{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,18]],"date-time":"2026-07-18T18:30:32Z","timestamp":1784399432305,"version":"3.55.0"},"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"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2025,10,19]]},"DOI":"10.1109\/iccv51701.2025.01206","type":"proceedings-article","created":{"date-parts":[[2026,4,29]],"date-time":"2026-04-29T19:45:49Z","timestamp":1777491949000},"page":"12976-12986","source":"Crossref","is-referenced-by-count":1,"title":["Unified Multimodal Understanding via Byte-Pair Visual Encoding"],"prefix":"10.1109","author":[{"given":"Wanpeng","family":"Zhang","sequence":"first","affiliation":[{"name":"School of Computer Science, Peking University"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yicheng","family":"Feng","sequence":"additional","affiliation":[{"name":"School of Computer Science, Peking University"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Hao","family":"Luo","sequence":"additional","affiliation":[{"name":"School of Computer Science, Peking University"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yijiang","family":"Li","sequence":"additional","affiliation":[{"name":"University of California,San Diego"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Zihao","family":"Yue","sequence":"additional","affiliation":[{"name":"School of Information, Renmin University of China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Sipeng","family":"Zheng","sequence":"additional","affiliation":[{"name":"BeingBeyond"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Zongqing","family":"Lu","sequence":"additional","affiliation":[{"name":"School of Computer Science, Peking University"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"263","reference":[{"key":"ref1","article-title":"Qwen-vl: A versatile vision-language model for understanding, localization, text reading, and beyond","author":"Bai","year":"2023","journal-title":"arXiv preprint"},{"key":"ref2","article-title":"Qwen2.5-vl technical report","author":"Bai","year":"2025","journal-title":"arXiv preprint"},{"key":"ref3","article-title":"Hallucination of multimodal large language models: A survey","author":"Bai","year":"2024","journal-title":"arXiv preprint"},{"key":"ref4","doi-asserted-by":"publisher","DOI":"10.1109\/TPAMI.2018.2798607"},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.1109\/ACCESS.2023.3243854"},{"key":"ref6","article-title":"Analyzing the language of visual tokens","author":"David","year":"2024","journal-title":"arXiv preprint"},{"key":"ref7","author":"Chen","year":"2024","journal-title":"Allava: Harnessing gpt4v-synthesized data for a lite vision-language model"},{"key":"ref8","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-031-72643-9_22"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.52202\/075280-2142"},{"key":"ref10","article-title":"The llama 3 herd of models","author":"Dubey","year":"2024","journal-title":"arXiv preprint"},{"key":"ref11","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR46437.2021.01268"},{"key":"ref12","article-title":"Videoorion: Tokenizing object dynamics in videos","author":"Feng","year":"2024","journal-title":"arXiv preprint"},{"key":"ref13","article-title":"Mme: A comprehensive evaluation benchmark for multimodal large language models","author":"Fu","year":"2023","journal-title":"arXiv preprint"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/D19-1141"},{"key":"ref15","article-title":"Llama-adapter v2: Parameter-efficient visual instruction model","author":"Gao","year":"2023","journal-title":"arXiv preprint"},{"key":"ref16","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2017.670"},{"key":"ref17","article-title":"A semantic-based layer freezing approach to efficient fine-tuning of language models","author":"Gu","year":"2024","journal-title":"arXiv preprint"},{"key":"ref18","article-title":"Infinity-mm: Scaling multimodal performance with large-scale and high-quality instruction data","author":"Gu","year":"2024","journal-title":"arXiv preprint"},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2018.00380"},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV.2015.123"},{"key":"ref21","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2024.findings-acl.573"},{"key":"ref22","article-title":"From specific-mllm to omni-mllm: A survey about the mllms alligned with multi-modality","author":"Jiang","year":"2024","journal-title":"arXiv preprint"},{"key":"ref23","article-title":"Unified language-vision pretraining with dynamic discrete visual tokenization","author":"Jin","year":"2023","journal-title":"arXiv preprint"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.3115\/v1\/D14-1086"},{"key":"ref25","article-title":"Theoretical analysis of byte-pair encoding","author":"Kozma","year":"2024","journal-title":"arXiv preprint"},{"key":"ref26","article-title":"Llava-onevision: Easy visual task transfer","author":"Li","year":"2024","journal-title":"arXiv preprint"},{"key":"ref27","article-title":"A survey on benchmarks of multimodal large language models","author":"Li","year":"2024","journal-title":"arXiv preprint"},{"key":"ref28","article-title":"A survey on multimodal benchmarks: In the era of large ai models","author":"Li","year":"2024","journal-title":"arXiv preprint"},{"key":"ref29","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2023.emnlp-main.20"},{"key":"ref30","author":"Lin","year":"2023","journal-title":"Vila: On pre-training for visual language models"},{"key":"ref31","doi-asserted-by":"publisher","DOI":"10.52202\/075280-1516"},{"key":"ref32","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52733.2024.02484"},{"key":"ref33","article-title":"Visual instruction tuning","author":"Liu","year":"2024","journal-title":"Advances in neural information processing systems, 36"},{"key":"ref34","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-031-72658-3_13"},{"key":"ref35","author":"Lu","year":"2024","journal-title":"Deepseek-vl: Towards real-world visionlanguage understanding"},{"key":"ref36","article-title":"Unified-io: A unified model for vision, language, and multi-modal tasks","author":"Lu","year":"2022","journal-title":"arXiv preprint"},{"key":"ref37","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52733.2024.02497"},{"key":"ref38","article-title":"Learn to explain: Multimodal reasoning via thought chains for science question answering","volume-title":"The 36th Conference on Neural Information Processing Systems (NeurIPS)","author":"Lu","year":"2022"},{"key":"ref39","article-title":"Attention with markov: A framework for principled analysis of transformers via markov chains","author":"Vardhan Makkuva","year":"2024","journal-title":"arXiv preprint"},{"key":"ref40","article-title":"Language models still struggle to zero-shot reason about time series","author":"Mike","year":"2024","journal-title":"arXiv preprint"},{"issue":"8","key":"ref41","article-title":"Language models are unsupervised multitask learners","volume":"1","author":"Radford","year":"2019","journal-title":"OpenAI blog"},{"key":"ref42","first-page":"8748","article-title":"Learning transferable visual models from natural language supervision","volume-title":"International conference on machine learning","author":"Radford","year":"2021"},{"key":"ref43","first-page":"18","article-title":"Vision language models are blind","volume-title":"Proceedings of the Asian Conference on Computer Vision","author":"Rahmanzadehgervi","year":"2024"},{"key":"ref44","article-title":"Toward a theory of tokenization in 11 ms","author":"Rajaraman","year":"2024","journal-title":"arXiv preprint"},{"key":"ref45","article-title":"Generating diverse high-fidelity images with vq-vae-2","author":"Razavi","year":"2019","journal-title":"Advances in neural information processing systems, 32"},{"key":"ref46","first-page":"26","article-title":"Pixel-level bpe for auto-regressive image generation","volume-title":"Proceedings of the First Workshop on Performance and Interpretability Evaluations of Multimodal, Multipurpose, Massive-Scale Models","author":"Razzhigaev","year":"2022"},{"key":"ref47","doi-asserted-by":"publisher","DOI":"10.52202\/068431-1833"},{"key":"ref48","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-031-20074-8_9"},{"key":"ref49","article-title":"Neural machine translation of rare words with subword units","author":"Sennrich","year":"2015","journal-title":"arXiv preprint"},{"key":"ref50","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/P18-1238"},{"key":"ref51","article-title":"How to bridge the gap between modalities: A comprehensive survey on multimodal large language model","author":"Song","year":"2023","journal-title":"arXiv preprint"},{"key":"ref52","article-title":"Emu: Generative pretraining in multimodality","author":"Sun","year":"2023","journal-title":"arXiv preprint"},{"key":"ref53","doi-asserted-by":"publisher","DOI":"10.1109\/isit63088.2025.11195372"},{"key":"ref54","article-title":"Chameleon: Mixed-modal early-fusion foundation models","author":"Team","year":"2024","journal-title":"arXiv preprint"},{"key":"ref55","article-title":"Neural discrete representation learning","author":"Van Den Oord","year":"2017","journal-title":"Advances in neural information processing systems"},{"key":"ref56","article-title":"Attention is all you need","author":"Vaswani","year":"2017","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref57","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":"ref58","article-title":"Emu3: Next-token prediction is all you need","author":"Wang","year":"2024","journal-title":"arXiv preprint"},{"key":"ref59","article-title":"Deepseek-vl2: Mixture-ofexperts vision-language models for advanced multimodal understanding","author":"Wu","year":"2024","journal-title":"arXiv preprint"},{"key":"ref60","article-title":"Vitae: Vision transformer advanced by exploring intrinsic inductive bias","author":"Xu","year":"2021","journal-title":"Advances in Neural Information Processing Systems, 34"},{"key":"ref61","article-title":"Embodiedbench: Comprehensive benchmarking multi-modal large language models for vision-driven embodied agents","author":"Yang","year":"2025","journal-title":"arXiv preprint"},{"key":"ref62","article-title":"Cognitive mirage: A review of hallucinations in large language models","author":"Ye","year":"2023","journal-title":"arXiv preprint"},{"key":"ref63","author":"Ye","year":"2023","journal-title":"mplug-owl: Modularization empowers large language models with multimodality"},{"key":"ref64","doi-asserted-by":"publisher","DOI":"10.1109\/cvpr52733.2024.01239"},{"key":"ref65","doi-asserted-by":"publisher","DOI":"10.1093\/nsr\/nwae403"},{"key":"ref66","article-title":"Can mllms perform text-to-image incontext learning?","author":"Zeng","year":"2024","journal-title":"arXiv preprint"},{"key":"ref67","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV51070.2023.01100"},{"key":"ref68","doi-asserted-by":"publisher","DOI":"10.1007\/s11263-022-01739-w"},{"key":"ref69","doi-asserted-by":"publisher","DOI":"10.1109\/TPAMI.2023.3347693"},{"key":"ref70","article-title":"Llama-adapter: Efficient fine-tuning of language models with zero-init attention","author":"Zhang","year":"2023","journal-title":"arXiv preprint"},{"key":"ref71","author":"Zhang","year":"2024","journal-title":"Hyperllava: Dynamic visual and language expert tuning for multimodal large language models"},{"key":"ref72","article-title":"From pixels to tokens: Byte-pair encoding on quantized visual modalities","volume-title":"The Thirteenth International Conference on Learning Representations","author":"Zhang","year":"2025"},{"key":"ref73","article-title":"Unicode: Learning a unified codebook for multimodal large language models","author":"Zheng","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\/11445004.pdf?arnumber=11445004","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,5,1]],"date-time":"2026-05-01T05:30:49Z","timestamp":1777613449000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11445004\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,10,19]]},"references-count":73,"URL":"https:\/\/doi.org\/10.1109\/iccv51701.2025.01206","relation":{},"subject":[],"published":{"date-parts":[[2025,10,19]]}}}