{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,8]],"date-time":"2026-07-08T06:06:53Z","timestamp":1783490813581,"version":"3.55.0"},"reference-count":87,"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"}],"funder":[{"DOI":"10.13039\/501100004183","name":"IITP","doi-asserted-by":"publisher","award":["RS-2024-00457882"],"award-info":[{"award-number":["RS-2024-00457882"]}],"id":[{"id":"10.13039\/501100004183","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2025,10,19]]},"DOI":"10.1109\/iccv51701.2025.01579","type":"proceedings-article","created":{"date-parts":[[2026,4,29]],"date-time":"2026-04-29T19:45:49Z","timestamp":1777491949000},"page":"17001-17012","source":"Crossref","is-referenced-by-count":1,"title":["MetaMorph: Multimodal Understanding and Generation via Instruction Tuning"],"prefix":"10.1109","author":[{"given":"Shengbang","family":"Tong","sequence":"first","affiliation":[{"name":"FAIR, Meta"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"David","family":"Fan","sequence":"additional","affiliation":[{"name":"FAIR, Meta"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jiachen","family":"Zhu","sequence":"additional","affiliation":[{"name":"FAIR, Meta"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yunyang","family":"Xiong","sequence":"additional","affiliation":[{"name":"Meta Reality Labs"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Xinlei","family":"Chen","sequence":"additional","affiliation":[{"name":"FAIR, Meta"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Koustuv","family":"Sinha","sequence":"additional","affiliation":[{"name":"FAIR, Meta"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Michael","family":"Rabbat","sequence":"additional","affiliation":[{"name":"FAIR, Meta"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yann","family":"LeCun","sequence":"additional","affiliation":[{"name":"FAIR, Meta"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Saining","family":"Xie","sequence":"additional","affiliation":[{"name":"New York University"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Zhuang","family":"Liu","sequence":"additional","affiliation":[{"name":"FAIR, Meta"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"263","reference":[{"key":"ref1","first-page":"8","article-title":"Cm3: A causal masked multimodal model of the internet","author":"Aghajanyan","year":"2022","journal-title":"arXiv preprint"},{"key":"ref2","first-page":"1","article-title":"Pixtral 12b","author":"Agrawal","year":"2024","journal-title":"arXiv preprint"},{"key":"ref3","year":"2024","journal-title":"Llama 3 model card"},{"key":"ref4","doi-asserted-by":"publisher","DOI":"10.52202\/068431-1723"},{"key":"ref5","volume-title":"ICML 2024 Tutorial: Physics of Language Models","author":"Allen-Zhu","year":"2024"},{"key":"ref6","first-page":"8","year":"2024","journal-title":"Claude"},{"key":"ref7","first-page":"13","article-title":"Layer normalization","author":"Ba","year":"2016","journal-title":"InProceedings of the Neural Information Processing Systems (NeurIPS)"},{"key":"ref8","first-page":"13","article-title":"Revisiting feature prediction for learning visual representations from video","author":"Bardes","year":"2024","journal-title":"Transactions on Machine Learning Research (TMLR)"},{"key":"ref9","first-page":"3","article-title":"High fidelity visualization of what your self-supervised representation knows about","author":"Bordes","year":"2022","journal-title":"Transactions on Machine Learning Research (TMLR)"},{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52729.2023.01764"},{"key":"ref11","first-page":"8","article-title":"Sharegpt4v: Improving large multi-modal models with better captions","author":"Chen","year":"2023","journal-title":"arXiv preprint"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.52202\/079017-0614"},{"key":"ref13","doi-asserted-by":"publisher","DOI":"10.1007\/s11432-024-4231-5"},{"key":"ref14","first-page":"1","article-title":"Instructableip: Towards general-purpose vision-language models with instruction tuning","author":"Dai","year":"2024","journal-title":"InProceedings of the Neural Information Processing Systems (NeurIPS)"},{"key":"ref15","first-page":"8","article-title":"Dreamllm: Synergistic multimodal comprehension and creation","volume-title":"InProceedings of the International Conference on Learning Representations (ICLR)","author":"Dong","year":"2024"},{"key":"ref16","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2010.11929"},{"key":"ref17","first-page":"1","article-title":"Datacomp: In search of the next generation of multimodal datasets","author":"Gadre","year":"2024","journal-title":"InProceedings of the Neural Information Processing Systems (NeurIPS)"},{"key":"ref18","first-page":"8","article-title":"Sphinx-x: Scaling data and parameters for a family of multi-modal large language models","author":"Gao","year":"2024","journal-title":"arXiv preprint"},{"key":"ref19","first-page":"13","article-title":"Planting a seed of vision in large language model","author":"Ge","year":"2023","journal-title":"arXiv preprint"},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV.2017.622"},{"key":"ref21","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2017.670"},{"key":"ref22","first-page":"13","article-title":"Gaussian error linear units (gelus)","author":"Hendrycks","year":"2016","journal-title":"arXiv preprint"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2021.emnlp-main.595"},{"key":"ref24","first-page":"13","article-title":"Gans trained by a two time-scale update rule converge to a local nash equilibrium","author":"Heusel","year":"2017","journal-title":"InProceedings of the Neural Information Processing Systems (NeurIPS)"},{"key":"ref25","first-page":"8","article-title":"The platonic representation hypothesis","volume-title":"Proceedings of the International Conference on Machine Learning (ICML)","author":"Huh","year":"2024"},{"key":"ref26","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-031-72640-8_7"},{"key":"ref27","first-page":"3","article-title":"Generating images with multimodal language models","author":"Koh","year":"2024","journal-title":"InProceedings of the Neural Information Processing Systems (NeurIPS)"},{"key":"ref28","article-title":"Learning action and reasoning-centric image editing from videos and simulations","author":"Krojer","year":"2024","journal-title":"InProceedings of the Neural Information Processing Systems (NeurIPS)"},{"key":"ref29","first-page":"8","article-title":"Obelics: An open web-scale filtered dataset of interleaved image-text documents","volume":"36","author":"Lauren\u00e7on","year":"2024","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref30","first-page":"8","article-title":"What matters when building vision-language models?","author":"Lauren\u00e7on","year":"2024","journal-title":"arXiv preprint"},{"issue":"1","key":"ref31","first-page":"1","article-title":"A path towards autonomous machine intelligence version 0.9. 2, 2022-06-27","volume":"62","author":"LeCun","year":"2022","journal-title":"Open Review"},{"key":"ref32","article-title":"Llava-onevision: Easy visual task transfer","author":"Li","year":"2024","journal-title":"arXiv preprint"},{"key":"ref33","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52733.2024.02095"},{"key":"ref34","first-page":"3","article-title":"Return of unconditional generation: A self-supervised representation generation method","author":"Li","year":"2024","journal-title":"InProceedings of the Neural Information Processing Systems (NeurIPS)"},{"key":"ref35","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-319-10602-1_48"},{"key":"ref36","doi-asserted-by":"publisher","DOI":"10.52202\/075280-1516"},{"key":"ref37","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52733.2024.02484"},{"key":"ref38","first-page":"8","author":"Liu","year":"2024","journal-title":"Llava-next: Improved reasoning, ocr, and world knowledge"},{"key":"ref39","article-title":"World model on million-length video and language with ringattention","author":"Liu","year":"2024","journal-title":"arXiv preprint"},{"key":"ref40","article-title":"Mmbench: Is your multi-modal model an all-around player?","volume-title":"Proceedings of the European Conference on Computer Vision (ECCV)","author":"Liu","year":"2024"},{"key":"ref41","first-page":"13","article-title":"Decoupled weight decay regularization","volume-title":"InProceedings of the International Conference on Learning Representations (ICLR)","author":"Loshchilov","year":"2019"},{"key":"ref42","first-page":"8","article-title":"Unified-io: A unified model for vision, language, and multi-modal tasks","volume-title":"InProceedings of the International Conference on Learning Representations (ICLR)","author":"Lu","year":"2022"},{"key":"ref43","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52733.2024.02497"},{"key":"ref44","first-page":"13","article-title":"Learn to explain: Multimodal reasoning via thought chains for science question answering","author":"Lu","year":"2022","journal-title":"InProceedings of the Neural Information Processing Systems (NeurIPS)"},{"key":"ref45","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2022.findings-acl.177"},{"key":"ref46","first-page":"13","article-title":"Mm1: Methods, analysis & insights from multimodal 11 m pre-training","author":"McKinzie","year":"2024","journal-title":"arXiv preprint"},{"key":"ref47","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV.2019.00272"},{"key":"ref48","first-page":"8","year":"2024","journal-title":"gpt4o"},{"key":"ref49","first-page":"8","article-title":"Autonomous evaluation and refinement of digital agents","volume-title":"InProceedings of the Conference on Language Modeling (COLM)","author":"Pan","year":"2024"},{"key":"ref50","first-page":"3","article-title":"Kosmos-g: Generating images in context with multimodal large language models","volume-title":"InProceedings of the International Conference on Learning Representations (ICLR)","author":"Pan","year":"2024"},{"key":"ref51","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52688.2022.01036"},{"key":"ref52","article-title":"Learning transferable visual models from natural language supervision","volume-title":"InProceedings of the International Conference on Machine Learning (ICML)","author":"Radford","year":"2021"},{"key":"ref53","doi-asserted-by":"publisher","DOI":"10.1109\/SC41405.2020.00024"},{"key":"ref54","first-page":"6","article-title":"Exploring the limits of transfer learning with a unified text-to-text transformer","author":"Roberts","year":"2019","journal-title":"Journal of Machine Learning Research (JMLR)"},{"key":"ref55","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52688.2022.01042"},{"key":"ref56","doi-asserted-by":"publisher","DOI":"10.52202\/068431-1833"},{"key":"ref57","article-title":"Visual cot: Advancing multi-modal language models with a comprehensive dataset and benchmark for chain-of-thought reasoning","author":"Shao","year":"2024","journal-title":"InProceedings of the Neural Information Processing Systems (NeurIPS)"},{"key":"ref58","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-58536-5_44"},{"key":"ref59","doi-asserted-by":"publisher","DOI":"10.1109\/cvpr52733.2024.01365"},{"key":"ref60","first-page":"8","article-title":"Generative pretraining in multimodality","volume-title":"InProceedings of the International Conference on Learning Representations (ICLR)","author":"Sun","year":"2024"},{"key":"ref61","author":"Taori","year":"2023","journal-title":"Alpaca: A strong, replicable instructionfollowing model"},{"key":"ref62","article-title":"Chameleon: Mixed-modal early-fusion foundation models","year":"2024","journal-title":"arXiv preprint"},{"key":"ref63","article-title":"Cambrian-1: A fully open, vision-centric exploration of multimodal 11 ms","author":"Tong","year":"2024","journal-title":"InProceedings of the Neural Information Processing Systems (NeurIPS)"},{"key":"ref64","first-page":"6","article-title":"Massproducing failures of multimodal systems with language models","author":"Tong","year":"2024","journal-title":"InProceedings of the Neural Information Processing Systems (NeurIPS)"},{"key":"ref65","doi-asserted-by":"publisher","DOI":"10.1109\/cvpr52733.2024.00914"},{"key":"ref66","first-page":"1","author":"Touvron","year":"2023","journal-title":"LLaMA 2: Open foundation and fine-tuned chat models"},{"key":"ref67","article-title":"Qwen2-vl: Enhancing vision-language model\u2013s perception of the world at any resolution","author":"Wang","year":"2024","journal-title":"arXiv preprint"},{"key":"ref68","article-title":"Emu3: Next-token prediction is all you need","author":"Wang","year":"2024","journal-title":"arXiv preprint"},{"key":"ref69","article-title":"Finetuned language models are zero-shot learners","volume-title":"InProceedings of the International Conference on Learning Representations (ICLR)","author":"Wei","year":"2022"},{"key":"ref70","first-page":"7","article-title":"Chain-ofthought prompting elicits reasoning in large language models","author":"Wei","year":"2022","journal-title":"InProceedings of the Neural Information Processing Systems (NeurIPS)"},{"key":"ref71","first-page":"1","article-title":"Janus: Decoupling visual encoding for unified multimodal understanding and generation","author":"Wu","year":"2024","journal-title":"arXiv preprint"},{"key":"ref72","doi-asserted-by":"publisher","DOI":"10.1109\/cvpr52733.2024.01243"},{"key":"ref73","article-title":"Vila-u: a unified foundation model integrating visual understanding and generation","author":"Wu","year":"2024","journal-title":"arXiv preprint"},{"key":"ref74","first-page":"13","year":"2024","journal-title":"grok"},{"key":"ref75","article-title":"Show-o: One single transformer to unify multimodal understanding and generation","author":"Xie","year":"2024","journal-title":"arXiv preprint"},{"key":"ref76","doi-asserted-by":"publisher","DOI":"10.5040\/9798881817916.ch-004"},{"key":"ref77","article-title":"Physics of Language Models: Part 2.1, Grade-School Math and the Hidden Reasoning Process","volume":"abs\/2407.20311","author":"Ye","year":"2024","journal-title":"ArXiv e-prints"},{"key":"ref78","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52733.2024.00913"},{"key":"ref79","first-page":"8","article-title":"Mmmu-pro: A more robust multidiscipline multimodal understanding benchmark","author":"Yue","year":"2024","journal-title":"arXiv preprint"},{"key":"ref80","first-page":"6","article-title":"When and why vision-language models behave like bags-of-words, and what to do about it?","volume-title":"Proceedings of the International Conference on Learning Representations (ICLR)","author":"Yuksekgonul","year":"2022"},{"key":"ref81","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV51070.2023.01100"},{"key":"ref82","first-page":"8","article-title":"Fine-tuning large vision-language models as decision-making agents via reinforcement learning","author":"Zhai","year":"2024","journal-title":"InProceedings of the Neural Information Processing Systems (NeurIPS)"},{"key":"ref83","article-title":"Direct preference optimization of video large multimodal models from language model reward","author":"Zhang","year":"2024","journal-title":"arXiv preprint"},{"key":"ref84","first-page":"4","article-title":"Pre-trained language models do not help auto-regressive text-to-image generation","author":"Zhang","year":"2023","journal-title":"InProceedings of the Empirical Methods in Natural Language Processing (EMNLP)"},{"key":"ref85","article-title":"Lima: Less is more for alignment","author":"Zhou","year":"2024","journal-title":"InProceedings of the Neural Information Processing Systems (NeurIPS)"},{"key":"ref86","article-title":"Transfusion: Predict the next token and diffuse images with one multi-modal model","author":"Zhou","year":"2024","journal-title":"arXiv preprint"},{"key":"ref87","article-title":"Video-star: Self-training enables video instruction tuning with any supervision","author":"Zohar","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\/11445400.pdf?arnumber=11445400","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,5,1]],"date-time":"2026-05-01T05:01:57Z","timestamp":1777611717000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11445400\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,10,19]]},"references-count":87,"URL":"https:\/\/doi.org\/10.1109\/iccv51701.2025.01579","relation":{},"subject":[],"published":{"date-parts":[[2025,10,19]]}}}