{"id":"https://openalex.org/W7214008351","doi":"https://doi.org/10.48550/arxiv.2609.22087","title":"When Is Availability-Aware Training Worth It? A Benchmark and Empirical Study of Interruption-Resilient Optimization Under Predictable Compute Schedules","display_name":"When Is Availability-Aware Training Worth It? A Benchmark and Empirical Study of Interruption-Resilient Optimization Under Predictable Compute Schedules","publication_year":2026,"publication_date":"2026-06-22","ids":{"openalex":"https://openalex.org/W7214008351","doi":"https://doi.org/10.48550/arxiv.2609.22087"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2609.22087","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.22087","pdf_url":"https://arxiv.org/pdf/2609.22087","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"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":"text"},"volume":null,"type":"preprint","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2609.22087","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5073413136","display_name":"Subhadip Mitra","orcid":"https://orcid.org/0000-0002-7107-0343","observed_orcids":["https://orcid.org/0000-0002-7107-0343"]},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Mitra, Subhadip","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":["https://openalex.org/A5073413136"],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.4731000065803528,"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":0.4731000065803528,"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/T12535","display_name":"Machine Learning and Data Classification","score":0.1023000031709671,"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/T10101","display_name":"Cloud Computing and Resource Management","score":0.052799999713897705,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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/learning-rate-schedules","display_name":"learning rate scheduling","score":0.6150000095367432},{"id":"https://openalex.org/keywords/training-interruption","display_name":"training interruption","score":0.3779999911785126},{"id":"https://openalex.org/keywords/satellite-eclipses","display_name":"satellite eclipses","score":0.3140000104904175},{"id":"https://openalex.org/keywords/distribution-shift","display_name":"distribution shift","score":0.48399999737739563},{"id":"https://openalex.org/keywords/large-language-model-training","display_name":"large language model training","score":0.5210000276565552},{"id":"https://openalex.org/keywords/gpt-2","display_name":"GPT-2","score":0.5889999866485596},{"id":"https://openalex.org/keywords/adamw","display_name":"AdamW","score":0.5440000295639038}],"concepts":[{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.8165000081062317},{"id":"https://openalex.org/C68387754","wikidata":"https://www.wikidata.org/wiki/Q7271585","display_name":"Schedule","level":2,"score":0.7444999814033508},{"id":"https://openalex.org/C12725497","wikidata":"https://www.wikidata.org/wiki/Q810247","display_name":"Baseline (sea)","level":2,"score":0.7364000082015991},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5978000164031982},{"id":"https://openalex.org/C179254644","wikidata":"https://www.wikidata.org/wiki/Q13222844","display_name":"Moment (physics)","level":2,"score":0.5957000255584717},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.5138000249862671},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.4846000075340271},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.42800000309944153},{"id":"https://openalex.org/C120936955","wikidata":"https://www.wikidata.org/wiki/Q2155640","display_name":"Empirical research","level":2,"score":0.40619999170303345},{"id":"https://openalex.org/C112758219","wikidata":"https://www.wikidata.org/wiki/Q16038819","display_name":"Duration (music)","level":2,"score":0.385699987411499},{"id":"https://openalex.org/C48103436","wikidata":"https://www.wikidata.org/wiki/Q599031","display_name":"State (computer science)","level":2,"score":0.37439998984336853},{"id":"https://openalex.org/C2779662365","wikidata":"https://www.wikidata.org/wiki/Q5416694","display_name":"Event (particle physics)","level":2,"score":0.334199994802475},{"id":"https://openalex.org/C193254401","wikidata":"https://www.wikidata.org/wiki/Q2160088","display_name":"Robust optimization","level":2,"score":0.32359999418258667},{"id":"https://openalex.org/C2776502983","wikidata":"https://www.wikidata.org/wiki/Q690182","display_name":"Contrast (vision)","level":2,"score":0.32100000977516174},{"id":"https://openalex.org/C137836250","wikidata":"https://www.wikidata.org/wiki/Q984063","display_name":"Optimization problem","level":2,"score":0.31520000100135803},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.3075999915599823},{"id":"https://openalex.org/C2777027219","wikidata":"https://www.wikidata.org/wiki/Q1284190","display_name":"Constant (computer programming)","level":2,"score":0.3018999993801117},{"id":"https://openalex.org/C139807058","wikidata":"https://www.wikidata.org/wiki/Q352374","display_name":"Adaptation (eye)","level":2,"score":0.29739999771118164},{"id":"https://openalex.org/C147764199","wikidata":"https://www.wikidata.org/wiki/Q6865248","display_name":"Minification","level":2,"score":0.2904999852180481},{"id":"https://openalex.org/C162307627","wikidata":"https://www.wikidata.org/wiki/Q204833","display_name":"Enhanced Data Rates for GSM Evolution","level":2,"score":0.2809000015258789},{"id":"https://openalex.org/C168167062","wikidata":"https://www.wikidata.org/wiki/Q1117970","display_name":"Component (thermodynamics)","level":2,"score":0.2777000069618225},{"id":"https://openalex.org/C149782125","wikidata":"https://www.wikidata.org/wiki/Q160039","display_name":"Econometrics","level":1,"score":0.27090001106262207},{"id":"https://openalex.org/C90329073","wikidata":"https://www.wikidata.org/wiki/Q914232","display_name":"Ask price","level":2,"score":0.26420000195503235}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2609.22087","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.22087","pdf_url":"https://arxiv.org/pdf/2609.22087","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"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":"text"},{"id":"doi:10.48550/arxiv.2609.22087","doi":"https://doi.org/10.48550/arxiv.2609.22087","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2609.22087","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2609.22087","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.22087","pdf_url":"https://arxiv.org/pdf/2609.22087","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"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":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7214008351.pdf","grobid_xml":"https://content.openalex.org/works/W7214008351.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4405562219","https://openalex.org/W7154907554","https://openalex.org/W4404171822","https://openalex.org/W7147624600","https://openalex.org/W7202357449","https://openalex.org/W4225004481","https://openalex.org/W7213519283","https://openalex.org/W4414976344","https://openalex.org/W2898501133","https://openalex.org/W6928821074"],"abstract_inverted_index":{"Training":[0],"under":[1,8],"non-stationary":[2],"but":[3],"predictable":[4],"compute":[5],"availability":[6,35,56],"(satellites":[7],"eclipse,":[9],"duty-cycled":[10],"edge":[11],"devices,":[12],"power-capped":[13],"datacenters)":[14],"is":[15,60,68,86,220,242,249],"often":[16],"framed":[17],"as":[18],"needing":[19],"specialized,":[20],"availability-aware":[21,126],"optimizers.":[22],"We":[23,27,165],"test":[24],"that":[25,93,143,185],"premise.":[26],"release":[28],"OrbitTrace,":[29],"a":[30,51,61,82,119,140,193,227,229,235],"benchmark":[31],"of":[32,40,200,238],"50":[33],"physics-grounded":[34],"traces":[36],"from":[37,137,222],"SGP4":[38],"propagation":[39],"live":[41],"two-line":[42],"element":[43],"sets":[44],"across":[45,81,159,182,209],"three":[46],"orbital":[47],"regimes,":[48],"and":[49,98,116,162,184,234,246],"ask":[50],"falsifiable":[52],"question:":[53],"when":[54,75,239,247],"an":[55],"gap":[57,85],"interrupts":[58],"training,":[59],"specialized":[62],"resumption":[63],"strategy":[64],"worth":[65,243],"it,":[66],"or":[67],"competent":[69],"checkpoint-and-resume":[70],"enough?":[71],"Our":[72,224],"central":[73],"finding:":[74],"optimizer":[76,96,177,195],"state":[77,97,178],"can":[78],"be":[79,180],"preserved":[80],"gap,":[83],"the":[84,151,201,217],"essentially":[87],"free.":[88],"A":[89],"strong":[90,152,230],"checkpoint":[91],"baseline":[92,142,232],"restores":[94],"full":[95],"indexes":[99,144],"its":[100,145,244],"learning-rate":[101],"schedule":[102,146],"in":[103],"effective":[104],"(active)":[105],"time":[106],"matches":[107],"uninterrupted":[108],"training":[109],"to":[110],"within":[111],"data-ordering":[112],"noise":[113],"on":[114,118,147,204],"CIFAR-10/ResNet-18,":[115],"exactly":[117],"GPT-2/AdamW":[120],"task.":[121],"Advantages":[122],"previously":[123],"reported":[124],"for":[125,173,213],"methods,":[127],"including":[128],"our":[129],"own":[130],"three-pillar":[131],"method":[132],"AAT,":[133],"arise":[134],"almost":[135],"entirely":[136],"comparison":[138],"against":[139],"weak":[141],"wall-clock":[148],"time.":[149],"Against":[150],"baseline,":[153],"reactive":[154],"adaptation":[155],"provides":[156],"no":[157],"advantage":[158],"stationary,":[160],"optimizer-state-loss,":[161],"distribution-drift":[163],"regimes.":[164],"isolate":[166],"one":[167],"narrow":[168],"regime":[169],"where":[170,216],"it":[171,248],"helps:":[172],"large":[174],"models":[175],"whose":[176],"cannot":[179],"persisted":[181],"gaps":[183],"are":[186,226],"interrupted":[187],"by":[188],"frequent,":[189],"short":[190],"pauses,":[191],"reconstructing":[192],"decayed":[194,218],"moment":[196,219],"recovers":[197],"only":[198],"~21%":[199],"state-loss":[202],"penalty":[203],"average":[205],"(and":[206],"not":[207],"robustly":[208],"seeds);":[210],"this":[211],"vanishes":[212],"eclipse-scale":[214],"gaps,":[215],"indistinguishable":[221],"zero.":[223],"contributions":[225],"benchmark,":[228],"reproducible":[231],"protocol,":[233],"clear":[236],"characterization":[237],"interruption-resilient":[240],"optimization":[241],"complexity,":[245],"not.":[250]},"counts_by_year":[],"updated_date":"2026-10-28T10:33:24.094632","created_date":"2026-09-23T00:00:00"}