{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,5]],"date-time":"2026-06-05T16:08:39Z","timestamp":1780675719584,"version":"3.54.1"},"reference-count":25,"publisher":"Institute of Electrical and Electronics Engineers (IEEE)","license":[{"start":{"date-parts":[[2024,1,1]],"date-time":"2024-01-01T00:00:00Z","timestamp":1704067200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by-nc-nd\/4.0\/"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IEEE Access"],"published-print":{"date-parts":[[2024]]},"DOI":"10.1109\/access.2024.3456788","type":"journal-article","created":{"date-parts":[[2024,9,10]],"date-time":"2024-09-10T19:21:56Z","timestamp":1725996116000},"page":"178460-178475","source":"Crossref","is-referenced-by-count":5,"title":["MSCH: Microbatch-Based Selective Activation Checkpointing With Recomputation Hidden for Efficient Training of LLM Models"],"prefix":"10.1109","volume":"12","author":[{"ORCID":"https:\/\/orcid.org\/0000-0001-5050-1899","authenticated-orcid":false,"given":"Cheng","family":"Zhang","sequence":"first","affiliation":[{"name":"Platform Capability Research and Development Department, China Mobile (Zhejiang) Innovation Research Institute Company Ltd., Hangzhou, Zhejiang, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0005-1010-0616","authenticated-orcid":false,"given":"Minjun","family":"Yu","sequence":"additional","affiliation":[{"name":"Platform Capability Research and Development Department, China Mobile (Zhejiang) Innovation Research Institute Company Ltd., Hangzhou, Zhejiang, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Li","family":"Yu","sequence":"additional","affiliation":[{"name":"Platform Capability Research and Development Department, China Mobile (Zhejiang) Innovation Research Institute Company Ltd., Hangzhou, Zhejiang, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Pengyu","family":"Cong","sequence":"additional","affiliation":[{"name":"Platform Capability Research and Development Department, China Mobile (Zhejiang) Innovation Research Institute Company Ltd., Hangzhou, Zhejiang, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yuchao","family":"Yan","sequence":"additional","affiliation":[{"name":"Platform Capability Research and Development Department, China Mobile (Zhejiang) Innovation Research Institute Company Ltd., Hangzhou, Zhejiang, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jie","family":"Bao","sequence":"additional","affiliation":[{"name":"Platform Capability Research and Development Department, China Mobile (Zhejiang) Innovation Research Institute Company Ltd., Hangzhou, Zhejiang, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jian","family":"Jiang","sequence":"additional","affiliation":[{"name":"Platform Capability Research and Development Department, China Mobile (Zhejiang) Innovation Research Institute Company Ltd., Hangzhou, Zhejiang, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Xiaozheng","family":"Wang","sequence":"additional","affiliation":[{"name":"Zhejiang Mobile Communications Company Ltd., Hangzhou, Zhejiang, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Xiaolong","family":"Ye","sequence":"additional","affiliation":[{"name":"Platform Capability Research and Development Department, China Mobile (Zhejiang) Innovation Research Institute Company Ltd., Hangzhou, Zhejiang, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Tao","family":"Tang","sequence":"additional","affiliation":[{"name":"Zhejiang Mobile Communications Company Ltd., Hangzhou, Zhejiang, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Liang","family":"Xiao","sequence":"additional","affiliation":[{"name":"Zhejiang Mobile Communications Company Ltd., Hangzhou, Zhejiang, China"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"263","reference":[{"key":"ref1","doi-asserted-by":"publisher","DOI":"10.1109\/TMM.2021.3092571"},{"key":"ref2","doi-asserted-by":"publisher","DOI":"10.1016\/j.eswa.2023.120143"},{"key":"ref3","doi-asserted-by":"publisher","DOI":"10.1109\/TCSVT.2021.3078436"},{"key":"ref4","doi-asserted-by":"publisher","DOI":"10.5539\/mas.v14n6p75"},{"key":"ref5","first-page":"1877","article-title":"Language models are few-shot learners","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"33","author":"Brown"},{"key":"ref6","article-title":"Bloom: A 176B-parameter open-access multilingual language model","author":"Workshop","year":"2022","journal-title":"arXiv:2211.05100"},{"key":"ref7","article-title":"Llama 2: Open foundation and fine-tuned chat models","author":"Touvron","year":"2023","journal-title":"arXiv:2307.09288"},{"key":"ref8","doi-asserted-by":"publisher","DOI":"10.1145\/3581784.3607102"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.1109\/SC41405.2020.00024"},{"key":"ref10","first-page":"341","article-title":"Reducing activation recomputation in large transformer models","volume-title":"Proc. Mach. Learn. Syst.","volume":"5","author":"Korthikanti"},{"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","doi-asserted-by":"publisher","DOI":"10.1109\/TPDS.2023.3247883"},{"key":"ref14","first-page":"16639","article-title":"BPIPE: Memory-balanced pipeline parallelism for training large language models","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"Kim"},{"key":"ref15","article-title":"Training deep nets with sublinear memory cost","author":"Chen","year":"2016","journal-title":"arXiv:1604.06174"},{"key":"ref16","first-page":"16344","article-title":"FlashAttention: Fast and memory-efficient exact attention with IO-awareness","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","volume":"35","author":"Dao"},{"key":"ref17","doi-asserted-by":"publisher","DOI":"10.14778\/3415478.3415530"},{"key":"ref18","first-page":"551","article-title":"ZeRO-offload: Democratizing billion-scale model training","volume-title":"Proc. USENIX Annu. Tech. Conf.","author":"Ren"},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.1145\/3458817.3476205"},{"key":"ref20","article-title":"Megatron-LM: Training multi-billion parameter language models using model parallelism","author":"Shoeybi","year":"2019","journal-title":"arXiv:1909.08053"},{"key":"ref21","article-title":"Sequence parallelism: Long sequence training from system perspective","author":"Li","year":"2021","journal-title":"arXiv:2105.13120"},{"key":"ref22","article-title":"FlashAttention-2: Faster attention with better parallelism and work partitioning","author":"Dao","year":"2023","journal-title":"arXiv:2307.08691"},{"key":"ref23","article-title":"Using DeepSpeed and megatron to train megatron-turing NLG 530B, a large-scale generative language model","author":"Smith","year":"2022","journal-title":"arXiv:2201.11990"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2022.bigscience-1.9"},{"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"}],"container-title":["IEEE Access"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/6287639\/10380310\/10670415.pdf?arnumber=10670415","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2024,12,11]],"date-time":"2024-12-11T02:16:20Z","timestamp":1733883380000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/10670415\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2024]]},"references-count":25,"URL":"https:\/\/doi.org\/10.1109\/access.2024.3456788","relation":{},"ISSN":["2169-3536"],"issn-type":[{"value":"2169-3536","type":"electronic"}],"subject":[],"published":{"date-parts":[[2024]]}}}