{"id":"https://openalex.org/W4401508784","doi":"https://doi.org/10.1109/infocom52122.2024.10621327","title":"Parm: Efficient Training of Large Sparsely-Activated Models with Dedicated Schedules","display_name":"Parm: Efficient Training of Large Sparsely-Activated Models with Dedicated Schedules","publication_year":2024,"publication_date":"2024-05-20","ids":{"openalex":"https://openalex.org/W4401508784","doi":"https://doi.org/10.1109/infocom52122.2024.10621327"},"language":"en","primary_location":{"id":"doi:10.1109/infocom52122.2024.10621327","is_oa":false,"landing_page_url":"https://doi.org/10.1109/infocom52122.2024.10621327","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE INFOCOM 2024 - IEEE Conference on Computer Communications","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":false,"oa_status":"closed","oa_url":null,"any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5101238767","display_name":"Xinglin Pan","orcid":null},"institutions":[{"id":"https://openalex.org/I200769079","display_name":"Hong Kong University of Science and Technology","ror":"https://ror.org/00q4vv597","country_code":"HK","type":"education","lineage":["https://openalex.org/I200769079"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Xinglin Pan","raw_affiliation_strings":["The Hong Kong University of Science and Technology (Guangzhou),Data Science and Analytics Thrust"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The Hong Kong University of Science and Technology (Guangzhou),Data Science and Analytics Thrust","institution_ids":["https://openalex.org/I200769079"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5110841034","display_name":"Wenxiang Lin","orcid":"https://orcid.org/0000-0001-5147-0844"},"institutions":[{"id":"https://openalex.org/I204983213","display_name":"Harbin Institute of Technology","ror":"https://ror.org/01yqg2h08","country_code":"CN","type":"education","lineage":["https://openalex.org/I204983213"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Wenxiang Lin","raw_affiliation_strings":["Harbin Institute of Technology,School of Computer Science and Technology,Shenzhen"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Harbin Institute of Technology,School of Computer Science and Technology,Shenzhen","institution_ids":["https://openalex.org/I204983213"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5016836702","display_name":"Shaohuai Shi","orcid":"https://orcid.org/0000-0002-1418-5160"},"institutions":[{"id":"https://openalex.org/I204983213","display_name":"Harbin Institute of Technology","ror":"https://ror.org/01yqg2h08","country_code":"CN","type":"education","lineage":["https://openalex.org/I204983213"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Shaohuai Shi","raw_affiliation_strings":["Harbin Institute of Technology,School of Computer Science and Technology,Shenzhen"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Harbin Institute of Technology,School of Computer Science and Technology,Shenzhen","institution_ids":["https://openalex.org/I204983213"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100730785","display_name":"Xiaowen Chu","orcid":"https://orcid.org/0000-0001-9745-4372"},"institutions":[{"id":"https://openalex.org/I200769079","display_name":"Hong Kong University of Science and Technology","ror":"https://ror.org/00q4vv597","country_code":"HK","type":"education","lineage":["https://openalex.org/I200769079"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Xiaowen Chu","raw_affiliation_strings":["The Hong Kong University of Science and Technology (Guangzhou),Data Science and Analytics Thrust"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The Hong Kong University of Science and Technology (Guangzhou),Data Science and Analytics Thrust","institution_ids":["https://openalex.org/I200769079"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5103761974","display_name":"Weinong Sun","orcid":null},"institutions":[{"id":"https://openalex.org/I200769079","display_name":"Hong Kong University of Science and Technology","ror":"https://ror.org/00q4vv597","country_code":"HK","type":"education","lineage":["https://openalex.org/I200769079"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Weinong Sun","raw_affiliation_strings":["The Hong Kong University of Science and Technology,Department of Computer Science and Engineering"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The Hong Kong University of Science and Technology,Department of Computer Science and Engineering","institution_ids":["https://openalex.org/I200769079"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5100346957","display_name":"Bo Li","orcid":"https://orcid.org/0000-0002-7848-790X"},"institutions":[{"id":"https://openalex.org/I200769079","display_name":"Hong Kong University of Science and Technology","ror":"https://ror.org/00q4vv597","country_code":"HK","type":"education","lineage":["https://openalex.org/I200769079"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Bo Li","raw_affiliation_strings":["The Hong Kong University of Science and Technology,Department of Computer Science and Engineering"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The Hong Kong University of Science and Technology,Department of Computer Science and Engineering","institution_ids":["https://openalex.org/I200769079"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":10,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"1880","last_page":"1889"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.9952999949455261,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.9952999949455261,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T12072","display_name":"Machine Learning and Algorithms","score":0.991599977016449,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.9901000261306763,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.7045108079910278},{"id":"https://openalex.org/keywords/training","display_name":"Training (meteorology)","score":0.5659998059272766}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7045108079910278},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.5659998059272766},{"id":"https://openalex.org/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"score":0.0},{"id":"https://openalex.org/C153294291","wikidata":"https://www.wikidata.org/wiki/Q25261","display_name":"Meteorology","level":1,"score":0.0}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.1109/infocom52122.2024.10621327","is_oa":false,"landing_page_url":"https://doi.org/10.1109/infocom52122.2024.10621327","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE INFOCOM 2024 - IEEE Conference on Computer Communications","raw_type":"proceedings-article"},{"id":"pmh:oai:repository.hkust.edu.hk:1783.1-139128","is_oa":false,"landing_page_url":"http://repository.hkust.edu.hk/ir/Record/1783.1-139128","pdf_url":null,"source":{"id":"https://openalex.org/S4306401796","display_name":"Rare & Special e-Zone (The Hong Kong University of Science and Technology)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I200769079","host_organization_name":"Hong Kong University of Science and Technology","host_organization_lineage":["https://openalex.org/I200769079"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Conference paper"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"}],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":42,"referenced_works":["https://openalex.org/W1572016165","https://openalex.org/W2105549957","https://openalex.org/W2166221608","https://openalex.org/W2168231600","https://openalex.org/W2767510344","https://openalex.org/W2896457183","https://openalex.org/W2985108934","https://openalex.org/W2991040477","https://openalex.org/W3001279689","https://openalex.org/W4220838824","https://openalex.org/W4220967350","https://openalex.org/W4221150592","https://openalex.org/W4224308101","https://openalex.org/W4226079124","https://openalex.org/W4281399297","https://openalex.org/W4287391717","https://openalex.org/W4292779060","https://openalex.org/W4308426193","https://openalex.org/W4321636575","https://openalex.org/W4323572061","https://openalex.org/W4376652719","https://openalex.org/W4386260498","https://openalex.org/W6684859321","https://openalex.org/W6732520560","https://openalex.org/W6772383348","https://openalex.org/W6778883912","https://openalex.org/W6780805062","https://openalex.org/W6788811087","https://openalex.org/W6793102544","https://openalex.org/W6799372109","https://openalex.org/W6802290083","https://openalex.org/W6810081322","https://openalex.org/W6810737565","https://openalex.org/W6810764171","https://openalex.org/W6811726652","https://openalex.org/W6811928498","https://openalex.org/W6838284419","https://openalex.org/W6839092827","https://openalex.org/W6839827798","https://openalex.org/W6846390784","https://openalex.org/W6850503672","https://openalex.org/W6851281298"],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2748952813","https://openalex.org/W230091440","https://openalex.org/W2390279801","https://openalex.org/W2233261550","https://openalex.org/W2358668433","https://openalex.org/W4396701345","https://openalex.org/W2810751659","https://openalex.org/W258997015","https://openalex.org/W2376932109"],"abstract_inverted_index":{"Sparsely-activated":[0],"Mixture-of-Expert":[1],"(MoE)":[2],"layers":[3,168],"have":[4],"found":[5],"practical":[6],"applications":[7],"in":[8,22,135],"enlarging":[9],"the":[10,26,51,104,109,152],"model":[11,34,46],"size":[12],"of":[13,29],"large-scale":[14],"foundation":[15],"models,":[16],"with":[17],"only":[18],"a":[19,71,145],"sub-linear":[20],"increase":[21],"computation":[23],"demands.":[24],"Despite":[25],"wide":[27],"adoption":[28],"hybrid":[30],"parallel":[31,62],"paradigms":[32],"like":[33],"parallelism,":[35,37],"expert":[36],"and":[38,92,94,99,121,125,144,169,181],"expert-sharding":[39],"parallelism":[40],"(i.e.,":[41],"MP+EP+ESP)":[42],"to":[43,128,160],"support":[44],"MoE":[45,154,167,176],"training":[47,52,76,106,155],"on":[48,140,163,173,179],"GPU":[49],"clusters,":[50],"efficiency":[53],"is":[54],"hindered":[55],"by":[56,60,77],"communication":[57,84],"costs":[58],"introduced":[59],"these":[61],"paradigms.":[63],"To":[64],"address":[65],"this":[66],"limitation,":[67],"we":[68,116],"propose":[69],"Parm,":[70],"system":[72],"that":[73,149],"accelerates":[74],"MP+EP+ESP":[75],"designing":[78],"two":[79,110,174],"dedicated":[80],"schedules":[81,88,111],"for":[82],"placing":[83],"tasks.":[85],"The":[86],"proposed":[87],"eliminate":[89],"redundant":[90],"computations":[91],"communications":[93],"enable":[95],"overlaps":[96],"between":[97],"intra-node":[98],"inter-node":[100],"communications,":[101],"ultimately":[102],"reducing":[103],"overall":[105],"time.":[107],"As":[108],"are":[112],"not":[113],"mutually":[114],"exclusive,":[115],"provide":[117],"comprehensive":[118],"theoretical":[119],"analyses":[120],"derive":[122],"an":[123,141],"automatic":[124],"accurate":[126],"solution":[127],"determine":[129],"which":[130],"schedule":[131],"should":[132],"be":[133],"applied":[134],"different":[136],"scenarios.":[137],"Experimental":[138],"results":[139],"8-GPU":[142],"server":[143],"32-GPU":[146],"cluster":[147],"demonstrate":[148],"Parm":[150],"outperforms":[151],"state-of-the-art":[153],"system,":[156],"DeepSpeed-MoE,":[157],"achieving":[158],"1.13\u00d7":[159],"5.77\u00d7":[161],"speedup":[162],"1296":[164],"manually":[165],"configured":[166],"approximately":[170],"3\u00d7":[171],"improvement":[172],"real-world":[175],"models":[177],"based":[178],"BERT":[180],"GPT-2.":[182]},"counts_by_year":[{"year":2026,"cited_by_count":6},{"year":2025,"cited_by_count":4}],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}