{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,16]],"date-time":"2026-06-16T05:08:34Z","timestamp":1781586514508,"version":"3.54.5"},"reference-count":65,"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\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["62302309,62171248"],"award-info":[{"award-number":["62302309,62171248"]}],"id":[{"id":"10.13039\/501100001809","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\/iccv51701.2025.01767","type":"proceedings-article","created":{"date-parts":[[2026,4,29]],"date-time":"2026-04-29T19:45:49Z","timestamp":1777491949000},"page":"19011-19021","source":"Crossref","is-referenced-by-count":2,"title":["FastVAR: Linear Visual Autoregressive Modeling Via Cached Token Pruning"],"prefix":"10.1109","author":[{"given":"Hang","family":"Guo","sequence":"first","affiliation":[{"name":"Tsinghua University"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yawei","family":"Li","sequence":"additional","affiliation":[{"name":"ETH Zurich"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Taolin","family":"Zhang","sequence":"additional","affiliation":[{"name":"Tsinghua University"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jiangshan","family":"Wang","sequence":"additional","affiliation":[{"name":"Tsinghua University"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Tao","family":"Dai","sequence":"additional","affiliation":[{"name":"Shenzhen University"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Shu-Tao","family":"Xia","sequence":"additional","affiliation":[{"name":"Tsinghua University"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Luca","family":"Benini","sequence":"additional","affiliation":[{"name":"ETH Zurich"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"263","reference":[{"key":"ref1","article-title":"GPT-4 technical report","author":"Achiam","year":"2023","journal-title":"arXiv preprint"},{"key":"ref2","doi-asserted-by":"publisher","DOI":"10.1109\/CVPRW59228.2023.00484"},{"key":"ref3","article-title":"Token Merging: Your ViT but faster","author":"Bolya","year":"2022","journal-title":"arXiv preprint"},{"key":"ref4","first-page":"18771901","article-title":"Language models are few-shot learners","volume":"33","author":"Brown","year":"2020","journal-title":"NeurIPS"},{"key":"ref5","article-title":"Medusa: Simple LLM inference acceleration framework with multiple decoding heads","author":"Cai","year":"2024","journal-title":"arXiv preprint"},{"key":"ref6","article-title":"Muse: Text-to-image generation via masked generative transformers","author":"Chang","year":"2023","journal-title":"arXiv preprint"},{"key":"ref7","article-title":"PixArt-Alpha: Fast training of diffusion transformer for photorealistic text-to-image synthesis","author":"Chen","year":"2023","journal-title":"arXiv preprint"},{"key":"ref8","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-031-73411-3_5"},{"key":"ref9","first-page":"1691","article-title":"Generative pretraining from pixels","volume-title":"ICML","author":"Chen","year":"2020"},{"key":"ref10","article-title":"Collaborative decoding makes visual auto-regressive modeling efficient","author":"Chen","year":"2024","journal-title":"arXiv preprint"},{"issue":"70","key":"ref11","first-page":"1","article-title":"Scaling instructionfinetuned language models","volume":"25","author":"Won Chung","year":"2024","journal-title":"Journal of Machine Learning Research"},{"key":"ref12","article-title":"FlashAttention-2: Faster attention with better parallelism and work partitioning","author":"Dao","year":"2024","journal-title":"ICLR"},{"key":"ref13","article-title":"Flash Attention: Fast and memory-efficient exact attention with IO-awareness","author":"Dao","year":"2022","journal-title":"NeurIPS"},{"key":"ref14","first-page":"41714186","article-title":"BERT: Pre-training of deep bidirectional transformers for language understanding","author":"Devlin","year":"2019","journal-title":"ACL"},{"key":"ref15","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR46437.2021.01268"},{"key":"ref16","article-title":"Scaling rectified flow transformers for high-resolution image synthesis","author":"Esser","year":"2024","journal-title":"ICML"},{"key":"ref17","doi-asserted-by":"publisher","DOI":"10.52202\/075280-0731"},{"key":"ref18","article-title":"LazyLLM: Dynamic token pruning for efficient long context 11 m inference","author":"Fu","year":"2024","journal-title":"arXiv preprint"},{"key":"ref19","article-title":"Model tells you what to discard: Adaptive KV cache compression for LLMs","author":"Ge","year":"2023","journal-title":"arXiv preprint"},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.52202\/075280-2270"},{"key":"ref21","article-title":"Intlora: Integral low-rank adaptation of quantized diffusion models","author":"Guo","year":"2024","journal-title":"arXiv preprint"},{"key":"ref22","article-title":"Infinity: Scaling bitwise autoregressive modeling for high-resolution image synthesis","author":"Han","year":"2024","journal-title":"arXiv preprint"},{"key":"ref23","article-title":"ZIPAR: Accelerating autoregressive image generation through spatial locality","author":"He","year":"2024","journal-title":"arXiv preprint"},{"key":"ref24","article-title":"Scaling laws for autoregressive generative modeling","author":"Henighan","journal-title":"arXiv preprint"},{"key":"ref25","first-page":"6840","article-title":"Denoising diffusion probabilistic models","volume":"33","author":"Ho","year":"2020","journal-title":"NeurIPS"},{"key":"ref26","article-title":"Scaling laws for neural language models","author":"Kaplan","journal-title":"arXiv preprint"},{"key":"ref27","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52688.2022.01123"},{"key":"ref28","first-page":"19274","article-title":"Fast inference from transformers via speculative decoding","volume-title":"ICML","author":"Leviathan","year":"2023"},{"key":"ref29","article-title":"Playground v2.5: Three insights towards enhancing aesthetic quality in text-to-image generation","author":"Li","year":"2024","journal-title":"arXiv preprint"},{"key":"ref30","article-title":"Faster Diffusion: Rethinking the role of unet encoder in diffusion models","author":"Li","year":"2023","journal-title":"NeurIPS"},{"key":"ref31","article-title":"Autoregressive image generation without vector quantization","author":"Li","year":"2024","journal-title":"arXiv preprint"},{"key":"ref32","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV51070.2023.01608"},{"key":"ref33","article-title":"Lumina-mgpt: Illuminate flexible photorealistic text-to-image generation with multimodal generative pretraining","author":"Liu","year":"2024","journal-title":"arXiv preprint"},{"key":"ref34","article-title":"Faster diffusion via temporal attention decomposition","author":"Liu","year":"2024","journal-title":"arXiv preprint"},{"key":"ref35","doi-asserted-by":"publisher","DOI":"10.48550\/ARXIV.1907.11692"},{"key":"ref36","article-title":"Learning-to-Cache: Accelerating diffusion transformer via layer caching","author":"Ma","year":"2024","journal-title":"arXiv preprint"},{"key":"ref37","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52733.2024.01492"},{"key":"ref38","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v38i5.28206"},{"key":"ref39","doi-asserted-by":"publisher","DOI":"10.1145\/3680528.3687587"},{"key":"ref40","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52729.2023.01374"},{"key":"ref41","article-title":"SDXL: Improving latent diffusion models for high-resolution image synthesis","author":"Podell","year":"2023","journal-title":"arXiv preprint"},{"key":"ref42","article-title":"Generating diverse high-fidelity images with VQ-VA E-2","volume":"32","author":"Razavi","year":"2019","journal-title":"NeurIPS"},{"key":"ref43","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52688.2022.01042"},{"key":"ref44","article-title":"Progressive distillation for fast sampling of diffusion models","author":"Salimans","year":"2022","journal-title":"arXiv preprint"},{"key":"ref45","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2023.acl-long.689"},{"key":"ref46","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52729.2023.00196"},{"key":"ref47","article-title":"Autoregressive model beats diffusion: Llama for scalable image generation","author":"Sun","year":"2024","journal-title":"arXiv preprint"},{"key":"ref48","article-title":"HART: Efficient visual generation with hybrid autoregressive transformer","author":"Tang","year":"2024","journal-title":"arXiv preprint"},{"key":"ref49","article-title":"Accelerating autoregressive text-to-image generation with training-free speculative jacobi decoding","author":"Teng","year":"2024","journal-title":"arXiv preprint"},{"key":"ref50","article-title":"Visual autoregressive modeling: Scalable image generation via next-scale prediction","author":"Tian","year":"2024","journal-title":"NeurIPS"},{"key":"ref51","article-title":"Conditional image generation with pixelcnn decoders","volume":"29","author":"Van den Oord","year":"2016","journal-title":"NeurIPS"},{"key":"ref52","article-title":"Neural discrete representation learning","volume":"30","author":"Van Den Oord","year":"2017","journal-title":"NeurIPS"},{"key":"ref53","article-title":"Attention is all you need","author":"Vaswani","year":"2017","journal-title":"NeurIPS"},{"key":"ref54","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR52733.2024.01522"},{"key":"ref55","article-title":"EMU 3: Next-token prediction is all you need","author":"Wang","year":"2024","journal-title":"arXiv preprint"},{"key":"ref56","doi-asserted-by":"publisher","DOI":"10.1109\/cvpr52734.2025.01210"},{"key":"ref57","article-title":"VILA-U: A unified foundation model integrating visual understanding and generation","author":"Wu","year":"2024","journal-title":"arXiv preprint"},{"key":"ref58","article-title":"Efficient streaming language models with attention sinks","author":"Xiao","year":"2023","journal-title":"arXiv preprint"},{"key":"ref59","article-title":"Show-o: One single transformer to unify multimodal understanding and generation","author":"Xie","year":"2024","journal-title":"arXiv preprint"},{"key":"ref60","article-title":"Qwen2.5 technical report","author":"Yang","year":"2024","journal-title":"arXiv preprint"},{"key":"ref61","article-title":"Vector-quantized image modeling with improved VQGAN","author":"Yu","year":"2021","journal-title":"arXiv preprint"},{"key":"ref62","article-title":"Token pruning for caching better: 9 times acceleration on stable diffusion for free","author":"Zhang","year":"2024","journal-title":"arXiv preprint"},{"key":"ref63","first-page":"34661","article-title":"H2O: Heavy-hitter oracle for efficient generative inference of large language models","volume":"36","author":"Zhang","year":"2023","journal-title":"NeurIPS"},{"key":"ref64","first-page":"23412","article-title":"MOVQ: Modulating quantized vectors for highfidelity image generation","volume":"35","author":"Zheng","year":"2022","journal-title":"NeurIPS"},{"key":"ref65","article-title":"Accelerating diffusion transformers with tokenwise feature caching","author":"Zou","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\/11444600.pdf?arnumber=11444600","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,4,30]],"date-time":"2026-04-30T06:16:54Z","timestamp":1777529814000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11444600\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,10,19]]},"references-count":65,"URL":"https:\/\/doi.org\/10.1109\/iccv51701.2025.01767","relation":{},"subject":[],"published":{"date-parts":[[2025,10,19]]}}}