{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,9]],"date-time":"2026-07-09T03:22:49Z","timestamp":1783567369532,"version":"3.55.0"},"reference-count":78,"publisher":"IEEE","license":[{"start":{"date-parts":[[2025,10,19]],"date-time":"2025-10-19T00:00:00Z","timestamp":1760832000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2025,10,19]],"date-time":"2025-10-19T00:00:00Z","timestamp":1760832000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2025,10,19]]},"DOI":"10.1109\/iccv51701.2025.01969","type":"proceedings-article","created":{"date-parts":[[2026,4,29]],"date-time":"2026-04-29T19:45:49Z","timestamp":1777491949000},"page":"21197-21208","source":"Crossref","is-referenced-by-count":4,"title":["VAMBA: Understanding Hour-Long Videos with Hybrid Mamba-Transformers"],"prefix":"10.1109","author":[{"given":"Weiming","family":"Ren","sequence":"first","affiliation":[{"name":"University of Waterloo"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Wentao","family":"Ma","sequence":"additional","affiliation":[{"name":"University of Toronto"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Huan","family":"Yang","sequence":"additional","affiliation":[{"name":"Kuaishou Technology"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Cong","family":"Wei","sequence":"additional","affiliation":[{"name":"University of Waterloo"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Ge","family":"Zhang","sequence":"additional","affiliation":[{"name":"University of Waterloo"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Wenhu","family":"Chen","sequence":"additional","affiliation":[{"name":"University of Waterloo"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"263","reference":[{"key":"ref1","author":"Abouelenin","year":"2025","journal-title":"Phi-4-mini technical report: Compact yet powerful multimodal language models via mixture-of-loras"},{"key":"ref2","article-title":"Gpt-4 technical report","author":"Achiam","year":"2023","journal-title":"arXiv preprint"},{"key":"ref3","doi-asserted-by":"publisher","DOI":"10.52202\/068431-1723"},{"key":"ref4","article-title":"Openflamingo: An opensource framework for training large autoregressive visionlanguage models","author":"Awadalla","year":"2023","journal-title":"arXiv preprint"},{"key":"ref5","article-title":"Qwen technical report","author":"Bai","year":"2023","journal-title":"arXiv preprint"},{"key":"ref6","article-title":"Token merging: Your vit but faster","author":"Bolya","year":"2023","journal-title":"ICLR"},{"key":"ref7","article-title":"A short note on the kinetics-700 human action dataset","author":"Carreira","journal-title":"arXiv preprint"},{"key":"ref8","doi-asserted-by":"publisher","DOI":"10.52202\/079017-1684"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.1109\/cvpr46437.2021.00356"},{"key":"ref10","article-title":"Allava: Harnessing gpt4v-synthesized data for lite vision-language models","author":"Chen","year":"2024","journal-title":"arXiv preprint"},{"key":"ref11","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-031-72643-9_22"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.52202\/079017-0614"},{"key":"ref13","article-title":"Training deep nets with sublinear memory cost","author":"Chen","year":"2016","journal-title":"arXiv preprint"},{"key":"ref14","article-title":"Microsoft coco captions: Data collection and evaluation server","author":"Chen","year":"2015","journal-title":"arXiv preprint"},{"key":"ref15","doi-asserted-by":"publisher","DOI":"10.1109\/cvpr52733.2024.02283"},{"key":"ref16","article-title":"Flashattention-2: Faster attention with better parallelism and work partitioning","author":"Dao","year":"2023","journal-title":"arXiv preprint"},{"key":"ref17","article-title":"Transformers are ssms: Generalized models and efficient algorithms through structured state space duality","author":"Dao","year":"2024","journal-title":"arXiv preprint"},{"key":"ref18","doi-asserted-by":"publisher","DOI":"10.52202\/075280-0106"},{"key":"ref19","article-title":"The llama 3 herd of models","author":"Dubey","year":"2024","journal-title":"arXiv preprint"},{"key":"ref20","article-title":"Video-ccam: Enhancing video-language understanding with causal cross-attention masks for short and long videos","author":"Fei","year":"2024","journal-title":"arXiv preprint"},{"key":"ref21","article-title":"Video-mme: The first-ever comprehensive evaluation benchmark of multi-modal 11 ms in video analysis","author":"Fu","year":"2024","journal-title":"arXiv preprint"},{"key":"ref22","article-title":"Zamba: A compact 7b ssm hybrid model","author":"Glorioso","year":"2024","journal-title":"arXiv preprint"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52688.2022.01842"},{"key":"ref24","article-title":"Mamba: Linear-time sequence modeling with selective state spaces","author":"Gu","year":"2023","journal-title":"arXiv preprint"},{"key":"ref25","first-page":"1474","article-title":"Hippo: Recurrent memory with optimal polynomial projections","volume":"33","author":"Gu","year":"2020","journal-title":"Advances in neural information processing systems"},{"key":"ref26","article-title":"Efficiently modeling long sequences with structured state spaces","author":"Gu","year":"2021","journal-title":"arXiv preprint"},{"key":"ref27","article-title":"Efficient multimodal learning from data-centric perspective","author":"He","year":"2024","journal-title":"arXiv preprint"},{"key":"ref28","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52734.2025.02709"},{"key":"ref29","article-title":"Mistral 7b","author":"Jiang","year":"2023","journal-title":"arXiv preprint"},{"key":"ref30","article-title":"Mantis: Interleaved multi-image instruction tuning","author":"Jiang","year":"2024","journal-title":"arXiv preprint"},{"key":"ref31","doi-asserted-by":"publisher","DOI":"10.1109\/ICCVW69036.2025.00614"},{"key":"ref32","doi-asserted-by":"publisher","DOI":"10.1007\/s11263-016-0981-7"},{"key":"ref33","doi-asserted-by":"publisher","DOI":"10.52202\/075280-3138"},{"key":"ref34","article-title":"What matters when building vision-language models?","author":"Lauren\u00e7on","year":"2024","journal-title":"arXiv preprint"},{"key":"ref35","article-title":"Llava-onevision: Easy visual task transfer","author":"Li","year":"2024","journal-title":"arXiv preprint"},{"key":"ref36","article-title":"Llava-next-interleave: Tackling multi-image, video, and 3d in large multimodal models","author":"Li","year":"2024","journal-title":"arXiv preprint"},{"key":"ref37","first-page":"1973019742","article-title":"Blip-2: Bootstrapping language-image pre-training with frozen image encoders and large language models","volume-title":"In International conference on machine learning","author":"Li"},{"key":"ref38","doi-asserted-by":"publisher","DOI":"10.1007\/s11432-024-4321-9"},{"key":"ref39","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-031-73347-5_14"},{"key":"ref40","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52733.2024.02095"},{"key":"ref41","article-title":"Jamba: A hybrid transformer-mamba language model","author":"Lieber","year":"2024","journal-title":"arXiv preprint"},{"key":"ref42","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2024.emnlp-main.342"},{"key":"ref43","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52733.2024.02484"},{"key":"ref44","author":"Liu","year":"2024","journal-title":"Llava-next"},{"key":"ref45","article-title":"Visual instruction tuning","volume":"36","author":"Liu","year":"2024","journal-title":"Advances in neural information processing systems"},{"key":"ref46","article-title":"Kangaroo: A powerful video-language model supporting long-context video input","author":"Liu","year":"2024","journal-title":"arXiv preprint"},{"key":"ref47","article-title":"Vmamba: Visual state space model 2024","author":"Liu","year":"2024","journal-title":"arXiv preprint"},{"key":"ref48","article-title":"Oryx mllm: On-demand spatial-temporal understanding at arbitrary resolution","author":"Liu","year":"2024","journal-title":"arXiv preprint"},{"key":"ref49","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2024.acl-long.679"},{"key":"ref50","volume-title":"OpenAI. Chatgpt","year":"2023"},{"key":"ref51","volume-title":"OpenAI. Gpt-4o","year":"2024"},{"key":"ref52","doi-asserted-by":"publisher","DOI":"10.1109\/SC41405.2020.00024"},{"key":"ref53","article-title":"Vista: Enhancing long-duration and high-resolution video understanding by video spatiotemporal augmentation","author":"Ren","year":"2024","journal-title":"arXiv preprint"},{"key":"ref54","article-title":"Laion-400m: Open dataset of clip-filtered 400 million image-text pairs","author":"Schuhmann","year":"2021","journal-title":"arXiv preprint"},{"key":"ref55","article-title":"Longvu: Spatiotemporal adaptive compression for long video-language understanding","author":"Shen","year":"2024","journal-title":"arXiv preprint"},{"key":"ref56","doi-asserted-by":"publisher","DOI":"10.1109\/cvpr52734.2025.02436"},{"key":"ref57","article-title":"From pixels to prose: A large dataset of dense image captions","author":"Singla","year":"2024","journal-title":"arXiv preprint"},{"key":"ref58","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52733.2024.01725"},{"key":"ref59","article-title":"Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context","author":"Team","year":"2024","journal-title":"arXiv preprint"},{"key":"ref60","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v39i7.32798"},{"key":"ref61","article-title":"Llama: Open and efficient foundation language models","author":"Touvron","year":"2023","journal-title":"arXiv preprint"},{"key":"ref62","article-title":"Tarsier: Recipes for training and evaluating large video description models","author":"Wang","year":"2024","journal-title":"arXiv preprint"},{"key":"ref63","article-title":"Qwen2-vl: Enhancing vision-language model\u2019s perception of the world at any resolution","author":"Wang","year":"2024","journal-title":"arXiv preprint"},{"key":"ref64","article-title":"Lvbench: An extreme long video understanding benchmark","author":"Wang","year":"2024","journal-title":"arXiv preprint"},{"key":"ref65","article-title":"Longllava: Scaling multi-modal 11 ms to 1000 images efficiently via a hybrid architecture","author":"Wang","year":"2024","journal-title":"arXiv preprint"},{"key":"ref66","article-title":"Internvideo2. 5: Empowering video mllms with long and rich context modeling","author":"Wang","year":"2025","journal-title":"arXiv preprint"},{"key":"ref67","first-page":"28828","article-title":"Longvideobench: A benchmark for long-context interleaved video-language understanding","volume":"37","author":"Wu","year":"2025","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref68","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR46437.2021.00965"},{"key":"ref69","article-title":"mplugowl3: Towards long image-sequence understanding in multimodal large language models","volume-title":"The Thirteenth International Conference on Learning Representations","author":"Ye"},{"key":"ref70","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2024.acl-long.142"},{"key":"ref71","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV51070.2023.01100"},{"key":"ref72","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2023.emnlp-demo.49"},{"key":"ref73","article-title":"Long context transfer from language to vision","author":"Zhang","year":"2024","journal-title":"arXiv preprint"},{"key":"ref74","article-title":"Llava-mini: Efficient image and video large multimodal models with one vision token","author":"Zhang","year":"2025","journal-title":"arXiv preprint"},{"key":"ref75","article-title":"Video instruction tuning with synthetic data","author":"Zhang","year":"2024","journal-title":"arXiv preprint"},{"key":"ref76","article-title":"Mlvu: A comprehensive benchmark for multi-task long video understanding","author":"Zhou","year":"2024","journal-title":"arXiv preprint"},{"key":"ref77","article-title":"Vision mamba: Efficient visual representation learning with bidirectional state space model","author":"Zhu","year":"2024","journal-title":"arXiv preprint"},{"key":"ref78","article-title":"Falcon mamba: The first competitive attention-free 7b language model","author":"Zuo","year":"2024","journal-title":"arXiv preprint"}],"event":{"name":"2025 IEEE\/CVF International Conference on Computer Vision (ICCV)","location":"Honolulu, HI, USA","start":{"date-parts":[[2025,10,19]]},"end":{"date-parts":[[2025,10,25]]}},"container-title":["2025 IEEE\/CVF International Conference on Computer Vision (ICCV)"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/11443115\/11443287\/11445594.pdf?arnumber=11445594","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,5,1]],"date-time":"2026-05-01T05:17:38Z","timestamp":1777612658000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11445594\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,10,19]]},"references-count":78,"URL":"https:\/\/doi.org\/10.1109\/iccv51701.2025.01969","relation":{},"subject":[],"published":{"date-parts":[[2025,10,19]]}}}