Papers › Revealing Single Frame Bias for Video-and-Language Learning

Revealing Single Frame Bias for Video-and-Language Learning

7 Jun 2022arXiv:2206.03428archive 2025-07-28

Jie Lei, Tamara L. Berg, Mohit Bansal

Training an effective video-and-language model intuitively requires multiple frames as model inputs. However, it is unclear whether using multiple frames is beneficial to downstream tasks, and if yes, whether the performance gain is worth the drastically-increased computation and memory costs resulting from using more frames. In this work, we explore single-frame models for video-and-language learning. On a diverse set of video-and-language tasks (including text-to-video retrieval and video question answering), we show the surprising result that, with large-scale pre-training and a proper frame ensemble strategy at inference time, a single-frame trained model that does not consider temporal information can achieve better performance than existing methods that use multiple frames for training. This result reveals the existence of a strong "static appearance bias" in popular video-and-language datasets. Therefore, to allow for a more comprehensive evaluation of video-and-language models, we propose two new retrieval tasks based on existing fine-grained action recognition datasets that encourage temporal modeling. Our code is available at https://github.com/jayleicn/singularity

PaperPDFCodeCode Syntology ran

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

No sample from the archive's official repository (jayleicn/singularity) is linked to this paper; the 12 samples below come from 1 community repository. Syntology harvested code from jayleicn/singularity for another paper.

jayleicn/singularity: Syntology's graph links these samples to Test of Time: Instilling Video-Language Models with a Sense of Time (arXiv:2301.02074; that paper's own run record: 10 ran (of which 0 constructed an object rather than computing a result; 9 with no instrument failure: 0 honoured, 0 violated, 9 with no contract checked; 1 where Syntology's instrument failed) · 0 unverified); nothing checks that they implement this paper's method.

Syntology 12 code samples harvested from 1 repository linked to this paper: 11 ran (of which 0 constructed an object rather than computing a result; 7 with no instrument failure: 0 honoured, 0 violated, 7 with no contract checked; 4 where Syntology's instrument failed) · 1 unverified. Where Syntology's instrument failed: 2 where our contract draft was wrong; 2 where the contract check could not run; failures of Syntology's instrument, not of the code.

By repository: community (archive-listed): 12 samples from 1 repository, 11 ran (7 with no instrument failure; 4 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.

No sample is linked to this paper from 1 of its repositories that Syntology harvested for other papers:

Levels written between 2026-09-24 and 2026-09-25 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed. 1 of the 11 samples that ran has no recorded date.

jayleicn/singularity officialmentioned in papermentioned on GitHubpytorchSyntology: no sample linked to this paper (harvested for another paper). Syntology's graph links these samples to Test of Time: Instilling Video-Language Models with a Sense of Time (arXiv:2301.02074; that paper's own run record: 10 ran (of which 0 constructed an object rather than computing a result; 9 with no instrument failure: 0 honoured, 0 violated, 9 with no contract checked; 1 where Syntology's instrument failed) · 0 unverified); nothing checks that they implement this paper's method. report
jayleicn/ClipBERT mentioned on GitHubpytorchMITSyntology: 12 samples, 11 ran (4 instrument-failed) · 1 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.

A paper named beside a repository with no sample linked to this paper is shown with that paper's own run record, not this paper's: “ran” means executed on a synthesized input, not that the code is correct, and “where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Papers are listed in arXiv-id order, at most three per repository.

Code Syntology ran Syntology

12 samples harvested: 11 ran (of which 0 constructed an object rather than computing a result; 7 with no instrument failure: 0 honoured, 0 violated, 7 with no contract checked; 4 where Syntology's instrument failed) · 1 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Syntology's record for this page last changed ; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

2ran · our draft was wrong
2ran; the contract check could not run
7ran
1unverified

Licence: 2 of the 12 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 jayleicn/ClipBERT. “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”. “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.

jayleicn/ClipBERT community (archive-listed) 12 samples, 11 ran (4 instrument-failed) · 1 unverified
conv3x3 in 1,376 papers src/modeling/grid_feat.py ran MIT (permissive) · 17e89eef72e37efe · report fixture: signature in_planes:int/0/int, out_planes:int/0/int, stride:int/0/int, groups:int/0/int, dilation:int/0/int, init_identity:bool/0/bool · returned Conv2d · level recorded 2026-09-25
get_padding in 84 papers src/datasets/data_utils.py ran MIT (permissive) · 0cf387da46369b1a · report fixture: signature image:Tensor/4/float, max_w:int/0/int, max_h:int/0/int, pad_all:bool/0/bool · returned tuple · level recorded 2026-09-24
get_random_sample_indices src/modeling/modeling.py ran MIT (permissive) · f31be4ddaa3ec6c2 · report fixture: signature seq_len:int/0/int, num_samples:int/0/int, device:device/0/? · returned Tensor · level recorded 2026-09-25
get_start_end_idx in 8 papers src/datasets/decoder.py ran fingerprinted MIT (permissive) · 707f6cd75833c65d · report fixture: signature video_size:int/0/int, clip_size:int/0/int, clip_idx:int/0/int, num_clips:int/0/int · returned tuple · level recorded 2026-09-25
get_video_decoding_kwargs src/datasets/dataset_base.py ran MIT (permissive) · b6331823c359e9d8 · report fixture: signature container:VideoContainer/0/?, num_frames:int/0/int, target_fps:int/0/int, num_clips:NoneType/0/?, clip_idx:NoneType/0/?, sampling_strategy:str/0/str, safeguard_duration:bool/0/bool, video_max_pts:NoneType/0/? · returned dict · level recorded 2026-09-25
image_to_tensor in 14 papers src/datasets/data_utils.py ran; the contract check could not run (RAISES) MIT (permissive) · e9242c50098a6ea2 · report fixture: signature image:ndarray/3/float, keepdim:bool/0/bool · returned Tensor · level recorded 2026-09-24
instance_bce_with_logits in 7 papers src/modeling/modeling.py ran MIT (permissive) · 5a4d93d8be97ccd9 · report fixture: signature logits:Tensor/2/float, labels:Tensor/2/float, reduction:str/0/str · returned Tensor · level recorded 2026-09-25
mask_batch_text_tokens in 6 papers src/datasets/data_utils.py ran · our draft was wrong MIT (permissive) here; the stored copy is pointer only · f828f173e6c2fea2 · report fixture: signature inputs:Tensor/2/int, tokenizer:MockTokenizer/0/?, mlm_probability:float/0/float, is_train:bool/0/bool · returned tuple · level recorded, date not recorded
move_to_cuda in 22 papers src/datasets/dataloader.py ran; the contract check could not run (RAISES) MIT (permissive) · b68769dd52ad6c44 · report fixture: signature batch:dict/0/? · returned dict · level recorded 2026-09-24
parse_with_config in 12 papers src/configs/config.py ran · our draft was wrong MIT (permissive) here; the stored copy is pointer only · 3e81145412a40e36 · report fixture: signature parsed_args:Namespace/0/? · returned EasyDict · level recorded 2026-09-25
temporal_sampling in 8 papers src/datasets/decoder.py ran MIT (permissive) · f307b4793e8204b7 · report fixture: signature frames:list/1/MockVideoFrame, start_idx:int/0/int, end_idx:int/0/int, num_samples:int/0/int · returned list · level recorded 2026-09-25
pyav_decode_stream in 5 papers src/datasets/decoder.py unverified MIT (permissive) · b28b46ef7776c434 · report

Tasks

Action RecognitionFine-grained Action RecognitionLanguage ModelingLanguage ModellingQuestion AnsweringRetrievalText to Video RetrievalVideo Question AnsweringVideo RetrievalZero-Shot Video Retrieval

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Video Question Answering ActivityNet-QA Singularity-temporal Accuracy 44.1 #20 of 36 Archive leaderboard report
Video Question Answering ActivityNet-QA Singularity Accuracy 43.1 #22 of 36 Archive leaderboard report
Video Question Answering MSRVTT-MC Singularity-temporal Accuracy 93.7 #5 of 7 Archive leaderboard report
Video Question Answering MSRVTT-MC Singularity Accuracy 92.1 #7 of 7 Archive leaderboard report
Video Question Answering MSRVTT-QA Singularity-temporal Accuracy 43.9 #12 of 14 Archive leaderboard report
Video Question Answering MSRVTT-QA Singularity Accuracy 43.5 #13 of 14 Archive leaderboard report
Video Retrieval ActivityNet Singularity text-to-video R@1 47.1 #19 of 31 Archive leaderboard report
Video Retrieval ActivityNet Singularity text-to-video R@10 85.5 #19 of 31 Archive leaderboard report
Video Retrieval ActivityNet Singularity text-to-video R@5 75.5 #19 of 31 Archive leaderboard report
Video Retrieval DiDeMo Singularity text-to-video R@1 53.9 #18 of 40 Archive leaderboard report
Video Retrieval DiDeMo Singularity text-to-video R@10 86.9 #18 of 40 Archive leaderboard report
Video Retrieval DiDeMo Singularity text-to-video R@5 79.4 #18 of 40 Archive leaderboard report
Video Retrieval MSR-VTT-1kA Singularity text-to-video R@1 41.5 #37 of 63 Archive leaderboard report
Video Retrieval MSR-VTT-1kA Singularity text-to-video R@10 77 #37 of 63 Archive leaderboard report
Video Retrieval MSR-VTT-1kA Singularity text-to-video R@5 68.7 #37 of 63 Archive leaderboard report
Video Retrieval SSv2-label retrieval Singularity-temporal text-to-video R@1 47.4 #5 of 5 Archive leaderboard report
Video Retrieval SSv2-label retrieval Singularity-temporal text-to-video R@10 84 #5 of 5 Archive leaderboard report
Video Retrieval SSv2-label retrieval Singularity-temporal text-to-video R@5 75.9 #5 of 5 Archive leaderboard report
Video Retrieval SSv2-template retrieval Singularity-temporal text-to-video R@1 77.6 #5 of 5 Archive leaderboard report
Video Retrieval SSv2-template retrieval Singularity-temporal text-to-video R@10 98.9 #5 of 5 Archive leaderboard report
Video Retrieval SSv2-template retrieval Singularity-temporal text-to-video R@5 96 #5 of 5 Archive leaderboard report
Zero-Shot Video Retrieval ActivityNet Singularity-temporal-5M text-to-video R@1 30.8 #10 of 12 Archive leaderboard report
Zero-Shot Video Retrieval ActivityNet Singularity-temporal-5M text-to-video R@10 66.3 #10 of 12 Archive leaderboard report
Zero-Shot Video Retrieval ActivityNet Singularity-temporal-5M text-to-video R@5 55.9 #10 of 12 Archive leaderboard report
Zero-Shot Video Retrieval ActivityNet Singularity-temporal-17M text-to-video R@1 30.6 #12 of 12 Archive leaderboard report
Zero-Shot Video Retrieval ActivityNet Singularity-temporal-17M text-to-video R@10 66.9 #12 of 12 Archive leaderboard report
Zero-Shot Video Retrieval ActivityNet Singularity-temporal-17M text-to-video R@5 55.6 #12 of 12 Archive leaderboard report
Zero-Shot Video Retrieval DiDeMo Singularity-17M text-to-video R@1 37.1 #11 of 26 Archive leaderboard report
Zero-Shot Video Retrieval DiDeMo Singularity-17M text-to-video R@10 69.9 #11 of 26 Archive leaderboard report
Zero-Shot Video Retrieval DiDeMo Singularity-17M text-to-video R@5 61.7 #11 of 26 Archive leaderboard report
Zero-Shot Video Retrieval DiDeMo Singularity-5M text-to-video R@1 36.9 #12 of 26 Archive leaderboard report
Zero-Shot Video Retrieval DiDeMo Singularity-5M text-to-video R@10 69.3 #12 of 26 Archive leaderboard report
Zero-Shot Video Retrieval DiDeMo Singularity-5M text-to-video R@5 61.1 #12 of 26 Archive leaderboard report
Zero-Shot Video Retrieval MSR-VTT Singularity-17M text-to-video R@1 34.0 #20 of 41 Archive leaderboard report
Zero-Shot Video Retrieval MSR-VTT Singularity-17M text-to-video R@10 66.7 #20 of 41 Archive leaderboard report
Zero-Shot Video Retrieval MSR-VTT Singularity-17M text-to-video R@5 56.7 #20 of 41 Archive leaderboard report
Zero-Shot Video Retrieval MSR-VTT Singularity-5M text-to-video R@1 28.4 #24 of 41 Archive leaderboard report
Zero-Shot Video Retrieval MSR-VTT Singularity-5M text-to-video R@10 59.5 #24 of 41 Archive leaderboard report
Zero-Shot Video Retrieval MSR-VTT Singularity-5M text-to-video R@5 50.2 #24 of 41 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