{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,4,23]],"date-time":"2026-04-23T14:46:49Z","timestamp":1776955609198,"version":"3.51.4"},"reference-count":28,"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"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2025,6,22]]},"DOI":"10.1109\/dac63849.2025.11132497","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":2,"title":["Move Less, Retrieve Fast: A Retrieval-in-Memory Architecture for Language Models"],"prefix":"10.1109","author":[{"given":"Jiaxian","family":"Chen","sequence":"first","affiliation":[{"name":"Shenzhen University,College of Computer Science and Software Engineering"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Yuxuan","family":"Qi","sequence":"additional","affiliation":[{"name":"Shenzhen University,College of Computer Science and Software Engineering"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Jianan","family":"Yuan","sequence":"additional","affiliation":[{"name":"Shenzhen University,College of Computer Science and Software Engineering"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Kaoyi","family":"Sun","sequence":"additional","affiliation":[{"name":"Shenzhen University,College of Computer Science and Software Engineering"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Tianyu","family":"Wang","sequence":"additional","affiliation":[{"name":"Shenzhen University,College of Computer Science and Software Engineering"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Chenlin","family":"Ma","sequence":"additional","affiliation":[{"name":"Shenzhen University,College of Computer Science and Software Engineering"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Yi","family":"Wang","sequence":"additional","affiliation":[{"name":"Shenzhen University,College of Computer Science and Software Engineering"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"263","reference":[{"key":"ref1","article-title":"CACTI","year":"2024"},{"key":"ref2","article-title":"GPT-2 Large","year":"2024"},{"key":"ref3","article-title":"LM-Eval","year":"2024"},{"key":"ref4","article-title":"The Pile","year":"2024"},{"key":"ref5","first-page":"2206","article-title":"Improving Language Models by Retrieving from Trillions of Tokens","volume-title":"International Conference on Machine Learning (ICML)","author":"Borgeaud"},{"key":"ref6","doi-asserted-by":"publisher","DOI":"10.1109\/DAC56929.2023.10247871"},{"key":"ref7","doi-asserted-by":"publisher","DOI":"10.1145\/3649329.3657375"},{"key":"ref8","first-page":"4171","article-title":"Bert: Pre-Training of Deep Bidirectional Transformers for Language Understanding","author":"Devlin","year":"2019","journal-title":"North American Chapter of the Association for Computational Linguistics (NAACL-HLT)"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2021.eacl-main.74"},{"issue":"251","key":"ref10","first-page":"1","article-title":"Atlas: Few-shot Learning with Retrieval Augmented Language Models","volume":"24","author":"Izacard","year":"2023","journal-title":"Journal of Machine Learning Research (JMLR)"},{"key":"ref11","first-page":"1","article-title":"Chameleon: A Heterogeneous and Disaggregated Accelerator System for Retrieval-Augmented Language Models","volume-title":"International Workshop on Heterogeneous High-Performance Reconfigurable Computing","author":"Jiang"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.1109\/TBDATA.2019.2921572"},{"key":"ref13","doi-asserted-by":"publisher","DOI":"10.1109\/TPAMI.2010.57"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.1109\/ISCA52012.2021.00013"},{"key":"ref15","doi-asserted-by":"publisher","DOI":"10.1109\/HPCA53966.2022.00021"},{"key":"ref16","doi-asserted-by":"publisher","DOI":"10.1109\/LCA.2020.2973991"},{"key":"ref17","doi-asserted-by":"publisher","DOI":"10.1109\/TPAMI.2018.2889473"},{"key":"ref18","doi-asserted-by":"publisher","DOI":"10.1145\/3079856.3080251"},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2020.emnlp-main.365"},{"key":"ref20","first-page":"1","article-title":"In-Context Pretraining: Language Modeling Beyond Document Boundaries","volume-title":"International Conference on Learning Representations (ICLR)","author":"Shi"},{"key":"ref21","doi-asserted-by":"publisher","DOI":"10.1109\/TCAD.2023.3262201"},{"key":"ref22","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2022.acl-long.226"},{"key":"ref23","first-page":"1","article-title":"Visually-Augmented Language Modeling","volume-title":"International Conference on Learning Representations (ICLR)","author":"Wang"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.1145\/3649329.3657357"},{"key":"ref25","first-page":"1","article-title":"Improving Natural Language Understanding with Computation-Efficient Retrieval Augmentation","volume-title":"International Conference on Learning Representations (ICLR)","author":"Wu"},{"key":"ref26","doi-asserted-by":"publisher","DOI":"10.1007\/s00778-021-00680-7"},{"key":"ref27","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2022.emnlp-main.382"},{"key":"ref28","doi-asserted-by":"publisher","DOI":"10.1109\/DAC56929.2023.10247755"}],"event":{"name":"2025 62nd ACM\/IEEE Design Automation Conference (DAC)","location":"San Francisco, CA, USA","start":{"date-parts":[[2025,6,22]]},"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\/11132497.pdf?arnumber=11132497","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,9,16]],"date-time":"2025-09-16T05:47:16Z","timestamp":1758001636000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11132497\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,6,22]]},"references-count":28,"URL":"https:\/\/doi.org\/10.1109\/dac63849.2025.11132497","relation":{},"subject":[],"published":{"date-parts":[[2025,6,22]]}}}