{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2025,9,17]],"date-time":"2025-09-17T06:13:09Z","timestamp":1758089589440,"version":"3.44.0"},"reference-count":21,"publisher":"IEEE","license":[{"start":{"date-parts":[[2025,6,22]],"date-time":"2025-06-22T00:00:00Z","timestamp":1750550400000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2025,6,22]],"date-time":"2025-06-22T00:00:00Z","timestamp":1750550400000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"funder":[{"DOI":"10.13039\/100006190","name":"Research and Development","doi-asserted-by":"publisher","id":[{"id":"10.13039\/100006190","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100003347","name":"Fudan University","doi-asserted-by":"publisher","id":[{"id":"10.13039\/501100003347","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2025,6,22]]},"DOI":"10.1109\/dac63849.2025.11133063","type":"proceedings-article","created":{"date-parts":[[2025,9,15]],"date-time":"2025-09-15T17:35:41Z","timestamp":1757957741000},"page":"1-7","source":"Crossref","is-referenced-by-count":0,"title":["Libra: A Hybrid-Sparse Attention Accelerator Featuring Multi-Level Workload Balance"],"prefix":"10.1109","author":[{"given":"Faxian","family":"Sun","sequence":"first","affiliation":[{"name":"Fudan University,State Key Lab of Integrated Chips & Systems, and School of Microelectronics,Shanghai,China"}]},{"given":"Runzhou","family":"Zhang","sequence":"additional","affiliation":[{"name":"Fudan University,State Key Lab of Integrated Chips & Systems, and School of Microelectronics,Shanghai,China"}]},{"given":"Zhenyu","family":"Liu","sequence":"additional","affiliation":[{"name":"Fudan University,State Key Lab of Integrated Chips & Systems, and School of Microelectronics,Shanghai,China"}]},{"given":"Heng","family":"Liao","sequence":"additional","affiliation":[{"name":"Fudan University,State Key Lab of Integrated Chips & Systems, and School of Microelectronics,Shanghai,China"}]},{"given":"Zhinan","family":"Qin","sequence":"additional","affiliation":[{"name":"Fudan University,State Key Lab of Integrated Chips & Systems, and School of Microelectronics,Shanghai,China"}]},{"given":"Jianli","family":"Chen","sequence":"additional","affiliation":[{"name":"Fudan University,State Key Lab of Integrated Chips & Systems, and School of Microelectronics,Shanghai,China"}]},{"given":"Jun","family":"Yu","sequence":"additional","affiliation":[{"name":"Fudan University,State Key Lab of Integrated Chips & Systems, and School of Microelectronics,Shanghai,China"}]},{"given":"Kun","family":"Wang","sequence":"additional","affiliation":[{"name":"Fudan University,State Key Lab of Integrated Chips & Systems, and School of Microelectronics,Shanghai,China"}]}],"member":"263","reference":[{"doi-asserted-by":"publisher","key":"ref1","DOI":"10.1145\/3085572"},{"year":"2017","author":"Paszke","article-title":"Automatic differentiation in pytorch","key":"ref2"},{"key":"ref3","article-title":"Attention is all you need","author":"Vaswani","year":"2017","journal-title":"Advances in Neural Information Processing Systems"},{"issue":"04805","key":"ref4","article-title":"Bert","volume-title":"Pre-training of deep bidirectional transformers for language understanding. arXiv preprint arXiv","volume":"1810","author":"Devlin","year":"2018"},{"year":"2018","author":"Radford","article-title":"Improving language understanding by generative pre-training","key":"ref5"},{"issue":"11942","key":"ref6","article-title":"Albert: A lite bert for self-supervised learning of language representations","volume":"1909","author":"Lan","year":"2019","journal-title":"ArXiv preprint arXiv"},{"issue":"11692","key":"ref7","article-title":"Roberta","volume-title":"A robustly optimized bert pretraining approach. arXiv preprint arXiv","volume":"1907","author":"Liu","year":"2019"},{"issue":"8","key":"ref8","article-title":"Language models are unsupervised multitask learners","volume":"1","author":"Radford","year":"2019","journal-title":"OpenAI blog"},{"key":"ref9","doi-asserted-by":"crossref","DOI":"10.1145\/3307650.3322255","article-title":"Laconic deep learning inference acceleration","volume-title":"Proceedings of the 46th International Symposium on Computer Architecture","author":"Sharify"},{"issue":"14165","key":"ref10","article-title":"Language models are few-shot learners","volume":"2005","author":"Tom","year":"2020","journal-title":"ArXiv preprint arXiv"},{"issue":"11929","key":"ref11","article-title":"An image is worth 16 16 words: Transformers for image recognition at scale","volume":"2010","author":"Dosovitskiy","year":"2020","journal-title":"ArXiv preprint arXiv"},{"doi-asserted-by":"publisher","key":"ref12","DOI":"10.18653\/v1\/2020.emnlp-demos.6"},{"volume-title":"International conference on machine learning","author":"Kim","article-title":"I-bert: Integer-only bert quantization","key":"ref13"},{"doi-asserted-by":"publisher","key":"ref14","DOI":"10.1145\/3466752.3480125"},{"issue":"11990","key":"ref15","article-title":"Using deepspeed and megatron to train megatron-turing nlg 530b, a large-scale generative language model","volume":"2201","author":"Smith","year":"2022","journal-title":"ArXiv preprint arXiv"},{"doi-asserted-by":"publisher","key":"ref16","DOI":"10.1109\/tcad.2022.3181541"},{"doi-asserted-by":"publisher","key":"ref17","DOI":"10.23919\/DATE54114.2022.9774692"},{"doi-asserted-by":"publisher","key":"ref18","DOI":"10.1109\/DAC56929.2023.10247662"},{"doi-asserted-by":"publisher","key":"ref19","DOI":"10.1109\/tc.2023.3324477"},{"doi-asserted-by":"publisher","key":"ref20","DOI":"10.1109\/ICCD58817.2023.00098"},{"doi-asserted-by":"publisher","key":"ref21","DOI":"10.1109\/TCSI.2024.3426653"}],"event":{"name":"2025 62nd ACM\/IEEE Design Automation Conference (DAC)","start":{"date-parts":[[2025,6,22]]},"location":"San Francisco, CA, USA","end":{"date-parts":[[2025,6,25]]}},"container-title":["2025 62nd ACM\/IEEE Design Automation Conference (DAC)"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/11132383\/11132091\/11133063.pdf?arnumber=11133063","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,9,16]],"date-time":"2025-09-16T05:49:01Z","timestamp":1758001741000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11133063\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,6,22]]},"references-count":21,"URL":"https:\/\/doi.org\/10.1109\/dac63849.2025.11133063","relation":{},"subject":[],"published":{"date-parts":[[2025,6,22]]}}}