{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,1,17]],"date-time":"2026-01-17T18:58:42Z","timestamp":1768676322091,"version":"3.49.0"},"reference-count":58,"publisher":"Institute of Electrical and Electronics Engineers (IEEE)","issue":"4","license":[{"start":{"date-parts":[[2024,4,1]],"date-time":"2024-04-01T00:00:00Z","timestamp":1711929600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/ieeexplore.ieee.org\/Xplorehelp\/downloads\/license-information\/IEEE.html"},{"start":{"date-parts":[[2024,4,1]],"date-time":"2024-04-01T00:00:00Z","timestamp":1711929600000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2024,4,1]],"date-time":"2024-04-01T00:00:00Z","timestamp":1711929600000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"funder":[{"name":"National Key R&D Program of China","award":["2021YFB0301200"],"award-info":[{"award-number":["2021YFB0301200"]}]},{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["62025208"],"award-info":[{"award-number":["62025208"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["61972409"],"award-info":[{"award-number":["61972409"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"name":"Xiangjiang Laboratory Project","award":["22XJ01012"],"award-info":[{"award-number":["22XJ01012"]}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IEEE Trans. Parallel Distrib. Syst."],"published-print":{"date-parts":[[2024,4]]},"DOI":"10.1109\/tpds.2023.3343570","type":"journal-article","created":{"date-parts":[[2023,12,15]],"date-time":"2023-12-15T19:55:21Z","timestamp":1702670121000},"page":"577-591","source":"Crossref","is-referenced-by-count":8,"title":["A Memory-Efficient Hybrid Parallel Framework for Deep Neural Network Training"],"prefix":"10.1109","volume":"35","author":[{"ORCID":"https:\/\/orcid.org\/0000-0001-9743-2034","authenticated-orcid":false,"given":"Dongsheng","family":"Li","sequence":"first","affiliation":[{"name":"National Key Laboratory of Parallel and Distributed Computing, College of Computer, National University of Defense Technology, Changsha, China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-7419-1511","authenticated-orcid":false,"given":"Shengwei","family":"Li","sequence":"additional","affiliation":[{"name":"National Key Laboratory of Parallel and Distributed Computing, College of Computer, National University of Defense Technology, Changsha, China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-3458-4732","authenticated-orcid":false,"given":"Zhiquan","family":"Lai","sequence":"additional","affiliation":[{"name":"National Key Laboratory of Parallel and Distributed Computing, College of Computer, National University of Defense Technology, Changsha, China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-7564-5239","authenticated-orcid":false,"given":"Yongquan","family":"Fu","sequence":"additional","affiliation":[{"name":"National Key Laboratory of Parallel and Distributed Computing, College of Computer, National University of Defense Technology, Changsha, China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0009-0008-1285-5176","authenticated-orcid":false,"given":"Xiangyu","family":"Ye","sequence":"additional","affiliation":[{"name":"National Key Laboratory of Parallel and Distributed Computing, College of Computer, National University of Defense Technology, Changsha, China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0009-0007-6965-2268","authenticated-orcid":false,"given":"Lei","family":"Cai","sequence":"additional","affiliation":[{"name":"National Key Laboratory of Parallel and Distributed Computing, College of Computer, National University of Defense Technology, Changsha, China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-8285-2738","authenticated-orcid":false,"given":"Linbo","family":"Qiao","sequence":"additional","affiliation":[{"name":"National Key Laboratory of Parallel and Distributed Computing, College of Computer, National University of Defense Technology, Changsha, China"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"263","reference":[{"key":"ref1","first-page":"1","article-title":"Improving language understanding by generative pre-training","author":"Radford","year":"2018"},{"issue":"8","key":"ref2","first-page":"9","article-title":"Language models are unsupervised multitask learners","volume":"1","author":"Radford","year":"2019","journal-title":"OpenAI Blog"},{"key":"ref3","first-page":"1877","article-title":"Language models are few-shot learners","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Brown"},{"key":"ref4","doi-asserted-by":"publisher","DOI":"10.1145\/2959100.2959190"},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.1145\/3219819.3219869"},{"key":"ref6","doi-asserted-by":"publisher","DOI":"10.5555\/2685048.2685095"},{"key":"ref7","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v29i1.9195"},{"key":"ref8","article-title":"Horovod: Fast and easy distributed deep learning in tensorflow","author":"Sergeev","year":"2018"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.1145\/2901318.2901331"},{"key":"ref10","first-page":"1223","article-title":"Large scale distributed deep networks","volume-title":"Proc. 25th Int. Conf. Neural Inf. Process. Syst.","author":"Dean"},{"key":"ref11","doi-asserted-by":"publisher","DOI":"10.1145\/3472456.3472497"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.1145\/3341301.3359646"},{"key":"ref13","doi-asserted-by":"publisher","DOI":"10.48550\/arxiv.1811.06965"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.1109\/CBD.2019.00020"},{"key":"ref15","first-page":"269","article-title":"PipeMare: Asynchronous pipeline parallel DNN training","volume-title":"Proc. Mach. Learn. Syst.","volume":"3","author":"Yang"},{"key":"ref16","doi-asserted-by":"publisher","DOI":"10.1145\/3410463.3414656"},{"key":"ref17","first-page":"587","article-title":"VersaPipe: A versatile programming framework for pipelined computing on GPU","volume-title":"Proc. IEEE\/ACM 50th Annu. Int. Symp. Microarchitecture","author":"Zheng"},{"key":"ref18","first-page":"2834","article-title":"On model parallelization and scheduling strategies for distributed machine learning","volume-title":"Proc. 27th Int. Conf. Neural Inf. Process. Syst.","author":"Lee"},{"key":"ref19","article-title":"Efficient and robust parallel DNN training through model parallelism on multi-GPU platform","author":"Chen","year":"2018"},{"key":"ref20","first-page":"7937","article-title":"Memory-efficient pipeline-parallel DNN training","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Narayanan"},{"key":"ref21","doi-asserted-by":"publisher","DOI":"10.1145\/3437801.3441593"},{"key":"ref22","doi-asserted-by":"publisher","DOI":"10.1109\/TPDS.2023.3247001"},{"key":"ref23","article-title":"Training deep nets with sublinear memory cost","author":"Chen","year":"2016"},{"key":"ref24","first-page":"497","article-title":"Checkmate: Breaking the memory wall with optimal tensor rematerialization","volume-title":"Proc. Mach. Learn. Syst.","volume":"2","author":"Jain"},{"key":"ref25","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"},{"key":"ref26","doi-asserted-by":"publisher","DOI":"10.1109\/CLUSTER51413.2022.00042"},{"key":"ref27","first-page":"183","article-title":"Pipelined model parallelism: Complexity results and memory considerations","volume-title":"Proc. Parallel Process. 27th Int. Conf. Parallel Distrib. Comput.","author":"Beaumont"},{"key":"ref28","first-page":"6183","article-title":"Asynchronous coordinate descent under more realistic assumptions","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Sun"},{"key":"ref29","article-title":"Very deep convolutional networks for large-scale image recognition","author":"Simonyan","year":"2014"},{"key":"ref30","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v33i01.33014780"},{"key":"ref31","doi-asserted-by":"publisher","DOI":"10.1007\/s11263-015-0816-y"},{"key":"ref32","article-title":"Googles neural machine translation system: Bridging the gap between human and machine translation","author":"Wu","year":"2016"},{"key":"ref33","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/W16-2323"},{"key":"ref34","article-title":"NCCL","year":"2019"},{"key":"ref35","doi-asserted-by":"publisher","DOI":"10.1145\/3394486.3406703"},{"key":"ref36","doi-asserted-by":"publisher","DOI":"10.1109\/SC41405.2020.00024"},{"key":"ref37","doi-asserted-by":"publisher","DOI":"10.1145\/3458817.3476205"},{"key":"ref38","doi-asserted-by":"publisher","DOI":"10.1109\/TPDS.2022.3219819"},{"key":"ref39","doi-asserted-by":"publisher","DOI":"10.1006\/jpdc.1994.1085"},{"issue":"1","key":"ref40","first-page":"1235","article-title":"MLlib: Machine learning in apache spark","volume":"17","author":"Meng","year":"2016","journal-title":"J. Mach. Learn. Res."},{"key":"ref41","article-title":"Accurate, large minibatch SGD: Training imagenet in 1 hour","author":"Goyal","year":"2017"},{"key":"ref42","first-page":"3043","article-title":"Asynchronous decentralized parallel stochastic gradient descent","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Lian"},{"key":"ref43","first-page":"8056","article-title":"Pipe-SGD: A decentralized pipelined SGD framework for distributed deep net training","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Li"},{"key":"ref44","first-page":"8024","article-title":"PyTorch: An imperative style, high-performance deep learning library","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"32","author":"Paszke","year":"2019"},{"key":"ref45","first-page":"265","article-title":"TensorFlow: Large-scale machine learning on heterogeneous systems","volume-title":"Proc. 12th USENIX Conf. Operating Syst. Des. Implementation","author":"Abadi"},{"key":"ref46","first-page":"693","article-title":"HOGWILD!: A lock-free approach to parallelizing stochastic gradient descent","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Recht"},{"key":"ref47","first-page":"1223","article-title":"More effective distributed ML via a stale synchronous parallel parameter server","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Ho"},{"key":"ref48","first-page":"37","article-title":"Exploiting bounded staleness to speed up Big Data analytics","volume-title":"Proc. USENIX Annu. Tech. Conf.","author":"Cui"},{"key":"ref49","article-title":"One weird trick for parallelizing convolutional neural networks","author":"Krizhevsky","year":"2014"},{"key":"ref50","first-page":"1","article-title":"Beyond data and model parallelism for deep neural networks","volume-title":"Proc. Mach. Learn. Syst.","author":"Jia"},{"key":"ref51","first-page":"571","article-title":"Project adam: Building an efficient and scalable deep learning training system","volume-title":"Proc. 11th USENIX Symp. Operating Syst. Des. Implementation","author":"Chilimbi"},{"key":"ref52","doi-asserted-by":"publisher","DOI":"10.1109\/MM.2019.2935967"},{"key":"ref53","first-page":"307","article-title":"HetPipe: Enabling large DNN training on (whimpy) heterogeneous GPU clusters through integration of pipelined model parallelism and data parallelism","volume-title":"Proc. USENIX Annu. Tech. Conf.","author":"Park"},{"key":"ref54","first-page":"559","article-title":"Alpa: Automating inter-and intra-operator parallelism for distributed deep learning","volume-title":"Proc. 16th USENIX Symp. Operating Syst. Des. Implementation","author":"Zheng"},{"key":"ref55","first-page":"4150","article-title":"PipeTransformer: Automated elastic pipelining for distributed training of large-scale models","volume-title":"Proc. 38th Int. Conf. Mach. Learn.","author":"He"},{"key":"ref56","doi-asserted-by":"publisher","DOI":"10.1145\/3458817.3476145"},{"key":"ref57","doi-asserted-by":"publisher","DOI":"10.1109\/TPDS.2023.3247883"},{"key":"ref58","doi-asserted-by":"publisher","DOI":"10.1145\/3572848.3577484"}],"container-title":["IEEE Transactions on Parallel and Distributed Systems"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx7\/71\/10438218\/10361608.pdf?arnumber=10361608","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2024,3,14]],"date-time":"2024-03-14T00:21:41Z","timestamp":1710375701000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/10361608\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2024,4]]},"references-count":58,"journal-issue":{"issue":"4"},"URL":"https:\/\/doi.org\/10.1109\/tpds.2023.3343570","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":[[2024,4]]}}}