{"id":"https://openalex.org/W4394923484","doi":"https://doi.org/10.1145/3627703.3650085","title":"Just-In-Time Checkpointing: Low Cost Error Recovery from Deep Learning Training Failures","display_name":"Just-In-Time Checkpointing: Low Cost Error Recovery from Deep Learning Training Failures","publication_year":2024,"publication_date":"2024-04-18","ids":{"openalex":"https://openalex.org/W4394923484","doi":"https://doi.org/10.1145/3627703.3650085"},"language":"en","primary_location":{"id":"doi:10.1145/3627703.3650085","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3627703.3650085","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the Nineteenth European Conference on Computer Systems","raw_type":"proceedings-article"},"type":"article","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/A5063233370","display_name":"Tanmaey Gupta","orcid":"https://orcid.org/0009-0004-2944-1632"},"institutions":[{"id":"https://openalex.org/I4210164937","display_name":"Microsoft Research (United Kingdom)","ror":"https://ror.org/05k87vq12","country_code":"GB","type":"company","lineage":["https://openalex.org/I1290206253","https://openalex.org/I4210164937"]}],"countries":["GB"],"is_corresponding":true,"raw_author_name":"Tanmaey Gupta","raw_affiliation_strings":["Microsoft Research"],"affiliations":[{"raw_affiliation_string":"Microsoft Research","institution_ids":["https://openalex.org/I4210164937"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5026456033","display_name":"Sanjeev Krishnan","orcid":"https://orcid.org/0009-0009-0125-1956"},"institutions":[{"id":"https://openalex.org/I4210164937","display_name":"Microsoft Research (United Kingdom)","ror":"https://ror.org/05k87vq12","country_code":"GB","type":"company","lineage":["https://openalex.org/I1290206253","https://openalex.org/I4210164937"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Sanjeev Krishnan","raw_affiliation_strings":["Microsoft Research"],"affiliations":[{"raw_affiliation_string":"Microsoft Research","institution_ids":["https://openalex.org/I4210164937"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100638000","display_name":"Rituraj Kumar","orcid":null},"institutions":[{"id":"https://openalex.org/I4210164937","display_name":"Microsoft Research (United Kingdom)","ror":"https://ror.org/05k87vq12","country_code":"GB","type":"company","lineage":["https://openalex.org/I1290206253","https://openalex.org/I4210164937"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Rituraj Kumar","raw_affiliation_strings":["Microsoft Research"],"affiliations":[{"raw_affiliation_string":"Microsoft Research","institution_ids":["https://openalex.org/I4210164937"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5003509558","display_name":"Abhishek Vijeev","orcid":"https://orcid.org/0009-0006-8570-4577"},"institutions":[{"id":"https://openalex.org/I4210164937","display_name":"Microsoft Research (United Kingdom)","ror":"https://ror.org/05k87vq12","country_code":"GB","type":"company","lineage":["https://openalex.org/I1290206253","https://openalex.org/I4210164937"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Abhishek Vijeev","raw_affiliation_strings":["Microsoft Research"],"affiliations":[{"raw_affiliation_string":"Microsoft Research","institution_ids":["https://openalex.org/I4210164937"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5059874199","display_name":"Bhargav Gulavani","orcid":"https://orcid.org/0009-0003-4862-0057"},"institutions":[{"id":"https://openalex.org/I4210164937","display_name":"Microsoft Research (United Kingdom)","ror":"https://ror.org/05k87vq12","country_code":"GB","type":"company","lineage":["https://openalex.org/I1290206253","https://openalex.org/I4210164937"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Bhargav Gulavani","raw_affiliation_strings":["Microsoft Research"],"affiliations":[{"raw_affiliation_string":"Microsoft Research","institution_ids":["https://openalex.org/I4210164937"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5060166632","display_name":"Nipun Kwatra","orcid":"https://orcid.org/0000-0003-0354-6204"},"institutions":[{"id":"https://openalex.org/I4210164937","display_name":"Microsoft Research (United Kingdom)","ror":"https://ror.org/05k87vq12","country_code":"GB","type":"company","lineage":["https://openalex.org/I1290206253","https://openalex.org/I4210164937"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Nipun Kwatra","raw_affiliation_strings":["Microsoft Research"],"affiliations":[{"raw_affiliation_string":"Microsoft Research","institution_ids":["https://openalex.org/I4210164937"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5015355530","display_name":"Ramachandran Ramjee","orcid":"https://orcid.org/0000-0003-0007-6040"},"institutions":[{"id":"https://openalex.org/I4210164937","display_name":"Microsoft Research (United Kingdom)","ror":"https://ror.org/05k87vq12","country_code":"GB","type":"company","lineage":["https://openalex.org/I1290206253","https://openalex.org/I4210164937"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Ramachandran Ramjee","raw_affiliation_strings":["Microsoft Research"],"affiliations":[{"raw_affiliation_string":"Microsoft Research","institution_ids":["https://openalex.org/I4210164937"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5049023640","display_name":"Muthian Sivathanu","orcid":"https://orcid.org/0009-0004-6190-9733"},"institutions":[{"id":"https://openalex.org/I4210164937","display_name":"Microsoft Research (United Kingdom)","ror":"https://ror.org/05k87vq12","country_code":"GB","type":"company","lineage":["https://openalex.org/I1290206253","https://openalex.org/I4210164937"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Muthian Sivathanu","raw_affiliation_strings":["Microsoft Research"],"affiliations":[{"raw_affiliation_string":"Microsoft Research","institution_ids":["https://openalex.org/I4210164937"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":8,"corresponding_author_ids":["https://openalex.org/A5063233370"],"corresponding_institution_ids":["https://openalex.org/I4210164937"],"apc_list":null,"apc_paid":null,"fwci":5.5129,"has_fulltext":false,"cited_by_count":15,"citation_normalized_percentile":{"value":0.96123035,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":94,"max":100},"biblio":{"volume":null,"issue":null,"first_page":"1110","last_page":"1125"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10772","display_name":"Distributed systems and fault tolerance","score":0.9984999895095825,"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"}},"topics":[{"id":"https://openalex.org/T10772","display_name":"Distributed systems and fault tolerance","score":0.9984999895095825,"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"}},{"id":"https://openalex.org/T13553","display_name":"Age of Information Optimization","score":0.996999979019165,"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"}},{"id":"https://openalex.org/T11005","display_name":"Radiation Effects in Electronics","score":0.9908999800682068,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.8812257051467896},{"id":"https://openalex.org/keywords/overhead","display_name":"Overhead (engineering)","score":0.7799793481826782},{"id":"https://openalex.org/keywords/code","display_name":"Code (set theory)","score":0.5701773166656494},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.5177432894706726},{"id":"https://openalex.org/keywords/state","display_name":"State (computer science)","score":0.5121686458587646},{"id":"https://openalex.org/keywords/training","display_name":"Training (meteorology)","score":0.4778706431388855},{"id":"https://openalex.org/keywords/process","display_name":"Process (computing)","score":0.4676278233528137},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.4471398890018463},{"id":"https://openalex.org/keywords/distributed-computing","display_name":"Distributed computing","score":0.4414197504520416},{"id":"https://openalex.org/keywords/software","display_name":"Software","score":0.43556880950927734},{"id":"https://openalex.org/keywords/deep-learning","display_name":"Deep learning","score":0.4292181730270386},{"id":"https://openalex.org/keywords/computer-engineering","display_name":"Computer engineering","score":0.34346386790275574},{"id":"https://openalex.org/keywords/operating-system","display_name":"Operating system","score":0.29192522168159485},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.2845911681652069},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.1786629557609558},{"id":"https://openalex.org/keywords/programming-language","display_name":"Programming language","score":0.1135924756526947}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8812257051467896},{"id":"https://openalex.org/C2779960059","wikidata":"https://www.wikidata.org/wiki/Q7113681","display_name":"Overhead (engineering)","level":2,"score":0.7799793481826782},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.5701773166656494},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.5177432894706726},{"id":"https://openalex.org/C48103436","wikidata":"https://www.wikidata.org/wiki/Q599031","display_name":"State (computer science)","level":2,"score":0.5121686458587646},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.4778706431388855},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.4676278233528137},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.4471398890018463},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.4414197504520416},{"id":"https://openalex.org/C2777904410","wikidata":"https://www.wikidata.org/wiki/Q7397","display_name":"Software","level":2,"score":0.43556880950927734},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.4292181730270386},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.34346386790275574},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.29192522168159485},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.2845911681652069},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.1786629557609558},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.1135924756526947},{"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},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3627703.3650085","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3627703.3650085","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the Nineteenth European Conference on Computer Systems","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/9","display_name":"Industry, innovation and infrastructure","score":0.6299999952316284}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":21,"referenced_works":["https://openalex.org/W95608104","https://openalex.org/W1491268894","https://openalex.org/W1999900893","https://openalex.org/W2035492130","https://openalex.org/W2039631162","https://openalex.org/W2055913812","https://openalex.org/W2100970777","https://openalex.org/W2146854584","https://openalex.org/W2898046929","https://openalex.org/W2910100551","https://openalex.org/W2977720775","https://openalex.org/W2985374077","https://openalex.org/W3021124033","https://openalex.org/W3115029474","https://openalex.org/W3158296418","https://openalex.org/W3189830509","https://openalex.org/W3193156221","https://openalex.org/W4224308101","https://openalex.org/W4320854356","https://openalex.org/W4387302750","https://openalex.org/W6811340617"],"related_works":["https://openalex.org/W4375867731","https://openalex.org/W2611989081","https://openalex.org/W4230611425","https://openalex.org/W2731899572","https://openalex.org/W4294635752","https://openalex.org/W4304166257","https://openalex.org/W4383066092","https://openalex.org/W3215138031","https://openalex.org/W2056396287","https://openalex.org/W4390846322"],"abstract_inverted_index":{"Deep":[0,37],"Learning":[1],"training":[2,8,52,98,182],"jobs":[3,27,99],"process":[4],"large":[5,57,70,91,218],"amounts":[6],"of":[7,72,111,128,138,161,188,205,220],"data":[9],"using":[10,100],"many":[11,101],"GPU":[12,203],"devices,":[13],"often":[14],"running":[15],"for":[16,35,56,90,191,217],"weeks":[17],"or":[18,22],"months.":[19],"When":[20],"hardware":[21],"software":[23],"failures":[24,87,115,121],"happen,":[25,116],"these":[26],"need":[28,74],"to":[29,50,75,144],"restart,":[30],"losing":[31],"the":[32,36,79,136,159,201],"memory":[33],"state":[34,53,64,154],"Neural":[38],"Network":[39],"(DNN)":[40],"model":[41],"trained":[42],"so":[43],"far,":[44],"unless":[45],"checkpointing":[46,60,113,164,177,194,207,216],"mechanisms":[47,190],"are":[48,88],"used":[49],"save":[51],"periodically.":[54],"However,":[55],"models,":[58],"periodic":[59,215],"incurs":[61],"significant":[62],"steady":[63,153],"overhead,":[65],"and":[66,208],"during":[67],"recovery,":[68],"a":[69,108,124,145,163],"number":[71],"GPUs":[73],"redo":[76],"work":[77,129,204],"since":[78,166],"last":[80],"checkpoint.":[81],"This":[82,134,156],"is":[83,212],"especially":[84],"problematic":[85],"when":[86,114],"frequent":[89],"DNN":[92],"(such":[93],"as":[94,184,186],"Large":[95],"Language":[96],"Model)":[97],"GPUs.":[102,133,221],"In":[103],"this":[104],"paper,":[105],"we":[106],"present":[107,223],"novel":[109],"approach":[110],"just-in-time":[112,176,193,206],"which":[117],"enables":[118],"recovery":[119,140],"from":[120,141,225],"with":[122,150],"just":[123],"single":[125],"minibatch":[126],"iteration":[127],"replayed":[130],"by":[131],"all":[132],"reduces":[135],"cost":[137],"error":[139],"several":[142],"minutes":[143],"few":[146],"seconds":[147],"per":[148],"GPU,":[149],"nearly":[151],"zero":[152],"overhead.":[155],"also":[157],"avoids":[158],"guesswork":[160],"choosing":[162],"frequency":[165],"failure":[167],"rates":[168],"usually":[169],"have":[170],"high":[171],"variance.":[172],"We":[173,199,222],"discuss":[174],"how":[175],"can":[178],"be":[179],"enabled":[180],"in":[181,228],"code,":[183],"well":[185],"design":[187],"key":[189],"transparent":[192],"without":[195],"user":[196],"code":[197],"change.":[198],"analyze":[200],"wasted":[202],"show":[209],"that":[210],"it":[211],"less":[213],"than":[214],"numbers":[219],"results":[224],"our":[226],"implementation":[227],"modern":[229],"AI":[230],"cluster":[231],"infrastructure.":[232]},"counts_by_year":[{"year":2025,"cited_by_count":13},{"year":2024,"cited_by_count":2}],"updated_date":"2025-11-06T03:46:38.306776","created_date":"2025-10-10T00:00:00"}