{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,1,9]],"date-time":"2026-01-09T19:32:17Z","timestamp":1767987137440,"version":"3.49.0"},"reference-count":56,"publisher":"Institute of Electrical and Electronics Engineers (IEEE)","issue":"6","license":[{"start":{"date-parts":[[2025,6,1]],"date-time":"2025-06-01T00:00:00Z","timestamp":1748736000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/ieeexplore.ieee.org\/Xplorehelp\/downloads\/license-information\/IEEE.html"},{"start":{"date-parts":[[2025,6,1]],"date-time":"2025-06-01T00:00:00Z","timestamp":1748736000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2025,6,1]],"date-time":"2025-06-01T00:00:00Z","timestamp":1748736000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"funder":[{"name":"Industry Collaboration Projects (IAF-ICP) Funding Initiative"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["IEEE Trans. Parallel Distrib. Syst."],"published-print":{"date-parts":[[2025,6]]},"DOI":"10.1109\/tpds.2025.3553137","type":"journal-article","created":{"date-parts":[[2025,3,21]],"date-time":"2025-03-21T19:30:46Z","timestamp":1742585446000},"page":"1071-1086","source":"Crossref","is-referenced-by-count":1,"title":["IceFrog: A Layer-Elastic Scheduling System for Deep Learning Training in GPU Clusters"],"prefix":"10.1109","volume":"36","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-7048-1722","authenticated-orcid":false,"given":"Wei","family":"Gao","sequence":"first","affiliation":[{"name":"College of Computational and Data Science, Nanyang Technological University, Singapore"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0009-0009-8998-8628","authenticated-orcid":false,"given":"Zhuoyuan","family":"Ouyang","sequence":"additional","affiliation":[{"name":"College of Computational and Data Science, Nanyang Technological University, Singapore"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-8456-0491","authenticated-orcid":false,"given":"Peng","family":"Sun","sequence":"additional","affiliation":[{"name":"Shanghai AI Lab & Sensetime, Shanghai, China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-6595-6650","authenticated-orcid":false,"given":"Tianwei","family":"Zhang","sequence":"additional","affiliation":[{"name":"College of Computational and Data Science, Nanyang Technological University, Singapore"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-2751-5114","authenticated-orcid":false,"given":"Yonggang","family":"Wen","sequence":"additional","affiliation":[{"name":"College of Computational and Data Science, Nanyang Technological University, Singapore"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"263","reference":[{"key":"ref1","first-page":"595","article-title":"Gandiva: Introspective cluster scheduling for deep learning","volume-title":"Proc. USENIX Symp. Operating Syst. Des. Implementation","author":"Xiao"},{"key":"ref2","doi-asserted-by":"publisher","DOI":"10.1145\/3575693.3575705"},{"key":"ref3","doi-asserted-by":"publisher","DOI":"10.1145\/3190508.3190517"},{"key":"ref4","article-title":"Pollux: Co-adaptive cluster scheduling for goodput-optimized deep learning","volume-title":"Proc. USENIX Symp. Operating Syst. Des. Implementation","author":"Qiao"},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.1145\/3458817.3480859"},{"key":"ref6","doi-asserted-by":"publisher","DOI":"10.1145\/3650200.3656599"},{"key":"ref7","doi-asserted-by":"publisher","DOI":"10.1145\/3542929.3563460"},{"key":"ref8","first-page":"721","article-title":"Elastic resource sharing for distributed deep learning","volume-title":"Proc. USENIX Symp. Netw. Syst. Des. Implementation","author":"Hwang"},{"key":"ref9","first-page":"4150","article-title":"Pipetransformer: Automated elastic pipelining for distributed training of large-scale models","volume-title":"Proc. Int. Conf. Mach. Learn.","author":"He"},{"key":"ref10","article-title":"Autofreeze: Automatically freezing model blocks to accelerate fine-tuning","author":"Liu","year":"2021"},{"key":"ref11","article-title":"Freezeout: Accelerate training by progressively freezing layers","author":"Brock","year":"2017"},{"key":"ref12","first-page":"6078","article-title":"SVCCA: Singular vector canonical correlation analysis for deep learning dynamics and interpretability","volume-title":"Proc. Int. Conf. Neural Inf. Process. Syst.","author":"Raghu"},{"key":"ref13","first-page":"5732","article-title":"Insights on representational similarity in neural networks with canonical correlation","volume-title":"Proc. Int. Conf. Neural Inf. Process. Syst.","author":"Morcos"},{"key":"ref14","article-title":"Efficient self-supervised continual learning with progressive task-correlated layer freezing","author":"Yang","year":"2023"},{"key":"ref15","doi-asserted-by":"publisher","DOI":"10.1145\/3552326.3587451"},{"key":"ref16","doi-asserted-by":"publisher","DOI":"10.1201\/9781003162810-13"},{"key":"ref17","doi-asserted-by":"publisher","DOI":"10.1109\/TPDS.2021.3079202"},{"key":"ref18","first-page":"945","article-title":"MLaaS in the wild: Workload analysis and scheduling in Large-Scale heterogeneous GPU clusters","volume-title":"Proc. USENIX Symp. Netw. Syst. Des. Implementation","author":"Weng"},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.1109\/TPDS.2021.3064966"},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.14778\/3570690.3570697"},{"key":"ref21","doi-asserted-by":"publisher","DOI":"10.1109\/SC41405.2020.00024"},{"key":"ref22","doi-asserted-by":"publisher","DOI":"10.14778\/3611540.3611569"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.48550\/arxiv.1811.06965"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.1145\/3341301.3359646"},{"key":"ref25","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1810.04805"},{"key":"ref26","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2016.91"},{"key":"ref27","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2018.00474"},{"key":"ref28","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2016.90"},{"key":"ref29","article-title":"Very deep convolutional networks for large-scale image recognition","author":"Simonyan","year":"2014"},{"key":"ref30","doi-asserted-by":"publisher","DOI":"10.4324\/9781410605337-29"},{"key":"ref31","article-title":"Learning multiple layers of features from tiny images","author":"Krizhevsky","year":"2012"},{"key":"ref32","doi-asserted-by":"publisher","DOI":"10.1007\/s11263-009-0275-4"},{"key":"ref33","doi-asserted-by":"publisher","DOI":"10.1109\/CVPR.2009.5206848"},{"key":"ref34","first-page":"947","article-title":"Analysis of large-scale multi-tenant GPU clusters for DNN training workloads","volume-title":"Proc. USENIX Annu. Techn. Conf.","author":"Jeon"},{"key":"ref35","doi-asserted-by":"publisher","DOI":"10.1109\/iThings\/GreenCom\/CPSCom\/SmartData.2019.00205"},{"key":"ref36","article-title":"Layer freezing & data sieving: Missing pieces of a generic framework for sparse training","author":"Yuan","year":"2022"},{"key":"ref37","doi-asserted-by":"publisher","DOI":"10.1145\/3472883.3486978"},{"key":"ref38","doi-asserted-by":"publisher","DOI":"10.1109\/TC.2024.3371794"},{"key":"ref39","article-title":"Fairscale: A general purpose modular pytorch library for high performance and large scale training","year":"2021"},{"key":"ref40","article-title":"NCCL","year":"2022"},{"key":"ref41","first-page":"6630","article-title":"AMP: Automatically finding model parallel strategies with heterogeneity awareness","volume-title":"Proc. Adv. Neural Inf. Process. Syst.","author":"Li"},{"key":"ref42","doi-asserted-by":"publisher","DOI":"10.1145\/3373376.3378525"},{"key":"ref43","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/P18-2124"},{"key":"ref44","first-page":"1631","article-title":"Recursive deep models for semantic compositionality over a sentiment treebank","volume-title":"Proc. 2013 Conf. Empirical Methods Natural Lang. Process.","author":"Socher"},{"key":"ref45","first-page":"703","article-title":"Shockwave: Fair and efficient cluster scheduling for dynamic adaptation in machine learning","volume-title":"Proc. USENIX Symp. Netw. Syst. Des. Implementation","author":"Zheng"},{"key":"ref46","article-title":"Naruto blip captions","author":"Cervenka","year":"2022"},{"key":"ref47","article-title":"Denoising diffusion probabilistic models","author":"Ho","year":"2020"},{"key":"ref48","article-title":"Pointer sentinel mixture models","author":"Merity","year":"2016"},{"key":"ref49","first-page":"481","article-title":"Heterogeneity-aware cluster scheduling policies for deep learning workloads","volume-title":"Proc. 14th USENIX Symp. Operating Syst. Des. Implementation","author":"Narayanan"},{"key":"ref50","doi-asserted-by":"publisher","DOI":"10.1145\/3477132.3483588"},{"key":"ref51","first-page":"485","article-title":"Tiresias: A GPU cluster manager for distributed deep learning","volume-title":"Proc. USENIX Symp. Netw. Syst. Des. Implementation","author":"Gu"},{"key":"ref52","first-page":"515","article-title":"Hived: Sharing a GPU cluster for deep learning with guarantees","volume-title":"Proc. USENIX Symp. Operating Syst. Des. Implementation","author":"Zhao"},{"key":"ref53","doi-asserted-by":"publisher","DOI":"10.1145\/3650200.3656598"},{"key":"ref54","doi-asserted-by":"publisher","DOI":"10.1145\/3656012"},{"key":"ref55","doi-asserted-by":"publisher","DOI":"10.1145\/3600006.3613175"},{"key":"ref56","doi-asserted-by":"publisher","DOI":"10.1007\/s42979-020-00312-x"}],"container-title":["IEEE Transactions on Parallel and Distributed Systems"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/71\/10969501\/10935732.pdf?arnumber=10935732","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,4,18]],"date-time":"2025-04-18T17:39:00Z","timestamp":1744997940000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/10935732\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,6]]},"references-count":56,"journal-issue":{"issue":"6"},"URL":"https:\/\/doi.org\/10.1109\/tpds.2025.3553137","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":[[2025,6]]}}}