot Video Question Answer ActivityNet-QA FrozenBiLM Accuracy 24.7 #27 of 28 Archive leaderboard report
Zero-Shot Video Question Answer ActivityNet-QA FrozenBiLM Confidence Score - #27 of 28 Archive leaderboard report
Zero-Shot Video Question Answer EgoSchema (fullset) FrozenBiLM Accuracy 26.9 #27 of 29 Archive leaderboard report
Zero-Shot Video Question Answer MSVD-QA FrozenBiLM Accuracy 33.8 #28 of 28 Archive leaderboard report
Zero-Shot Video Question Answer TGIF-QA FrozenBiLM Accuracy 41.9 #13 of 14 Archive leaderboard report
Zero-Shot Video Question Answer TVQA FrozenBiLM (with speech) Accuracy 59.7 #1 of 9 Archive leaderboard report
Zero-Shot Video Question Answer TVQA FrozenBILM (no speech) Accuracy 29.7 #9 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.

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