{"id":"https://openalex.org/W7220732105","doi":"https://doi.org/10.48550/arxiv.2610.05126","title":"Selecting Repetition Counts Across Model Scales in Data-Constrained Pretraining","display_name":"Selecting Repetition Counts Across Model Scales in Data-Constrained Pretraining","publication_year":2026,"publication_date":"2026-10-04","ids":{"openalex":"https://openalex.org/W7220732105","doi":"https://doi.org/10.48550/arxiv.2610.05126"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2610.05126","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2610.05126","pdf_url":"https://arxiv.org/pdf/2610.05126","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/2610.05126","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5157702346","display_name":"Ziyue WANG","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I4400009020","display_name":"Institute of Science Tokyo","ror":"https://ror.org/05dqf9946","country_code":"JP","type":"education","lineage":["https://openalex.org/I4400009020"]}],"countries":["JP"],"is_corresponding":false,"raw_author_name":"WANG, Ziyue","raw_affiliation_strings":["Institute of Science Tokyo"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Institute of Science Tokyo","institution_ids":["https://openalex.org/I4400009020"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5157696828","display_name":"T. Kanamori","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I4400009020","display_name":"Institute of Science Tokyo","ror":"https://ror.org/05dqf9946","country_code":"JP","type":"education","lineage":["https://openalex.org/I4400009020"]}],"countries":["JP"],"is_corresponding":false,"raw_author_name":"Kanamori, T.","raw_affiliation_strings":["Institute of Science Tokyo"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Institute of Science Tokyo","institution_ids":["https://openalex.org/I4400009020"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I4400009020"],"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/T10028","display_name":"Topic Modeling","score":0.9301999807357788,"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/T10028","display_name":"Topic Modeling","score":0.9301999807357788,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.05460000038146973,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.006200000178068876,"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/small-language-models","display_name":"small language models","score":0.7440000176429749},{"id":"https://openalex.org/keywords/model-scaling","display_name":"model scaling","score":0.6899999976158142},{"id":"https://openalex.org/keywords/loss-curves","display_name":"loss curves","score":0.6759999990463257},{"id":"https://openalex.org/keywords/pubmed","display_name":"PubMed","score":0.48100000619888306}],"concepts":[{"id":"https://openalex.org/C2776141515","wikidata":"https://www.wikidata.org/wiki/Q1274479","display_name":"Repetition (rhetorical device)","level":2,"score":0.8094000220298767},{"id":"https://openalex.org/C189430467","wikidata":"https://www.wikidata.org/wiki/Q7293293","display_name":"Ranking (information retrieval)","level":2,"score":0.5569000244140625},{"id":"https://openalex.org/C105795698","wikidata":"https://www.wikidata.org/wiki/Q12483","display_name":"Statistics","level":1,"score":0.5175999999046326},{"id":"https://openalex.org/C81917197","wikidata":"https://www.wikidata.org/wiki/Q628760","display_name":"Selection (genetic algorithm)","level":2,"score":0.5128999948501587},{"id":"https://openalex.org/C48921125","wikidata":"https://www.wikidata.org/wiki/Q10861030","display_name":"Linear regression","level":2,"score":0.49459999799728394},{"id":"https://openalex.org/C108010975","wikidata":"https://www.wikidata.org/wiki/Q500094","display_name":"Pruning","level":2,"score":0.4912000000476837},{"id":"https://openalex.org/C83546350","wikidata":"https://www.wikidata.org/wiki/Q1139051","display_name":"Regression","level":2,"score":0.48249998688697815},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.4787999987602234},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.47269999980926514},{"id":"https://openalex.org/C99844830","wikidata":"https://www.wikidata.org/wiki/Q102441924","display_name":"Scaling","level":2,"score":0.4691999852657318},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.45980000495910645},{"id":"https://openalex.org/C152877465","wikidata":"https://www.wikidata.org/wiki/Q208042","display_name":"Regression analysis","level":2,"score":0.4401000142097473},{"id":"https://openalex.org/C28719098","wikidata":"https://www.wikidata.org/wiki/Q44946","display_name":"Point (geometry)","level":2,"score":0.42969998717308044},{"id":"https://openalex.org/C2778755073","wikidata":"https://www.wikidata.org/wiki/Q10858537","display_name":"Scale (ratio)","level":2,"score":0.4025999903678894},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.38920000195503235},{"id":"https://openalex.org/C163175372","wikidata":"https://www.wikidata.org/wiki/Q3339222","display_name":"Linear model","level":2,"score":0.36800000071525574},{"id":"https://openalex.org/C93959086","wikidata":"https://www.wikidata.org/wiki/Q6888345","display_name":"Model selection","level":2,"score":0.3598000109195709},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.3517000079154968},{"id":"https://openalex.org/C527412718","wikidata":"https://www.wikidata.org/wiki/Q855395","display_name":"Interpretation (philosophy)","level":2,"score":0.3375000059604645},{"id":"https://openalex.org/C117220453","wikidata":"https://www.wikidata.org/wiki/Q5172842","display_name":"Correlation","level":2,"score":0.3237999975681305},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.3212999999523163},{"id":"https://openalex.org/C149364088","wikidata":"https://www.wikidata.org/wiki/Q185917","display_name":"Translation (biology)","level":4,"score":0.3111000061035156},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.30469998717308044},{"id":"https://openalex.org/C2779466056","wikidata":"https://www.wikidata.org/wiki/Q107630651","display_name":"Time point","level":2,"score":0.2802000045776367},{"id":"https://openalex.org/C2776502983","wikidata":"https://www.wikidata.org/wiki/Q690182","display_name":"Contrast (vision)","level":2,"score":0.2777000069618225},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.2766000032424927},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.2734000086784363},{"id":"https://openalex.org/C70519679","wikidata":"https://www.wikidata.org/wiki/Q6666755","display_name":"Log-linear model","level":3,"score":0.2667999863624573},{"id":"https://openalex.org/C3020672099","wikidata":"https://www.wikidata.org/wiki/Q857354","display_name":"Longitudinal data","level":2,"score":0.2653999924659729},{"id":"https://openalex.org/C16012445","wikidata":"https://www.wikidata.org/wiki/Q1501135","display_name":"Mixed model","level":2,"score":0.2590000033378601}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2610.05126","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2610.05126","pdf_url":"https://arxiv.org/pdf/2610.05126","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.2610.05126","doi":"https://doi.org/10.48550/arxiv.2610.05126","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2610.05126","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:2610.05126","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2610.05126","pdf_url":"https://arxiv.org/pdf/2610.05126","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":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7220732105.pdf","grobid_xml":"https://content.openalex.org/works/W7220732105.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W7203605131","https://openalex.org/W7165876075","https://openalex.org/W7220742388","https://openalex.org/W7161173344","https://openalex.org/W4385570327","https://openalex.org/W4406129401","https://openalex.org/W7133188041","https://openalex.org/W4416052770","https://openalex.org/W4312089714","https://openalex.org/W7160282514"],"abstract_inverted_index":{"The":[0],"repetition":[1,48,62,89],"count":[2,111],"that":[3,60],"works":[4],"best":[5,14],"for":[6,91],"a":[7,16,26,35,55,84,94,127,168],"small":[8],"language":[9],"model":[10,52,144,156],"may":[11],"not":[12],"remain":[13],"at":[15,34,93,117],"larger":[17,95],"scale.":[18,96],"We":[19,74,134],"study":[20],"this":[21],"effect":[22],"in":[23,154],"pretraining":[24],"with":[25,31,51,145],"finite":[27],"target":[28,37],"corpus":[29],"mixed":[30],"generic":[32],"data":[33,41],"fixed":[36,103],"fraction.":[38],"On":[39,97],"Wikipedia-derived":[40],"and":[42,54,99,120],"Proof-Pile-2,":[43],"the":[44,108,113,137,159,164,172],"ranking":[45],"of":[46,87,171],"measured":[47,110],"counts":[49,90],"changes":[50],"size,":[53],"520M":[56],"Proof-Pile-2":[57],"experiment":[58],"confirms":[59],"reducing":[61],"from":[63,78],"sixteen":[64],"to":[65,82,130,140],"eight":[66],"improves":[67],"loss":[68,76,149],"while":[69],"using":[70],"fewer":[71],"training":[72,106],"tokens.":[73],"use":[75],"curves":[77],"several":[79],"smaller":[80],"models":[81],"retain":[83,107],"short":[85],"list":[86],"promising":[88],"evaluation":[92,115],"PubMed":[98],"Caselaw,":[100],"candidate":[101,124],"sets":[102],"before":[104],"target-model":[105],"lowest-loss":[109],"on":[112],"original":[114],"grids":[116],"both":[118],"200M":[119],"520M.":[121],"This":[122],"supports":[123],"retention":[125],"as":[126],"practical":[128],"alternative":[129],"exact":[131],"point":[132],"prediction.":[133],"also":[135],"relate":[136],"pruning":[138],"regression":[139],"an":[141],"empirical":[142],"scaling":[143],"two":[146],"opposing":[147],"repetition-dependent":[148],"terms.":[150],"A":[151],"first-order":[152],"expansion":[153],"log":[155],"size":[157],"yields":[158],"linear":[160],"form":[161],"used":[162],"by":[163],"selection":[165],"rule,":[166],"providing":[167],"scaling-based":[169],"interpretation":[170],"candidate-selection":[173],"procedure.":[174]},"counts_by_year":[],"updated_date":"2026-10-28T10:33:24.094632","created_date":"2026-10-07T00:00:00"}