{"url":"/sota/video-question-answering-on-activitynet-qa","task":{"name":"Video Question Answering","url":"/task/video-question-answering","note":null},"dataset":{"name":"ActivityNet-QA","url":"/dataset/activitynet-qa"},"category":"Computer Vision","categories":["Computer Vision","Reasoning"],"category_note":null,"description":null,"description_from":null,"source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","rank":"the archive's row order at snapshot; not re-ranked","rows_end_at":"2025-07-28","rows_withheld_as_spam":0,"metric_values":"the archive's strings, untouched"},"metrics":["Accuracy","Confidence score"],"metric_direction":{"note":"inferred from the metric name only (the archive records no direction); null = not inferred, chart draws points only","by_metric":{"Accuracy":"higher","Confidence score":"higher"}},"counts":{"rows":36,"rows_with_code":32,"rows_with_paper_page":36,"rows_dated":36,"rows_using_additional_data":10},"rows":[{"rank_in_archive_order":1,"model":"GPT-2 + CLIP-14 + CLIP-multilingual (Zero-Shot)","metrics":{"Accuracy":"61.2"},"uses_additional_data":false,"paper_date":"2022-10-20","paper":"/paper/composing-ensembles-of-pre-trained-models-via","paper_url":"https://arxiv.org/abs/2210.11522v1","paper_title":"Composing Ensembles of Pre-trained Models via Iterative Consensus","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":2,"model":"GPT-2 + CLIP-32 (Zero-Shot)","metrics":{"Accuracy":"58.4"},"uses_additional_data":false,"paper_date":"2022-10-20","paper":"/paper/composing-ensembles-of-pre-trained-models-via","paper_url":"https://arxiv.org/abs/2210.11522v1","paper_title":"Composing Ensembles of Pre-trained Models via Iterative Consensus","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":3,"model":"VideoCoCa","metrics":{"Accuracy":"56.1"},"uses_additional_data":true,"paper_date":"2022-12-09","paper":"/paper/video-text-modeling-with-zero-shot-transfer","paper_url":"https://arxiv.org/abs/2212.04979v3","paper_title":"VideoCoCa: Video-Text Modeling with Zero-Shot Transfer from Contrastive Captioners","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":4,"model":"Mirasol3B","metrics":{"Accuracy":"51.13"},"uses_additional_data":false,"paper_date":"2023-11-09","paper":"/paper/mirasol3b-a-multimodal-autoregressive-model","paper_url":"https://arxiv.org/abs/2311.05698v3","paper_title":"Mirasol3B: A Multimodal Autoregressive model for time-aligned and contextual modalities","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":5,"model":"VAST","metrics":{"Accuracy":"50.4"},"uses_additional_data":true,"paper_date":"2023-05-29","paper":"/paper/vast-a-vision-audio-subtitle-text-omni-1","paper_url":"https://arxiv.org/abs/2305.18500v2","paper_title":"VAST: A Vision-Audio-Subtitle-Text Omni-Modality Foundation Model and Dataset","code":"https://github.com/TXH-mercury/VALOR","n_code_links":2,"syntology":{"n_ran":15,"n_unverified":27,"n_samples":42,"n_pointer_only_licence":0}},{"rank_in_archive_order":6,"model":"COSA","metrics":{"Accuracy":"49.9"},"uses_additional_data":true,"paper_date":"2023-06-15","paper":"/paper/cosa-concatenated-sample-pretrained-vision","paper_url":"https://arxiv.org/abs/2306.09085v1","paper_title":"COSA: Concatenated Sample Pretrained Vision-Language Foundation Model","code":"https://github.com/txh-mercury/cosa","n_code_links":1,"syntology":null},{"rank_in_archive_order":7,"model":"MA-LMM","metrics":{"Accuracy":"49.8"},"uses_additional_data":false,"paper_date":"2024-04-08","paper":"/paper/ma-lmm-memory-augmented-large-multimodal","paper_url":"https://arxiv.org/abs/2404.05726v2","paper_title":"MA-LMM: Memory-Augmented Large Multimodal Model for Long-Term Video Understanding","code":"https://github.com/boheumd/MA-LMM","n_code_links":1,"syntology":{"n_ran":7,"n_unverified":2,"n_samples":9,"n_pointer_only_licence":0}},{"rank_in_archive_order":8,"model":"VideoChat2","metrics":{"Accuracy":"49.1","Confidence score":"3.3"},"uses_additional_data":false,"paper_date":"2023-11-28","paper":"/paper/mvbench-a-comprehensive-multi-modal-video","paper_url":"https://arxiv.org/abs/2311.17005v4","paper_title":"MVBench: A Comprehensive Multi-modal Video Understanding Benchmark","code":"https://github.com/opengvlab/ask-anything","n_code_links":3,"syntology":{"n_ran":7,"n_unverified":3,"n_samples":10,"n_pointer_only_licence":0}},{"rank_in_archive_order":9,"model":"VALOR","metrics":{"Accuracy":"48.6"},"uses_additional_data":true,"paper_date":"2023-04-17","paper":"/paper/valor-vision-audio-language-omni-perception","paper_url":"https://arxiv.org/abs/2304.08345v2","paper_title":"VALOR: Vision-Audio-Language Omni-Perception Pretraining Model and Dataset","code":"https://github.com/TXH-mercury/VALOR","n_code_links":1,"syntology":null},{"rank_in_archive_order":10,"model":"UMT-L (ViT-L/16)","metrics":{"Accuracy":"47.9"},"uses_additional_data":true,"paper_date":"2023-03-28","paper":"/paper/unmasked-teacher-towards-training-efficient","paper_url":"https://arxiv.org/abs/2303.16058v2","paper_title":"Unmasked Teacher: Towards Training-Efficient Video Foundation Models","code":"https://github.com/opengvlab/unmasked_teacher","n_code_links":1,"syntology":{"n_ran":5,"n_unverified":3,"n_samples":8,"n_pointer_only_licence":0}},{"rank_in_archive_order":11,"model":"LLaMA-VID-13B (2 Token)","metrics":{"Accuracy":"47.5","Confidence score":"3.3"},"uses_additional_data":false,"paper_date":"2023-11-28","paper":"/paper/llama-vid-an-image-is-worth-2-tokens-in-large","paper_url":"https://arxiv.org/abs/2311.17043v1","paper_title":"LLaMA-VID: An Image is Worth 2 Tokens in Large Language Models","code":"https://github.com/lastmile-ai/aiconfig","n_code_links":2,"syntology":{"n_ran":3,"n_unverified":1,"n_samples":4,"n_pointer_only_licence":0}},{"rank_in_archive_order":12,"model":"LLaMA-VID-7B (2 Token)","metrics":{"Accuracy":"47.4","Confidence score":"3.3"},"uses_additional_data":false,"paper_date":"2023-11-28","paper":"/paper/llama-vid-an-image-is-worth-2-tokens-in-large","paper_url":"https://arxiv.org/abs/2311.17043v1","paper_title":"LLaMA-VID: An Image is Worth 2 Tokens in Large Language Models","code":"https://github.com/lastmile-ai/aiconfig","n_code_links":2,"syntology":{"n_ran":3,"n_unverified":1,"n_samples":4,"n_pointer_only_licence":0}},{"rank_in_archive_order":13,"model":"Chat-UniVi-13B","metrics":{"Accuracy":"46.4","Confidence score":"3.3"},"uses_additional_data":false,"paper_date":"2023-11-14","paper":"/paper/chat-univi-unified-visual-representation","paper_url":"https://arxiv.org/abs/2311.08046v3","paper_title":"Chat-UniVi: Unified Visual Representation Empowers Large Language Models with Image and Video Understanding","code":"https://github.com/pku-yuangroup/chat-univi","n_code_links":4,"syntology":null},{"rank_in_archive_order":14,"model":"BT-Adapter (zero-shot)","metrics":{"Accuracy":"46.1","Confidence score":"3.6"},"uses_additional_data":false,"paper_date":"2023-09-27","paper":"/paper/one-for-all-video-conversation-is-feasible","paper_url":"https://arxiv.org/abs/2309.15785v2","paper_title":"BT-Adapter: Video Conversation is Feasible Without Video Instruction Tuning","code":"https://github.com/farewellthree/BT-Adapter","n_code_links":1,"syntology":{"n_ran":1,"n_unverified":3,"n_samples":4,"n_pointer_only_licence":0}},{"rank_in_archive_order":15,"model":"MovieChat","metrics":{"Accuracy":"45.7","Confidence score":"3.1"},"uses_additional_data":false,"paper_date":"2023-07-31","paper":"/paper/moviechat-from-dense-token-to-sparse-memory","paper_url":"https://arxiv.org/abs/2307.16449v4","paper_title":"MovieChat: From Dense Token to Sparse Memory for Long Video Understanding","code":"https://github.com/rese1f/MovieChat","n_code_links":1,"syntology":null},{"rank_in_archive_order":16,"model":"Video-LLaVA","metrics":{"Accuracy":"45.3","Confidence score":"3.3"},"uses_additional_data":false,"paper_date":"2023-11-16","paper":"/paper/video-llava-learning-united-visual-1","paper_url":"https://arxiv.org/abs/2311.10122v3","paper_title":"Video-LLaVA: Learning United Visual Representation by Alignment Before Projection","code":"https://github.com/PKU-YuanGroup/Video-LLaVA","n_code_links":6,"syntology":{"n_ran":4,"n_unverified":3,"n_samples":7,"n_pointer_only_licence":1}},{"rank_in_archive_order":17,"model":"TESTA (ViT-B/16)","metrics":{"Accuracy":"45"},"uses_additional_data":true,"paper_date":"2023-10-29","paper":"/paper/testa-temporal-spatial-token-aggregation-for","paper_url":"https://arxiv.org/abs/2310.19060v1","paper_title":"TESTA: Temporal-Spatial Token Aggregation for Long-form Video-Language Understanding","code":"https://github.com/renshuhuai-andy/testa","n_code_links":1,"syntology":{"n_ran":11,"n_unverified":4,"n_samples":15,"n_pointer_only_licence":0}},{"rank_in_archive_order":18,"model":"FrozenBiLM+","metrics":{"Accuracy":"44.8"},"uses_additional_data":false,"paper_date":"2023-08-18","paper":"/paper/open-vocabulary-video-question-answering-a","paper_url":"https://arxiv.org/abs/2308.09363v1","paper_title":"Open-vocabulary Video Question Answering: A New Benchmark for Evaluating the Generalizability of Video Question Answering Models","code":"https://github.com/mlvlab/ovqa","n_code_links":1,"syntology":{"n_ran":3,"n_unverified":5,"n_samples":8,"n_pointer_only_licence":8}},{"rank_in_archive_order":19,"model":"VindLU","metrics":{"Accuracy":"44.7"},"uses_additional_data":true,"paper_date":"2022-12-09","paper":"/paper/vindlu-a-recipe-for-effective-video-and","paper_url":"https://arxiv.org/abs/2212.05051v2","paper_title":"VindLU: A Recipe for Effective Video-and-Language Pretraining","code":"https://github.com/klauscc/vindlu","n_code_links":1,"syntology":null},{"rank_in_archive_order":20,"model":"Singularity-temporal","metrics":{"Accuracy":"44.1"},"uses_additional_data":true,"paper_date":"2022-06-07","paper":"/paper/revealing-single-frame-bias-for-video-and","paper_url":"https://arxiv.org/abs/2206.03428v1","paper_title":"Revealing Single Frame Bias for Video-and-Language Learning","code":"https://github.com/jayleicn/ClipBERT","n_code_links":2,"syntology":{"n_ran":9,"n_unverified":3,"n_samples":12,"n_pointer_only_licence":0}},{"rank_in_archive_order":21,"model":"FrozenBiLM","metrics":{"Accuracy":"43.2"},"uses_additional_data":true,"paper_date":"2022-06-16","paper":"/paper/zero-shot-video-question-answering-via-frozen","paper_url":"https://arxiv.org/abs/2206.08155v2","paper_title":"Zero-Shot Video Question Answering via Frozen Bidirectional Language Models","code":"https://github.com/antoyang/FrozenBiLM","n_code_links":3,"syntology":{"n_ran":14,"n_unverified":20,"n_samples":34,"n_pointer_only_licence":1}},{"rank_in_archive_order":22,"model":"Singularity","metrics":{"Accuracy":"43.1"},"uses_additional_data":true,"paper_date":"2022-06-07","paper":"/paper/revealing-single-frame-bias-for-video-and","paper_url":"https://arxiv.org/abs/2206.03428v1","paper_title":"Revealing Single Frame Bias for Video-and-Language Learning","code":"https://github.com/jayleicn/ClipBERT","n_code_links":2,"syntology":{"n_ran":9,"n_unverified":3,"n_samples":12,"n_pointer_only_licence":0}},{"rank_in_archive_order":23,"model":"Text + Text (no Multimodal Pretext Training)","metrics":{"Accuracy":"41.4"},"uses_additional_data":false,"paper_date":"2022-06-05","paper":"/paper/towards-fast-adaptation-of-pretrained","paper_url":"https://arxiv.org/abs/2206.02082v4","paper_title":"Towards Fast Adaptation of Pretrained Contrastive Models for Multi-channel Video-Language Retrieval","code":"https://github.com/xudonglinthu/upgradable-multimodal-intelligence","n_code_links":1,"syntology":{"n_ran":1,"n_unverified":0,"n_samples":1,"n_pointer_only_licence":0}},{"rank_in_archive_order":24,"model":"All-in-one+","metrics":{"Accuracy":"40.0"},"uses_additional_data":false,"paper_date":"2023-08-18","paper":"/paper/open-vocabulary-video-question-answering-a","paper_url":"https://arxiv.org/abs/2308.09363v1","paper_title":"Open-vocabulary Video Question Answering: A New Benchmark for Evaluating the Generalizability of Video Question Answering Models","code":"https://github.com/mlvlab/ovqa","n_code_links":1,"syntology":{"n_ran":3,"n_unverified":5,"n_samples":8,"n_pointer_only_licence":8}},{"rank_in_archive_order":25,"model":"VIOLET+","metrics":{"Accuracy":"39.7"},"uses_additional_data":false,"paper_date":"2023-08-18","paper":"/paper/open-vocabulary-video-question-answering-a","paper_url":"https://arxiv.org/abs/2308.09363v1","paper_title":"Open-vocabulary Video Question Answering: A New Benchmark for Evaluating the Generalizability of Video Question Answering Models","code":"https://github.com/mlvlab/ovqa","n_code_links":1,"syntology":{"n_ran":3,"n_unverified":5,"n_samples":8,"n_pointer_only_licence":8}},{"rank_in_archive_order":26,"model":"Just Ask (fine-tune)","metrics":{"Accuracy":"38.9"},"uses_additional_data":false,"paper_date":"2020-12-01","paper":"/paper/just-ask-learning-to-answer-questions-from","paper_url":"https://arxiv.org/abs/2012.00451v3","paper_title":"Just Ask: Learning to Answer Questions from Millions of Narrated Videos","code":"https://github.com/antoyang/just-ask","n_code_links":1,"syntology":null},{"rank_in_archive_order":27,"model":"LocVLM-Vid-B+","metrics":{"Accuracy":"38.2"},"uses_additional_data":false,"paper_date":"2024-04-11","paper":"/paper/learning-to-localize-objects-improves-spatial","paper_url":"https://arxiv.org/abs/2404.07449v1","paper_title":"Learning to Localize Objects Improves Spatial Reasoning in Visual-LLMs","code":"https://github.com/kahnchana/locvlm","n_code_links":1,"syntology":null},{"rank_in_archive_order":28,"model":"LocVLM-Vid-B","metrics":{"Accuracy":"37.4"},"uses_additional_data":false,"paper_date":"2024-04-11","paper":"/paper/learning-to-localize-objects-improves-spatial","paper_url":"https://arxiv.org/abs/2404.07449v1","paper_title":"Learning to Localize Objects Improves Spatial Reasoning in Visual-LLMs","code":"https://github.com/kahnchana/locvlm","n_code_links":1,"syntology":null},{"rank_in_archive_order":29,"model":"Video-ChatGPT","metrics":{"Accuracy":"35.2","Confidence score":"2.7"},"uses_additional_data":false,"paper_date":"2023-06-08","paper":"/paper/video-chatgpt-towards-detailed-video","paper_url":"https://arxiv.org/abs/2306.05424v2","paper_title":"Video-ChatGPT: Towards Detailed Video Understanding via Large Vision and Language Models","code":"https://github.com/mbzuai-oryx/video-chatgpt","n_code_links":2,"syntology":null},{"rank_in_archive_order":30,"model":"LLaMA Adapter V2","metrics":{"Accuracy":"34.2","Confidence score":"2.7"},"uses_additional_data":false,"paper_date":"2023-04-28","paper":"/paper/llama-adapter-v2-parameter-efficient-visual","paper_url":"https://arxiv.org/abs/2304.15010v1","paper_title":"LLaMA-Adapter V2: Parameter-Efficient Visual Instruction Model","code":"https://github.com/opengvlab/llama-adapter","n_code_links":3,"syntology":{"n_ran":0,"n_unverified":1,"n_samples":1,"n_pointer_only_licence":1}},{"rank_in_archive_order":31,"model":"E-SA","metrics":{"Accuracy":"31.8"},"uses_additional_data":false,"paper_date":"2019-06-06","paper":"/paper/activitynet-qa-a-dataset-for-understanding","paper_url":"https://arxiv.org/abs/1906.02467v1","paper_title":"ActivityNet-QA: A Dataset for Understanding Complex Web Videos via Question Answering","code":"https://github.com/MILVLG/activitynet-qa","n_code_links":1,"syntology":null},{"rank_in_archive_order":32,"model":"E-MN","metrics":{"Accuracy":"27.1"},"uses_additional_data":false,"paper_date":"2019-06-06","paper":"/paper/activitynet-qa-a-dataset-for-understanding","paper_url":"https://arxiv.org/abs/1906.02467v1","paper_title":"ActivityNet-QA: A Dataset for Understanding Complex Web Videos via Question Answering","code":"https://github.com/MILVLG/activitynet-qa","n_code_links":1,"syntology":null},{"rank_in_archive_order":33,"model":"Video Chat","metrics":{"Accuracy":"26.5","Confidence score":"2.2"},"uses_additional_data":false,"paper_date":"2023-05-10","paper":"/paper/videochat-chat-centric-video-understanding","paper_url":"https://arxiv.org/abs/2305.06355v2","paper_title":"VideoChat: Chat-Centric Video Understanding","code":"https://github.com/opengvlab/ask-anything","n_code_links":1,"syntology":null},{"rank_in_archive_order":34,"model":"FrozenBiLM (0-shot)","metrics":{"Accuracy":"25.9"},"uses_additional_data":false,"paper_date":"2022-06-16","paper":"/paper/zero-shot-video-question-answering-via-frozen","paper_url":"https://arxiv.org/abs/2206.08155v2","paper_title":"Zero-Shot Video Question Answering via Frozen Bidirectional Language Models","code":"https://github.com/antoyang/FrozenBiLM","n_code_links":3,"syntology":{"n_ran":14,"n_unverified":20,"n_samples":34,"n_pointer_only_licence":1}},{"rank_in_archive_order":35,"model":"E-VQA","metrics":{"Accuracy":"25.1"},"uses_additional_data":false,"paper_date":"2019-06-06","paper":"/paper/activitynet-qa-a-dataset-for-understanding","paper_url":"https://arxiv.org/abs/1906.02467v1","paper_title":"ActivityNet-QA: A Dataset for Understanding Complex Web Videos via Question Answering","code":"https://github.com/MILVLG/activitynet-qa","n_code_links":1,"syntology":null},{"rank_in_archive_order":36,"model":"Just Ask (0-shot)","metrics":{"Accuracy":"12.2"},"uses_additional_data":false,"paper_date":"2020-12-01","paper":"/paper/just-ask-learning-to-answer-questions-from","paper_url":"https://arxiv.org/abs/2012.00451v3","paper_title":"Just Ask: Learning to Answer Questions from Millions of Narrated Videos","code":"https://github.com/antoyang/just-ask","n_code_links":1,"syntology":null}],"since_archive":{"claim":"Results that newer papers report for their own method, placed here by Syntology. A model pointed at the cell in the paper's own table; the number was read from that cell and checked against this leaderboard's metric, dataset, split and scale; an independent check that saw this leaderboard's other rows and every other leaderboard on the same dataset accepted it. Not reviewed by the paper's authors or by the archive's editors, and not ranked against the archive rows.","extraction_file_present":true,"measurement":{"test_papers":883,"papers_with_output":881,"judged_true":108,"judged":110,"wilson95_lower":0.9361,"measured_on":"2026-09-24","frozen_commit":"0e3de0df94"},"measurement_note":"blind adjudication of accepted entries on a held-out split of archive papers, rules frozen before the test","coverage":{"sentence":"Syntology has checked 6,885 of the 9,623 papers on this site that are newer than the archive; results from the others appear after they are checked.","complete":false,"papers_newer_than_archive":9623,"papers_checked":6885,"papers_extracted_not_yet_verified":0,"boards_without_verdict":2,"papers_not_yet_extracted":2737},"order":"newest first by month (arXiv date, else the arXiv-id month), then arXiv id descending","columns":[],"entries":[]},"syntology":{"read_at":"2026-09-25T09:33:49+00:00","claim":"Per row: N of M harvested code samples from that row's paper executed on a synthesized fixture; the other M-N are unverified. Not a reproduction of the row's number; not a correctness claim. n_pointer_only_licence counts samples the site points at rather than redistributes (a licence axis, independent of ran/unverified).","rows_with_graph_line":18,"rows_with_any_sample_ran":17,"distinct_papers_with_graph_line":13,"distinct_papers_with_any_sample_ran":12,"samples_over_distinct_papers":{"n_ran":80,"n_unverified":75,"n_samples":155,"n_pointer_only_licence":11,"note":"each paper (arXiv id) counted once, however many rows it is behind; this is the page-level figure"},"samples_row_weighted":{"n_ran":112,"n_unverified":109,"n_samples":221,"n_pointer_only_licence":28,"note":"row-weighted: a paper behind several rows is counted once per row; inflated relative to samples_over_distinct_papers by design, kept for readers summing the per-row syntology blocks"}}}