Retrieving-to-Answer: Zero-Shot Video Question Answering with Frozen Large Language Models
The DOI matches its Crossref record; 22 of the 22 references it cites were found in a registry or an index.
Abstract
Video Question Answering (VideoQA) has been significantly advanced from the scaling of recent Large Language Models (LLMs). The key idea is to convert the visual information into the language feature space so that the capacity of LLMs can be fully exploited. Existing VideoQA methods typically take two paradigms: (1) learning cross-modal alignment, and (2) using an off-the-shelf captioning model to describe the visual data. However, the first design needs costly training on many extra multi-modal data, whilst the second is further limited by limited domain generalization. To address these limitations, a simple yet effective Retrieving-to-Answer (R2A) framework is proposed. Given an input video, R2A first retrieves a set of semantically similar texts from a generic text corpus using a pre-trained multi-modal model (e.g., CLIP). With both the question and the retrieved texts, a LLM (e.g., DeBERTa) can be directly used to yield a desired answer. Without the need for cross-modal fine-tuning, R2A allows for all the key components (e.g., LLM, retrieval model, and text corpus) to plug-and-play. Extensive experiments on several VideoQA benchmarks show that despite with 1.3B parameters and no fine-tuning, our R2A can outperform the 61× larger Flamingo-80B model [1] even additionally trained on nearly 2.1B multi-modal data.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“R2A [318] retrieves semantically similar texts by CLIP, and prompts LLM with both the query and the retrieved texts.”
From Retrieval-Augmented Generation for AI-Generated Content: A Survey (2024) · cites this paper as [318]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2019) · Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers) · doi:10.18653/v1/n19-1423 · Cited by 54,325
Language Models are Few-Shot Learners
Brown, T. B., Mann, B., Ryder, N., Subbiah, M., Kaplan, J., Dhariwal, P., Neelakantan, A., Shyam, P., Sastry, G., Askell, A., et al. (2020) · arXiv · doi:10.48550/arxiv.2005.14165 · Cited by 23,372
Training language models to follow instructions with human feedback
Ouyang, L., Wu, J., Jiang, X., Almeida, D., Wainwright, C. L., Mishkin, P., Zhang, C., Agarwal, S., Slama, K., Ray, A., et al. (2022) · arXiv · doi:10.48550/arxiv.2203.02155 · Cited by 4,208
Learning to Prompt for Vision-Language Models
Zhou, K., Yang, J., Loy, C. C., & Liu, Z. (2022) · International Journal of Computer Vision · doi:10.1007/s11263-022-01653-1 · Cited by 3,304 · cited 2×
“We hypothesize that this is due to the prior distributions of some retrieval datasets, which may serve as task-specific prompts [69] for certain target datasets.”
BackgroundTraining Language Models to Follow Instructions with Human Feedback
Ouyang, L., Wu, J., Jiang, X., Almeida, D., Wainwright, C., Mishkin, P., Zhang, C., Agarwal, S., Slama, K., Ray, A., et al. (2022) · Advances in Neural Information Processing Systems 35 · doi:10.52202/068431-2011 · Cited by 3,117
Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks
Lewis, P., Perez, E., Piktus, A., Petroni, F., Karpukhin, V., Goyal, N., Küttler, H., Lewis, M., Yih, W.-T., Rocktäschel, T., et al. (2020) · arXiv · doi:10.48550/arxiv.2005.11401 · Cited by 3,101 · cited 2×
“Retrieval-augmented language models have gained attention in NLP [4, 14, 26, 27, 27, 54].”
BackgroundOn the Opportunities and Risks of Foundation Models
Bommasani, R., Hudson, D. A., Adeli, E., Altman, R., Arora, S., von Arx, S., Bernstein, M. S., Bohg, J., Bosselut, A., Brunskill, E., et al. (2021) · arXiv · doi:10.48550/arxiv.2108.07258 · Cited by 2,676
Scaling Up Visual and Vision-Language Representation Learning With Noisy Text Supervision
Jia, C., Yang, Y., Xia, Y., Chen, Y.-T., Parekh, Z., Pham, H., Le, Q. V., Sung, Y., Li, Z., & Duerig, T. (2021) · arXiv · doi:10.48550/arxiv.2102.05918 · Cited by 1,977
Conceptual Captions: A Cleaned, Hypernymed, Image Alt-text Dataset For Automatic Image Captioning
Sharma, P., Ding, N., Goodman, S., & Soricut, R. (2018) · Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · doi:10.18653/v1/p18-1238 · Cited by 1,879 · cited 3×
“…various multi-modal datasets, including two datasets scraped from web: (1) WebVid [2] that contains approximately 10M video-text pairs, (2) CC3M [47] and CC12M [6] that consists of 3M and 12M image-text pairs, as well as a human-annotated dataset: (3) COCO Caption dataset [30] with 1.5M…”
BackgroundMethodLi, J., Li, D., Xiong, C., & Hoi, S. (2022) · arXiv · doi:10.48550/arxiv.2201.12086 · Cited by 1,741
Research works printed but not linked: 43
- 10.1109/CVPR46437.2021.00356 · DOI · Paper page
- 10.1109/CVPR52688.2022.01750 · DOI · Paper page
- 10.1609/aaai.v35i2.16203 · DOI
- 10.1109/IJCNN52387.2021.9533961 · DOI
- Guo, Accelerating large-scale inference with anisotropic vector quantization, 3887
- Guo, Accelerating large-scale inference with anisotropic vector quantization, 3887
- Hu, 2023, Open-domain visual entity recognition: Towards recognizing millions of wikipedia entities, 10.1109/ICCV51070.2023.01108 · DOI
- 10.1109/ICCV.2013.457 · DOI
- 10.18653/v1/2021.eacl-main.74 · DOI · Paper page
- 10.1109/CVPR.2017.149 · DOI · Paper page
- 10.1109/cvpr52729.2023.00638 · DOI · Paper page
- 10.1609/aaai.v35i14.17556 · DOI
- 10.1109/CVPR46437.2021.00725 · DOI · Paper page
- Lewis, 2020, Pre-training via paraphrasing, Advances in Neural Information Processing Systems, 33, 18470
- 10.1109/CVPR52688.2022.00294 · DOI
- 10.1007/978-3-319-10602-1_48 · DOI · Paper page
- 10.1109/CVPR46437.2021.00693 · DOI
- 10.1109/CVPR.2015.7298748 · DOI · Paper page
- 10.1109/CVPR52688.2022.00683 · DOI · Paper page
- 10.18653/v1/2023.eacl-main.185 · DOI
- 10.18653/v1/2022.findings-emnlp.67 · DOI · Paper page
- 10.1007/978-3-031-19781-9_24 · DOI
- 10.1109/CVPR46437.2021.01527 · DOI
- 10.1109/CVPR.2018.00918 · DOI · Paper page
- Radford, Learning transferable visual models from natural language supervision, 139, 8748
- 10.1109/cvpr52729.2023.00278 · DOI · Paper page
- 10.18653/v1/2021.naacl-main.196 · DOI
- 10.1145/3549555.3549585 · DOI · Paper page
- 10.1109/CVPR52688.2022.01743 · DOI · Paper page
- 10.1109/CVPR46437.2021.01660 · DOI
- 10.1109/ICCV.2019.00756 · DOI · Paper page
- 10.1145/2812802 · DOI · Paper page
- 10.1109/CVPR46437.2021.00965 · DOI · Paper page
- 10.1609/aaai.v36i3.20184 · DOI
- 10.1145/3123266.3123427 · DOI · Paper page
- 10.1145/3123266.3123427 · DOI · Paper page
- 10.18653/v1/2021.emnlp-main.544 · DOI · Paper page
- 10.1109/ICCV48922.2021.00171 · DOI · Paper page
- 10.1609/aaai.v36i3.20215 · DOI · Paper page
- 10.1109/CVPR52688.2022.01589 · DOI · Paper page
- Zellers, 2021, Merlot: Multimodal neural script knowledge models, Advances in Neural Information Processing Systems, 34
- 10.1109/CVPR46437.2021.00971 · DOI
- 10.18653/v1/2022.emnlp-main.432 · DOI
Cited by
Foundation Models Defining a New Era in Vision: A Survey and Outlook
Awais, M., Naseer, M., Khan, S., Anwer, R. M., Cholakkal, H., Shah, M., Yang, M.-H., & Khan, F. S. (2025) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2024.3506283 · Cited by 304
“Strategies tomitigatehallucinationincludeprovidingexplicitinstructions (orso-called system commands ), the chain of thought prompting [175], [176], [177], self-consistency (voting) [178], [179] and use of knowledge bases for retrieval augmented generation [180], [181], [182].”
Retrieval-Augmented Generation for AI-Generated Content: A Survey
Zhao, P., Zhang, H., Yu, Q., Wang, Z., Geng, Y., Fu, F., Yang, L., Zhang, W., Jiang, J., & Cui, B. (2024) · arXiv · doi:10.48550/arxiv.2402.19473 · Cited by 240
“The research advancements in prompt compression [202] and long-context support [374] have partially mitigated these challenges, albeit with a slight trade-off in accuracy or costs.”
MLVU: Benchmarking Multi-task Long Video Understanding
Zhou, J., Shu, Y., Zhao, B., Wu, B., Liang, Z., Xiao, S., Qin, M., Yang, X., Xiong, Y., Zhang, B., et al. (2025) · 2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr52734.2025.01278 · Cited by 44
Metadata
- Authors
- Pan, J., Lin, Z., Ge, Y., Zhu, X., Zhang, R., Wang, Y., Qiao, Y., & Li, H.
- Affiliations
- The Chinese University of Hong Kong,Multimedia Lab
- University of Hong Kong
- University of Surrey
- Shanghai AI Laboratory
- Venue
- 2023 IEEE/CVF International Conference on Computer Vision Workshops (ICCVW)
- Publisher
- IEEE
- Published
- 2023-10-02
- Pages
- 272-283
- Cited by
- 27
- References
- 22
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Retrieving-to-Answer: Zero-Shot Video Question Answering with Frozen Large Language Models. https://app.cochat.ai/papers/10.1109/iccvw60793.2023.00035
- The paper
- Pan, J., Lin, Z., Ge, Y., Zhu, X., Zhang, R., Wang, Y., Qiao, Y., & Li, H. (2023). Retrieving-to-Answer: Zero-Shot Video Question Answering with Frozen Large Language Models. 2023 IEEE/CVF International Conference on Computer Vision Workshops (ICCVW). https://doi.org/10.1109/iccvw60793.2023.00035
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.