{"id":"https://openalex.org/W4417120068","doi":"https://doi.org/10.48550/arxiv.2510.09103","title":"AdaPM: a Partial Momentum Algorithm for LLM Training","display_name":"AdaPM: a Partial Momentum Algorithm for LLM Training","publication_year":2025,"publication_date":"2025-10-10","ids":{"openalex":"https://openalex.org/W4417120068","doi":"https://doi.org/10.48550/arxiv.2510.09103"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2510.09103","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2510.09103","pdf_url":"https://arxiv.org/pdf/2510.09103","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"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"},"type":"preprint","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2510.09103","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":null,"display_name":"Zhang, Yimu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Yimu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5056367195","display_name":"Yuanshi Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Yuanshi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5101691432","display_name":"Cong Fang","orcid":"https://orcid.org/0000-0002-3821-0084"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Fang, Cong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"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/T12535","display_name":"Machine Learning and Data Classification","score":0.28049999475479126,"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/T12535","display_name":"Machine Learning and Data Classification","score":0.28049999475479126,"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/T13650","display_name":"Computational Physics and Python Applications","score":0.07970000058412552,"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/T11948","display_name":"Machine Learning in Materials Science","score":0.07249999791383743,"subfield":{"id":"https://openalex.org/subfields/2505","display_name":"Materials Chemistry"},"field":{"id":"https://openalex.org/fields/25","display_name":"Materials Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/momentum","display_name":"Momentum (technical analysis)","score":0.8203999996185303},{"id":"https://openalex.org/keywords/training","display_name":"Training (meteorology)","score":0.4796999990940094},{"id":"https://openalex.org/keywords/ranging","display_name":"Ranging","score":0.31209999322891235}],"concepts":[{"id":"https://openalex.org/C60718061","wikidata":"https://www.wikidata.org/wiki/Q1414747","display_name":"Momentum (technical analysis)","level":2,"score":0.8203999996185303},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5544000267982483},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.48159998655319214},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.4796999990940094},{"id":"https://openalex.org/C115051666","wikidata":"https://www.wikidata.org/wiki/Q6522493","display_name":"Ranging","level":2,"score":0.31209999322891235},{"id":"https://openalex.org/C44154836","wikidata":"https://www.wikidata.org/wiki/Q45045","display_name":"Simulation","level":1,"score":0.2696000039577484},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.2678000032901764},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.26570001244544983},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.25119999051094055},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.23479999601840973}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2510.09103","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2510.09103","pdf_url":"https://arxiv.org/pdf/2510.09103","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"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.2510.09103","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2510.09103","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2510.09103","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2510.09103","pdf_url":"https://arxiv.org/pdf/2510.09103","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"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":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"In":[0,25,68],"the":[1,63,66,69,76,87,146,150],"training":[2,33],"of":[3,65,72],"large":[4],"language":[5,117],"models,":[6],"momentum":[7,20,38,51,57,70,89,107],"is":[8,58],"widely":[9],"used":[10],"and":[11,78,112,130],"often":[12],"demonstrated":[13],"to":[14,39,61,74,122,139],"achieve":[15],"significant":[16],"acceleration.":[17],"However,":[18],"storing":[19],"typically":[21],"presents":[22],"memory":[23,102,136],"challenges.":[24],"this":[26,45],"paper,":[27],"we":[28,85,96],"propose":[29],"AdaPM,":[30,73],"an":[31],"adaptive":[32],"strategy":[34],"that":[35,98],"leverages":[36],"partial":[37,83,88],"implement":[40],"a":[41,49,91],"memory-efficient":[42,147],"optimizer.":[43],"To":[44],"end,":[46],"AdaPM":[47,132],"utilizes":[48],"non-uniform":[50],"design:":[52],"for":[53,114,127,158],"most":[54],"blocks,":[55],"full":[56],"not":[59],"necessary":[60],"preserve":[62],"performance":[64,79,113],"optimization.":[67],"design":[71],"mitigate":[75],"bias":[77,92],"loss":[80],"caused":[81],"by":[82,90,103,137,144],"momentum,":[84],"enhance":[86],"correction":[93],"technique.":[94],"Empirically,":[95],"verify":[97],"our":[99],"approach":[100],"reduces":[101],"over":[104,154],"$90\\%$":[105],"in":[106,141],"while":[108],"maintaining":[109],"both":[110],"efficiency":[111],"pretraining":[115,159],"various":[116],"models":[118],"ranging":[119],"from":[120],"60M":[121],"1.5B,":[123],"as":[124,126],"well":[125],"supervised":[128],"fine-tuning":[129],"RLHF.":[131],"can":[133],"further":[134],"reduce":[135],"up":[138],"$95\\%$":[140],"optimizer":[142],"states":[143],"combining":[145],"technique":[148],"on":[149],"second-order":[151],"statistic,":[152],"saving":[153],"$30\\%$":[155],"GPU":[156],"hours":[157],"GPT-2":[160],"1.5B.":[161]},"counts_by_year":[],"updated_date":"2026-07-03T08:13:44.112507","created_date":"2025-10-14T00:00:00"}