{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2025,11,21]],"date-time":"2025-11-21T05:56:53Z","timestamp":1763704613527,"version":"3.45.0"},"reference-count":46,"publisher":"IEEE","license":[{"start":{"date-parts":[[2025,10,26]],"date-time":"2025-10-26T00:00:00Z","timestamp":1761436800000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2025,10,26]],"date-time":"2025-10-26T00:00:00Z","timestamp":1761436800000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"funder":[{"DOI":"10.13039\/100000001","name":"National Science Foundation","doi-asserted-by":"publisher","id":[{"id":"10.13039\/100000001","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2025,10,26]]},"DOI":"10.1109\/iccad66269.2025.11240907","type":"proceedings-article","created":{"date-parts":[[2025,11,20]],"date-time":"2025-11-20T18:39:34Z","timestamp":1763663974000},"page":"1-9","source":"Crossref","is-referenced-by-count":0,"title":["SAGE: Saliency-Aware Grouping for Efficient Mapping of LLMs on Analog Compute-in-Memory"],"prefix":"10.1109","author":[{"given":"Yayue","family":"Hou","sequence":"first","affiliation":[{"name":"Rensselaer Polytechnic Institute,Troy,USA"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Zhenyu","family":"Liu","sequence":"additional","affiliation":[{"name":"Rensselaer Polytechnic Institute,Troy,USA"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Garrett","family":"Gagnon","sequence":"additional","affiliation":[{"name":"Rensselaer Polytechnic Institute,Troy,USA"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Hsinyu","family":"Tsai","sequence":"additional","affiliation":[{"name":"IBM Research,San Jose,USA"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Kaoutar","family":"El Maghraoui","sequence":"additional","affiliation":[{"name":"IBM Research,USA"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Geoffrey W.","family":"Burr","sequence":"additional","affiliation":[{"name":"IBM Research,San Jose,USA"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Liu","family":"Liu","sequence":"additional","affiliation":[{"name":"Rensselaer Polytechnic Institute,Troy,USA"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"263","reference":[{"key":"ref1","doi-asserted-by":"publisher","DOI":"10.1038\/s41467-024-45670-9"},{"year":"2024","key":"ref2","article-title":"Llama 3 model card"},{"key":"ref3","doi-asserted-by":"publisher","DOI":"10.1145\/3579371.3589062"},{"article-title":"Piqa: Reasoning about physical commonsense in natural language","year":"2019","author":"Bisk","key":"ref4"},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.1038\/s43588-024-00753-x"},{"key":"ref6","doi-asserted-by":"publisher","DOI":"10.1109\/ISSCC42615.2023.10067289"},{"key":"ref7","doi-asserted-by":"publisher","DOI":"10.1145\/3007787.3001140"},{"key":"ref8","doi-asserted-by":"publisher","DOI":"10.1109\/VLSITechnologyandCir46769.2022.9830490"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.1109\/TCAD.2024.3416256"},{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.1109\/TCAD.2025.3531255"},{"key":"ref11","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2021.emnlp-main.98"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.1109\/IRPS48203.2023.10117874"},{"article-title":"A framework for few-shot language model evaluation","year":"2024","author":"Gao","key":"ref13"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.3389\/fnins.2017.00538"},{"key":"ref15","doi-asserted-by":"publisher","DOI":"10.1109\/MICRO61859.2024.00121"},{"key":"ref16","doi-asserted-by":"publisher","DOI":"10.23919\/DATE64628.2025.10993217"},{"key":"ref17","doi-asserted-by":"publisher","DOI":"10.1109\/TVLSI.2022.3221390"},{"article-title":"Mistral 7b","year":"2023","author":"Jiang","key":"ref18"},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.1109\/JSSC.2022.3140414"},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.1109\/JSSC.2024.3362699"},{"key":"ref21","doi-asserted-by":"publisher","DOI":"10.1038\/s41928-023-01010-1"},{"key":"ref22","doi-asserted-by":"publisher","DOI":"10.1109\/HPCA57654.2024.00065"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.1109\/TCAD.2023.3344524"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.1109\/TVLSI.2023.3299509"},{"key":"ref25","doi-asserted-by":"publisher","DOI":"10.1109\/ISCA45697.2020.00073"},{"key":"ref26","doi-asserted-by":"publisher","DOI":"10.1109\/TVLSI.2023.3337777"},{"article-title":"Pointer sentinel mixture models","year":"2016","author":"Merity","key":"ref27"},{"key":"ref28","doi-asserted-by":"publisher","DOI":"10.1109\/ICECS46596.2019.8964852"},{"key":"ref29","doi-asserted-by":"publisher","DOI":"10.1109\/IEDM19573.2019.8993491"},{"key":"ref30","doi-asserted-by":"publisher","DOI":"10.1145\/3503222.3507738"},{"key":"ref31","doi-asserted-by":"publisher","DOI":"10.1038\/s41467-023-40770-4"},{"key":"ref32","doi-asserted-by":"publisher","DOI":"10.1109\/AICAS51828.2021.9458494"},{"key":"ref33","doi-asserted-by":"crossref","DOI":"10.1162\/tacl_a_00266","article-title":"Coqa: A conversational question answering challenge","author":"Reddy","year":"2019"},{"article-title":"Winogrande: An adversarial winograd schema challenge at scale","year":"2019","author":"Sakaguchi","key":"ref34"},{"key":"ref35","doi-asserted-by":"publisher","DOI":"10.1145\/3007787.3001139"},{"key":"ref36","doi-asserted-by":"publisher","DOI":"10.1109\/HPCA.2017.55"},{"article-title":"Llama 2: Open foundation and fine-tuned chat models","year":"2023","author":"Touvron","key":"ref37"},{"key":"ref38","doi-asserted-by":"publisher","DOI":"10.1109\/ISSCC42614.2022.9731716"},{"key":"ref39","doi-asserted-by":"publisher","DOI":"10.1038\/s41586-022-04992-8"},{"key":"ref40","doi-asserted-by":"publisher","DOI":"10.1109\/TVLSI.2025.3526363"},{"key":"ref41","first-page":"1","article-title":"Retransformer: Reram-based processing-in-memory architecture for transformer acceleration","volume-title":"Proceedings of the 39th International Conference on Computer-Aided Design","author":"Yang"},{"key":"ref42","doi-asserted-by":"publisher","DOI":"10.1038.\/s41586-020-1942-4"},{"key":"ref43","doi-asserted-by":"publisher","DOI":"10.1109\/JSSC.2024.3457898"},{"key":"ref44","first-page":"521","article-title":"Orca: A distributed serving system for {Transformer-Based} generative models","volume-title":"16th USENIX Symposium on Operating Systems Design and Implementation (OSDI 22)","author":"Yu"},{"key":"ref45","doi-asserted-by":"publisher","DOI":"10.1109\/ISCA52012.2021.00029"},{"key":"ref46","doi-asserted-by":"publisher","DOI":"10.1109\/TCAD.2022.3184464"}],"event":{"name":"2025 IEEE\/ACM International Conference On Computer Aided Design (ICCAD)","start":{"date-parts":[[2025,10,26]]},"location":"Munich, Germany","end":{"date-parts":[[2025,10,30]]}},"container-title":["2025 IEEE\/ACM International Conference On Computer Aided Design (ICCAD)"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/11240608\/11240621\/11240907.pdf?arnumber=11240907","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,11,21]],"date-time":"2025-11-21T05:46:06Z","timestamp":1763703966000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11240907\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,10,26]]},"references-count":46,"URL":"https:\/\/doi.org\/10.1109\/iccad66269.2025.11240907","relation":{},"subject":[],"published":{"date-parts":[[2025,10,26]]}}}