{"id":"https://openalex.org/W7155534315","doi":"https://doi.org/10.1145/3767295.3803590","title":"HetAuto: Cross-Cluster Auto-Parallelism for Heterogeneous Distributed Training","display_name":"HetAuto: Cross-Cluster Auto-Parallelism for Heterogeneous Distributed Training","publication_year":2026,"publication_date":"2026-04-24","ids":{"openalex":"https://openalex.org/W7155534315","doi":"https://doi.org/10.1145/3767295.3803590"},"language":null,"primary_location":{"id":"doi:10.1145/3767295.3803590","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3767295.3803590","pdf_url":null,"source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 21st European Conference on Computer Systems","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://doi.org/10.1145/3767295.3803590","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5119879247","display_name":"Guicheng Qi","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"]},{"id":"https://openalex.org/I889458895","display_name":"University of Hong Kong","ror":"https://ror.org/02zhqgq86","country_code":"HK","type":"education","lineage":["https://openalex.org/I889458895"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Guicheng Qi","raw_affiliation_strings":["Computer Science, The University of Hong Kong, Hong Kong, Hong Kong"],"raw_orcid":"https://orcid.org/0009-0005-0606-5109","affiliations":[{"raw_affiliation_string":"Computer Science, The University of Hong Kong, Hong Kong, Hong Kong","institution_ids":["https://openalex.org/I200769079","https://openalex.org/I889458895"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5019328628","display_name":"Junwei Su","orcid":"https://orcid.org/0000-0001-5185-3976"},"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"]},{"id":"https://openalex.org/I889458895","display_name":"University of Hong Kong","ror":"https://ror.org/02zhqgq86","country_code":"HK","type":"education","lineage":["https://openalex.org/I889458895"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Junwei Su","raw_affiliation_strings":["Computer Science, The University of Hong Kong, Hong Kong, Hong Kong"],"raw_orcid":"https://orcid.org/0000-0003-0537-4004","affiliations":[{"raw_affiliation_string":"Computer Science, The University of Hong Kong, Hong Kong, Hong Kong","institution_ids":["https://openalex.org/I200769079","https://openalex.org/I889458895"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134473191","display_name":"Liqi Yang","orcid":"https://orcid.org/0009-0001-0152-8304"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liqi Yang","raw_affiliation_strings":["Meituan Corporation, Beijing, China"],"raw_orcid":"https://orcid.org/0009-0001-0152-8304","affiliations":[{"raw_affiliation_string":"Meituan Corporation, Beijing, China","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134526273","display_name":"Li Tao","orcid":"https://orcid.org/0009-0002-9918-6003"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tao Li","raw_affiliation_strings":["Meituan Corporation, Beijing, China"],"raw_orcid":"https://orcid.org/0009-0002-9918-6003","affiliations":[{"raw_affiliation_string":"Meituan Corporation, Beijing, China","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134497854","display_name":"Tingwen Xie","orcid":"https://orcid.org/0009-0003-4739-9450"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tingwen Xie","raw_affiliation_strings":["Meituan Corporation, Beijing, China"],"raw_orcid":"https://orcid.org/0009-0003-4739-9450","affiliations":[{"raw_affiliation_string":"Meituan Corporation, Beijing, China","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134499574","display_name":"Yerui Sun","orcid":"https://orcid.org/0009-0003-5452-001X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yerui Sun","raw_affiliation_strings":["Meituan Corporation, Beijing, China"],"raw_orcid":"https://orcid.org/0009-0003-5452-001X","affiliations":[{"raw_affiliation_string":"Meituan Corporation, Beijing, China","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134483975","display_name":"Yuchen Xie","orcid":"https://orcid.org/0009-0003-6978-064X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yuchen Xie","raw_affiliation_strings":["Meituan Corporation, Beijing, China"],"raw_orcid":"https://orcid.org/0009-0003-6978-064X","affiliations":[{"raw_affiliation_string":"Meituan Corporation, Beijing, China","institution_ids":[]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5012597518","display_name":"Chuan Wu","orcid":"https://orcid.org/0000-0002-3144-4398"},"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"]},{"id":"https://openalex.org/I889458895","display_name":"University of Hong Kong","ror":"https://ror.org/02zhqgq86","country_code":"HK","type":"education","lineage":["https://openalex.org/I889458895"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Chuan Wu","raw_affiliation_strings":["Computer Science, The University of Hong Kong, Hong Kong, Hong Kong"],"raw_orcid":"https://orcid.org/0000-0002-3144-4398","affiliations":[{"raw_affiliation_string":"Computer Science, The University of Hong Kong, Hong Kong, Hong Kong","institution_ids":["https://openalex.org/I200769079","https://openalex.org/I889458895"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":8,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.6232962,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"759","last_page":"779"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.2281000018119812,"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"}},"topics":[{"id":"https://openalex.org/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.2281000018119812,"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/T10764","display_name":"Privacy-Preserving Technologies in Data","score":0.042399998754262924,"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/T10772","display_name":"Distributed systems and fault tolerance","score":0.040300000458955765,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"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/training","display_name":"Training (meteorology)","score":0.3625999987125397},{"id":"https://openalex.org/keywords/process","display_name":"Process (computing)","score":0.3091000020503998},{"id":"https://openalex.org/keywords/identification","display_name":"Identification (biology)","score":0.2605000138282776},{"id":"https://openalex.org/keywords/component","display_name":"Component (thermodynamics)","score":0.24060000479221344},{"id":"https://openalex.org/keywords/set","display_name":"Set (abstract data type)","score":0.23770000040531158}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5867999792098999},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.3625999987125397},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.3091000020503998},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.2685000002384186},{"id":"https://openalex.org/C116834253","wikidata":"https://www.wikidata.org/wiki/Q2039217","display_name":"Identification (biology)","level":2,"score":0.2605000138282776},{"id":"https://openalex.org/C168167062","wikidata":"https://www.wikidata.org/wiki/Q1117970","display_name":"Component (thermodynamics)","level":2,"score":0.24060000479221344},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.23770000040531158},{"id":"https://openalex.org/C115901376","wikidata":"https://www.wikidata.org/wiki/Q184199","display_name":"Automation","level":2,"score":0.2361000031232834},{"id":"https://openalex.org/C18762648","wikidata":"https://www.wikidata.org/wiki/Q42213","display_name":"Work (physics)","level":2,"score":0.23250000178813934},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.23229999840259552}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3767295.3803590","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3767295.3803590","pdf_url":null,"source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 21st European Conference on Computer Systems","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.1145/3767295.3803590","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3767295.3803590","pdf_url":null,"source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 21st European Conference on Computer Systems","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":8,"referenced_works":["https://openalex.org/W2126316555","https://openalex.org/W2133046454","https://openalex.org/W4247885026","https://openalex.org/W4285796263","https://openalex.org/W4310282800","https://openalex.org/W4401567198","https://openalex.org/W4414898584","https://openalex.org/W7133227460"],"related_works":[],"abstract_inverted_index":{"As":[0],"large":[1,56],"neural":[2],"network":[3],"models":[4],"(e.g.,":[5],"LLMs)":[6],"grow":[7],"in":[8,36],"scale,":[9],"single-cluster":[10],"resources":[11],"become":[12],"insufficient,":[13],"making":[14],"cross-cluster":[15,38,54,89],"distributed":[16],"training":[17,20,43,142],"essential.":[18],"Cross-cluster":[19],"is":[21],"challenging:":[22],"hardware":[23,33],"heterogeneity":[24],"complicates":[25],"load":[26],"balancing":[27],"and":[28,31,80,100,108,148,156],"parallelization":[29,50,78],"strategy":[30,104],"introduces":[32],"compatibility":[34],"issues":[35],"implementation;":[37],"communication":[39,90,97],"bottlenecks":[40],"severely":[41],"impact":[42],"throughput.":[44],"We":[45,119],"present":[46],"HetAuto,":[47],"an":[48,101,150],"automatic":[49],"system":[51],"for":[52,105],"efficient":[53,151],"heterogeneous":[55,86,130],"model":[57,74],"training.":[58],"HetAuto":[59,121,137],"contributes":[60],"three":[61],"key":[62],"innovations:":[63],"(1)":[64],"a":[65,70,110],"principle-guided":[66],"MCTS":[67],"algorithm":[68],"with":[69,98,126],"random":[71],"forest-enhanced":[72],"cost":[73],"that":[75,95,136],"efficiently":[76],"searches":[77],"strategies":[79],"quickly":[81],"evaluates":[82],"their":[83],"performance":[84],"under":[85],"configurations;":[87],"(2)":[88],"optimizations":[91],"including":[92],"Virtual-1F1B":[93],"scheduling":[94],"overlaps":[96],"computation":[99],"optimized":[102],"resharding":[103],"inter-stage":[106],"communication;":[107],"(3)":[109],"unified":[111],"API":[112],"enabling":[113],"seamless":[114],"integration":[115],"of":[116],"diverse":[117],"accelerators.":[118],"evaluate":[120],"across":[122],"4":[123],"different":[124],"clusters":[125],"up":[127,139],"to":[128,140],"736":[129],"devices.":[131],"The":[132],"evaluation":[133],"results":[134],"show":[135],"achieves":[138],"1.57\u00d7":[141],"throughput":[143],"improvement":[144],"over":[145],"representative":[146],"baselines,":[147],"strikes":[149],"balance":[152],"between":[153],"solution":[154],"quality":[155],"search":[157],"overhead.":[158]},"counts_by_year":[],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2026-04-25T00:00:00"}
