{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,5,4]],"date-time":"2026-05-04T10:13:07Z","timestamp":1777889587200,"version":"3.51.4"},"reference-count":73,"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\/100000001","name":"NSF","doi-asserted-by":"publisher","award":["IIS2404180"],"award-info":[{"award-number":["IIS2404180"]}],"id":[{"id":"10.13039\/100000001","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\/iccv51701.2025.00029","type":"proceedings-article","created":{"date-parts":[[2026,4,29]],"date-time":"2026-04-29T19:45:49Z","timestamp":1777491949000},"page":"228-238","source":"Crossref","is-referenced-by-count":0,"title":["X-Fusion: Introducing New Modality to Frozen Large Language Models"],"prefix":"10.1109","author":[{"given":"Sicheng","family":"Mo","sequence":"first","affiliation":[{"name":"University of California,Los Angeles"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Thao","family":"Nguyen","sequence":"additional","affiliation":[{"name":"University of Wisconsin,Madison"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Xun","family":"Huang","sequence":"additional","affiliation":[{"name":"Adobe Research"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Siddharth Srinivasan","family":"Iyer","sequence":"additional","affiliation":[{"name":"Adobe Research"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Yijun","family":"Li","sequence":"additional","affiliation":[{"name":"Adobe Research"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Yuchen","family":"Liu","sequence":"additional","affiliation":[{"name":"Adobe Research"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Abhishek","family":"Tandon","sequence":"additional","affiliation":[{"name":"Adobe Research"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Eli","family":"Shechtman","sequence":"additional","affiliation":[{"name":"Adobe Research"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Krishna Kumar","family":"Singh","sequence":"additional","affiliation":[{"name":"Adobe Research"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Yong Jae","family":"Lee","sequence":"additional","affiliation":[{"name":"University of Wisconsin,Madison"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Bolei","family":"Zhou","sequence":"additional","affiliation":[{"name":"University of California,Los Angeles"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Yuheng","family":"Li","sequence":"additional","affiliation":[{"name":"Adobe Research"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"263","reference":[{"key":"ref1","article-title":"Improving language understanding by generative pre-training","author":"Radford","journal-title":"2018. 1, 2"},{"key":"ref2","author":"Radford","journal-title":"Language models are unsupervised multitask learners"},{"key":"ref3","year":"2020","journal-title":"Language models are few-shot learners"},{"key":"ref4","article-title":"Exploring the limits of transfer learning with a unified text-to-text transformer","author":"Raffel","year":"2020","journal-title":"Journal of Machine Learning Research"},{"key":"ref5","year":"2022","journal-title":"Scaling language models: Methods, analysis & insights from training gopher"},{"key":"ref6","author":"Wei","year":"2022","journal-title":"Finetuned language models are zero-shot learners"},{"key":"ref7","author":"Hoffmann","year":"2022","journal-title":"Training compute-optimal large language models"},{"key":"ref8","article-title":"PaLM team","year":"2022","journal-title":"Palm: Scaling language modeling with pathways"},{"key":"ref9","article-title":"PaLM 2 team","year":"2023","journal-title":"Palm 2 technical report"},{"key":"ref10","year":"2024","journal-title":"Mixtral of experts"},{"key":"ref11","year":"2023","journal-title":"Mistral 7b"},{"key":"ref12","article-title":"DeepSeek-AI","year":"2024","journal-title":"Deepseek 11m: Scaling open-source language models with longtermism"},{"key":"ref13","article-title":"Gemma Team","year":"2024","journal-title":"Gemma: Open models based on gemini research and technology"},{"key":"ref14","author":"Chiang","year":"2023","journal-title":"Vicuna: An open-source chatbot impressing gpt-4 with 90 %* chatgpt quality"},{"key":"ref15","year":"2023","journal-title":"Llama 2: Open foundation and fine-tuned chat models"},{"key":"ref16","article-title":"Phi-3 team","year":"2024","journal-title":"Phi-3 technical report: A highly capable language model locally on your phone"},{"key":"ref17","article-title":"The llama 3 herd of models","volume":"abs\/2407.21783","author":"Dubey","year":"2024","journal-title":"ArXiv"},{"key":"ref18","article-title":"Qwen technical report","year":"2023","journal-title":"arXiv preprint"},{"key":"ref19","article-title":"InternLM2 team","year":"2024","journal-title":"Internlm2 technical report"},{"key":"ref20","article-title":"DeepSeek-AI","year":"2024","journal-title":"Deepseek-v2: A strong, economical, and efficient mixture-of-experts language model"},{"key":"ref21","year":"2021","journal-title":"Evaluating large language models trained on code"},{"key":"ref22","volume-title":"Codegen: An open large language model for code with multi-turn program synthesis","author":"Nijkamp","year":"2023"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.1108\/ws.2000.07949fab.004"},{"key":"ref24","author":"Yu","year":"2023","journal-title":"Scaling autoregressive multi-modal models: Pretraining and instruction tuning"},{"key":"ref25","article-title":"Gemini Team","year":"2024","journal-title":"Gemini: A family of highly capable multimodal models"},{"key":"ref26","article-title":"Chameleon: Mixed-modal early-fusion foundation models","author":"Team","year":"2024","journal-title":"arXiv preprint"},{"key":"ref27","article-title":"Emu3: Next-token prediction is all you need","author":"Wang","year":"2024","journal-title":"arXiv preprint"},{"key":"ref28","article-title":"Transfusion: Predict the next token and diffuse images with one multi-modal model","author":"Zhou","journal-title":"2024. 1, 2, 3, 6"},{"key":"ref29","author":"Dong","year":"2024","journal-title":"Dreamllm: Synergistic multimodal comprehension and creation"},{"key":"ref30","article-title":"Generating images with multimodal language models","author":"Yu Koh","year":"2023","journal-title":"NeurIPS"},{"key":"ref31","article-title":"Metamorph: Multimodal understanding and generation via instruction tuning","author":"Tong","year":"2024","journal-title":"arXiv preprint"},{"key":"ref32","doi-asserted-by":"publisher","DOI":"10.52202\/075280-1516"},{"key":"ref33","doi-asserted-by":"publisher","DOI":"10.1109\/cvpr52733.2024.02484"},{"key":"ref34","article-title":"MiniGPT-4: Enhancing vision-language understanding with advanced large language models","volume-title":"The Twelfth International Conference on Learning Representations","author":"Zhu","year":"2024"},{"key":"ref35","doi-asserted-by":"publisher","DOI":"10.1109\/cvpr52733.2024.02283"},{"key":"ref36","article-title":"Learning transferable visual models from natural language supervision","author":"Radford","year":"2021","journal-title":"arXiv preprint"},{"key":"ref37","article-title":"Investigating the catastrophic forgetting in multimodal large language model finetuning","volume-title":"Conference on Parsimony and Learning (Proceedings Track)","author":"Zhai","year":"2023"},{"key":"ref38","article-title":"Model tailor: mitigating catastrophic forgetting in multi-modal large language models","volume-title":"Proceedings of the 41st International Conference on Machine Learning, ICML\u201924. JMLR.org","author":"Zhu","year":"2024"},{"key":"ref39","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52733.2024.01365"},{"key":"ref40","article-title":"Seed-x: Multimodal models with unified multi-granularity comprehension and generation","author":"Ge","year":"2024","journal-title":"arXiv preprint"},{"key":"ref41","article-title":"Making llama see and draw with seed tokenizer","author":"Ge","year":"2023","journal-title":"arXiv preprint"},{"key":"ref42","author":"Tang","year":"2023","journal-title":"Codi-2: In-context, interleaved"},{"key":"ref43","article-title":"Januspro: Unified multimodal understanding and generation with data and model scaling","author":"Chen","year":"2025","journal-title":"arXiv preprint"},{"key":"ref44","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52688.2022.01042"},{"key":"ref45","doi-asserted-by":"publisher","DOI":"10.1109\/tpami.2018.2798607"},{"key":"ref46","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2016.91"},{"key":"ref47","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2019.00453"},{"key":"ref48","article-title":"Show-o: One single transformer to unify multimodal understanding and generation","author":"Xie","year":"2024","journal-title":"arXiv preprint"},{"key":"ref49","article-title":"Llamafusion: Adapting pretrained language models for multimodal generation","author":"Shi","year":"2024","journal-title":"arXiv preprint"},{"key":"ref50","article-title":"Outrageously large neural networks: The sparsely-gated mixture-of-experts layer","author":"Shazeer","year":"2017","journal-title":"ArXiv"},{"key":"ref51","article-title":"Gshard: Scaling giant models with conditional computation and automatic sharding","author":"Lepikhin","year":"2020","journal-title":"ArXiv, abs\/2006.16668"},{"key":"ref52","author":"Fedus","year":"2021","journal-title":"Switch transformers: Scaling to trillion parameter models with simple and efficient sparsity"},{"key":"ref53","author":"Du","year":"2021","journal-title":"Glam: Efficient scaling of language models with mixture-of-experts"},{"key":"ref54","article-title":"Scaling vision with sparse mixture of experts","author":"Riquelme","year":"2021","journal-title":"Neural Information Processing Systems"},{"key":"ref55","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV48922.2021.01415"},{"key":"ref56","article-title":"Exploring sparse moe in gans for text-conditioned image synthesis","volume-title":"ArXiv","volume":"abs\/2309.03904","author":"Zhu","year":"2023"},{"key":"ref57","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52729.2023.01457"},{"key":"ref58","article-title":"Image as a foreign language: Beit pretraining for all vision and vision-language tasks","author":"Wang","year":"2022","journal-title":"ArXiv"},{"key":"ref59","article-title":"Vlmo: Unified vision-language pre-training with mixture-of-modality-experts","volume":"abs\/2111.02358","author":"Wang","year":"2021","journal-title":"ArXiv"},{"key":"ref60","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2023.findings-emnlp.758"},{"key":"ref61","article-title":"Playground v3: Improving text-toimage alignment with deep-fusion large language models","volume":"abs\/2409.10695","author":"Liu","year":"2024","journal-title":"ArXiv"},{"key":"ref62","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52688.2022.01042"},{"key":"ref63","article-title":"Vo, Marc Szafraniec, Vasil Khalidov, Pierre Fernandez, Daniel Haziza, Francisco Massa, Alaaeldin El-Nouby, Mahmoud Assran, Nicolas Ballas, Wojciech Galuba, Russ Howes, Po-Yao (Bernie) Huang, Shang-Wen Li, Ishan Misra, Michael G","volume":"2","author":"Oquab","year":"2023","journal-title":"Rabbat, Vasu Sharma, Gabriel Synnaeve, Huijiao Xu, Herv\u00e9 J\u00e9gou, Julien Mairal, Patrick Labatut, Armand Joulin"},{"key":"ref64","article-title":"Scaling rectified flow transformers for high-resolution image synthesis","volume-title":"Forty-first international conference on machine learning","author":"Esser","year":"2024"},{"key":"ref65","doi-asserted-by":"publisher","DOI":"10.1145\/3394486.3406703"},{"key":"ref66","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-319-10602-1_48"},{"key":"ref67","article-title":"Gans trained by a two time-scale update rule converge to a local nash equilibrium","author":"Heusel","year":"2017","journal-title":"Advances in neural information processing systems, 30"},{"key":"ref68","article-title":"Blip-2: Bootstrapping language-image pre-training with frozen image encoders and large language models","volume-title":"International conference on machine learning","author":"Li","year":"2023"},{"key":"ref69","article-title":"Lawrence Zitnick, and Devi Parikh","author":"Vedantam","year":"2015","journal-title":"Cider: Consensus-based image description evaluation"},{"key":"ref70","author":"Zhang","year":"2020","journal-title":"Bertscore: Evaluating text generation with bert"},{"key":"ref71","doi-asserted-by":"publisher","DOI":"10.52202\/068431-1723"},{"key":"ref72","author":"Hendrycks","year":"2021","journal-title":"Measuring massive multitask language understanding"},{"key":"ref73","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV51070.2023.00387"}],"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\/11445488.pdf?arnumber=11445488","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,5,1]],"date-time":"2026-05-01T05:13:14Z","timestamp":1777612394000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11445488\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,10,19]]},"references-count":73,"URL":"https:\/\/doi.org\/10.1109\/iccv51701.2025.00029","relation":{},"subject":[],"published":{"date-parts":[[2025,10,19]]}}}