{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,22]],"date-time":"2026-07-22T16:15:10Z","timestamp":1784736910982,"version":"3.55.0"},"reference-count":17,"publisher":"IEEE","license":[{"start":{"date-parts":[[2025,5,26]],"date-time":"2025-05-26T00:00:00Z","timestamp":1748217600000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2025,5,26]],"date-time":"2025-05-26T00:00:00Z","timestamp":1748217600000},"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":[[2025,5,26]]},"DOI":"10.23919\/wiopt66569.2025.11123401","type":"proceedings-article","created":{"date-parts":[[2025,8,26]],"date-time":"2025-08-26T19:04:05Z","timestamp":1756235045000},"page":"1-8","source":"Crossref","is-referenced-by-count":5,"title":["Large Language Model Partitioning for Low-Latency Inference at the Edge"],"prefix":"10.23919","author":[{"given":"Dimitrios","family":"Kafetzis","sequence":"first","affiliation":[{"name":"Athens University of Economics and Business,Department of Informatics,Athens,Greece"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Ramin","family":"Khalili","sequence":"additional","affiliation":[{"name":"Huawei European Research Center,Munich,Germany"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Iordanis","family":"Koutsopoulos","sequence":"additional","affiliation":[{"name":"Athens University of Economics and Business,Department of Informatics,Athens,Greece"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"263","reference":[{"key":"ref1","doi-asserted-by":"publisher","DOI":"10.1109\/jiot.2024.3524255"},{"key":"ref2","article-title":"Splitllm: Collaborative inference of 11 ms for model placement and throughput optimization","author":"Mudvari","year":"2024","journal-title":"arXiv preprint"},{"key":"ref3","doi-asserted-by":"publisher","DOI":"10.1109\/INFOCOM52122.2024.10621342"},{"key":"ref4","first-page":"1877","article-title":"Language models are few-shot learners","volume":"33","author":"Brown","year":"2020","journal-title":"Advances in Neural Information Processing Systems"},{"key":"ref5","article-title":"6g technology overview","year":"2024","journal-title":"One6G"},{"key":"ref6","doi-asserted-by":"publisher","DOI":"10.1109\/EuCNC\/6GSummit60053.2024.10597111"},{"key":"ref7","doi-asserted-by":"publisher","DOI":"10.1109\/TMC.2022.3177569"},{"key":"ref8","doi-asserted-by":"publisher","DOI":"10.1109\/LNET.2023.3260567"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.1109\/ACCESS.2023.3257342"},{"key":"ref10","article-title":"GPipe: efficient training of giant neural networks using pipeline parallelism","volume-title":"Red Hook","author":"Huang","year":"2019"},{"key":"ref11","doi-asserted-by":"publisher","DOI":"10.1145\/3341301.3359646"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.1109\/CLUSTER51413.2022.00042"},{"key":"ref13","first-page":"7937","article-title":"Memory-efficient pipeline-parallel dnn training","volume-title":"International Conference on Machine Learning","author":"Narayanan"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.1109\/ICASSP48485.2024.10446916"},{"key":"ref15","author":"Kafetzis","year":"2025","journal-title":"Distributed Transformer Inference Simulator"},{"key":"ref16","article-title":"Google cluster-usage traces: format+ schema","author":"Reiss","year":"2012","journal-title":"Tech. Rep. ISTC-CC-TR-12-101, Google Inc."},{"key":"ref17","article-title":"Attention is all you need","author":"Vaswani","year":"2017","journal-title":"Neural Information Processing Systems"}],"event":{"name":"2025 23rd International Symposium on Modeling and Optimization in Mobile, Ad Hoc, and Wireless Networks (WiOpt)","location":"Linkoping, Sweden","start":{"date-parts":[[2025,5,26]]},"end":{"date-parts":[[2025,5,29]]}},"container-title":["2025 23rd International Symposium on Modeling and Optimization in Mobile, Ad Hoc, and Wireless Networks (WiOpt)"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/11123177\/11123184\/11123401.pdf?arnumber=11123401","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,8,27]],"date-time":"2025-08-27T05:11:48Z","timestamp":1756271508000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11123401\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,5,26]]},"references-count":17,"URL":"https:\/\/doi.org\/10.23919\/wiopt66569.2025.11123401","relation":{},"subject":[],"published":{"date-parts":[[2025,5,26]]}}}