{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,8,31]],"date-time":"2026-08-31T19:18:03Z","timestamp":1788203883187,"version":"build-2803163510"},"reference-count":47,"publisher":"Institute of Electrical and Electronics Engineers (IEEE)","issue":"10","license":[{"start":{"date-parts":[[2026,10,1]],"date-time":"2026-10-01T00:00:00Z","timestamp":1790812800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/ieeexplore.ieee.org\/Xplorehelp\/downloads\/license-information\/IEEE.html"},{"start":{"date-parts":[[2026,10,1]],"date-time":"2026-10-01T00:00:00Z","timestamp":1790812800000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2026,10,1]],"date-time":"2026-10-01T00:00:00Z","timestamp":1790812800000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"funder":[{"DOI":"10.13039\/501100000266","name":"Engineering and Physical Sciences Research Council","doi-asserted-by":"crossref","award":["EP\/X037304\/1"],"award-info":[{"award-number":["EP\/X037304\/1"]}],"id":[{"id":"10.13039\/501100000266","id-type":"DOI","asserted-by":"crossref"}]},{"name":"Royal Society Industry Fellowship"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IEEE Trans. Parallel Distrib. Syst."],"published-print":{"date-parts":[[2026,10]]},"DOI":"10.1109\/tpds.2026.3719351","type":"journal-article","created":{"date-parts":[[2026,8,3]],"date-time":"2026-08-03T19:03:16Z","timestamp":1785783796000},"page":"2205-2218","source":"Crossref","is-referenced-by-count":0,"title":["SynergyScale: Optimizing Offloading and Task Partitioning for Efficient Model Training"],"prefix":"10.1109","volume":"37","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-6193-809X","authenticated-orcid":false,"given":"Xiaoyang","family":"Sun","sequence":"first","affiliation":[{"name":"University of Leeds","place":["Leeds, U.K."]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-4102-233X","authenticated-orcid":false,"given":"Jie","family":"Xu","sequence":"additional","affiliation":[{"name":"University of Leeds","place":["Leeds, U.K."]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-6157-0662","authenticated-orcid":false,"given":"Zheng","family":"Wang","sequence":"additional","affiliation":[{"name":"University of Leeds","place":["Leeds, U.K."]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"263","reference":[{"key":"ref1","doi-asserted-by":"publisher","DOI":"10.1109\/ICCV48922.2021.00986"},{"key":"ref2","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1810.04805"},{"key":"ref3","first-page":"1877","article-title":"Language models are few-shot learners","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"33","author":"Brown","year":"2020"},{"key":"ref4","article-title":"Scaling laws for neural language models","author":"Kaplan","year":"2020"},{"key":"ref5","article-title":"Qwen3 technical report","author":"Yang","year":"2025"},{"key":"ref6","article-title":"Llama: Open and efficient foundation language models","author":"Touvron","year":"2023"},{"key":"ref7","article-title":"DeepSeek-V3 Technical Report","author":"Liu","year":"2024"},{"key":"ref8","first-page":"551","article-title":"Zero-offload: Democratizing billion-scale model training","volume-title":"Proc. USENIX Annu. Tech. Conf.","author":"Ren","year":"2021"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.14778\/3415478.3415530"},{"key":"ref10","article-title":"Large scale distributed deep networks","volume":"25","author":"Dean","year":"2012","journal-title":"NeurIPS"},{"key":"ref11","doi-asserted-by":"publisher","DOI":"10.48550\/arxiv.1811.06965"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.1145\/3341301.3359646"},{"key":"ref13","article-title":"Megatron-lm: Training multi-billion parameter language models using model parallelism","author":"Shoeybi","year":"2019"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.1145\/3302424.3303953"},{"key":"ref15","article-title":"Pipedream: Fast and efficient pipeline parallel DNN training","author":"Harlap","year":"2018"},{"key":"ref16","doi-asserted-by":"publisher","DOI":"10.1145\/3243904"},{"key":"ref17","doi-asserted-by":"publisher","DOI":"10.1145\/3373376.3378465"},{"key":"ref18","doi-asserted-by":"publisher","DOI":"10.1145\/3373376.3378530"},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.1109\/HPCA51647.2021.00057"},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.1145\/3442188.3445922"},{"key":"ref21","doi-asserted-by":"publisher","DOI":"10.1109\/SC41404.2022.00076"},{"key":"ref22","article-title":"Training large neural networks with constant memory using a new execution algorithm","author":"Pudipeddi","year":"2020"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.1145\/3458817.3476205"},{"key":"ref24","article-title":"Adam: A method for stochastic optimization","volume-title":"Proc. ICLR","author":"Kingma","year":"2015"},{"key":"ref25","doi-asserted-by":"publisher","DOI":"10.1109\/SC41405.2020.00024"},{"key":"ref26","doi-asserted-by":"publisher","DOI":"10.1145\/3676641.3715983"},{"key":"ref27","doi-asserted-by":"publisher","DOI":"10.1145\/3760250.3762217"},{"key":"ref28","doi-asserted-by":"publisher","DOI":"10.1145\/3712285.3759864"},{"key":"ref30","doi-asserted-by":"publisher","DOI":"10.48550\/ARXIV.1706.03762"},{"key":"ref31","article-title":"The Llama 4 herd: The beginning of a new era of natively multimodal AI innovation","year":"2025"},{"key":"ref34","article-title":"Zenflow: Enabling stall-free offloading training via asynchronous updates","author":"Lan","year":"2025"},{"key":"ref35","doi-asserted-by":"publisher","DOI":"10.1145\/3320060"},{"key":"ref37","doi-asserted-by":"publisher","DOI":"10.1109\/ICDCS54860.2022.00087"},{"key":"ref38","doi-asserted-by":"publisher","DOI":"10.1109\/IPDPS54959.2023.00031"},{"key":"ref39","first-page":"1","article-title":"Efficient large-scale language model training on GPU clusters using megatron-lm","volume-title":"Proc. Int. Conf. High Perform. Comput., Netw., Storage Anal.","author":"Narayanan","year":"2021"},{"key":"ref40","doi-asserted-by":"publisher","DOI":"10.1145\/3437801.3441593"},{"key":"ref41","first-page":"479","article-title":"Pipelined backpropagation at scale: Training large models without batches","volume-title":"Proc. Mach. Learn. Syst.","volume":"3","author":"Kosson","year":"2021"},{"key":"ref42","article-title":"PipeTransformer: Automated elastic pipelining for distributed training of transformers","author":"He","year":"2021"},{"key":"ref43","first-page":"6543","article-title":"TeraPipe: Token-level pipeline parallelism for training large-scale language models","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Li","year":"2021"},{"key":"ref44","first-page":"7937","article-title":"Memory-efficient pipeline-parallel DNN training","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Narayanan","year":"2021"},{"key":"ref45","article-title":"Training deep nets with sublinear memory cost","author":"Chen","year":"2016"},{"key":"ref46","article-title":"Mixed precision training","volume-title":"Proc. Int. Conf. Learn. Representations","author":"Micikevicius","year":"2018"},{"key":"ref47","article-title":"M6-10t: A Sharing-delinking paradigm for efficient multi-trillion parameter pretraining","author":"Lin","year":"2021"},{"key":"ref48","doi-asserted-by":"publisher","DOI":"10.1109\/IPDPS53621.2022.00065"},{"key":"ref49","doi-asserted-by":"publisher","DOI":"10.1145\/3178487.3178491"},{"key":"ref50","first-page":"495","article-title":"Dorylus: Affordable, scalable, and accurate GNN training with distributed cpu servers and serverless threads","volume-title":"Proc. 15th USENIX Symp. Operating Syst. Des. Implementation","author":"Thorpe","year":"2021"},{"key":"ref51","first-page":"4911","article-title":"Adascale SGD: A user-friendly algorithm for distributed training","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Johnson","year":"2020"}],"container-title":["IEEE Transactions on Parallel and Distributed Systems"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/71\/11636249\/11636265.pdf?arnumber=11636265","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,8,31]],"date-time":"2026-08-31T19:10:37Z","timestamp":1788203437000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11636265\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,10]]},"references-count":47,"journal-issue":{"issue":"10"},"URL":"https:\/\/doi.org\/10.1109\/tpds.2026.3719351","relation":{},"ISSN":["1045-9219","1558-2183","2161-9883"],"issn-type":[{"value":"1045-9219","type":"print"},{"value":"1558-2183","type":"electronic"},{"value":"2161-9883","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,10]]}}}