{"id":"https://openalex.org/W7214725969","doi":"https://doi.org/10.48550/arxiv.2609.33437","title":"SMAT: Simple and Efficient Merge-Aware Training","display_name":"SMAT: Simple and Efficient Merge-Aware Training","publication_year":2026,"publication_date":"2026-09-27","ids":{"openalex":"https://openalex.org/W7214725969","doi":"https://doi.org/10.48550/arxiv.2609.33437"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2609.33437","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.33437","pdf_url":"https://arxiv.org/pdf/2609.33437","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","is_traditional":true,"indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2609.33437","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5153487477","display_name":"Yanggan Gu","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I14243506","display_name":"Hong Kong Polytechnic University","ror":"https://ror.org/0030zas98","country_code":"HK","type":"education","lineage":["https://openalex.org/I14243506"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Gu, Yanggan","raw_affiliation_strings":["The Hong Kong Polytechnic University (PolyU)"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The Hong Kong Polytechnic University (PolyU)","institution_ids":["https://openalex.org/I14243506"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5153448908","display_name":"Yuanyi Wang","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I14243506","display_name":"Hong Kong Polytechnic University","ror":"https://ror.org/0030zas98","country_code":"HK","type":"education","lineage":["https://openalex.org/I14243506"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Wang, Yuanyi","raw_affiliation_strings":["The Hong Kong Polytechnic University (PolyU)"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The Hong Kong Polytechnic University (PolyU)","institution_ids":["https://openalex.org/I14243506"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5153431709","display_name":"Zhen Li","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I14243506","display_name":"Hong Kong Polytechnic University","ror":"https://ror.org/0030zas98","country_code":"HK","type":"education","lineage":["https://openalex.org/I14243506"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Li, Zhen","raw_affiliation_strings":["The Hong Kong Polytechnic University (PolyU)"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The Hong Kong Polytechnic University (PolyU)","institution_ids":["https://openalex.org/I14243506"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5153491463","display_name":"Shuo Cai","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I14243506","display_name":"Hong Kong Polytechnic University","ror":"https://ror.org/0030zas98","country_code":"HK","type":"education","lineage":["https://openalex.org/I14243506"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Cai, Shuo","raw_affiliation_strings":["The Hong Kong Polytechnic University (PolyU)"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The Hong Kong Polytechnic University (PolyU)","institution_ids":["https://openalex.org/I14243506"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5153461356","display_name":"Yuhang Liu","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I14243506","display_name":"Hong Kong Polytechnic University","ror":"https://ror.org/0030zas98","country_code":"HK","type":"education","lineage":["https://openalex.org/I14243506"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Liu, Yuhang","raw_affiliation_strings":["The Hong Kong Polytechnic University (PolyU)"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The Hong Kong Polytechnic University (PolyU)","institution_ids":["https://openalex.org/I14243506"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5153426164","display_name":"Junzhuo Li","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I4405261251","display_name":"The Hong Kong University of Science and Technology (Guangzhou)","ror":"https://ror.org/050h0vm43","country_code":"CN","type":"education","lineage":["https://openalex.org/I4405261251"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Li, Junzhuo","raw_affiliation_strings":["The Hong Kong University of Science and Technology ( Guangzhou )"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The Hong Kong University of Science and Technology ( Guangzhou )","institution_ids":["https://openalex.org/I4405261251"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5153484022","display_name":"Zihao Wang","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I177725633","display_name":"Chinese University of Hong Kong","ror":"https://ror.org/00t33hh48","country_code":"HK","type":"education","lineage":["https://openalex.org/I177725633"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Wang, Zihao","raw_affiliation_strings":["The Chinese University of Hong Kong"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The Chinese University of Hong Kong","institution_ids":["https://openalex.org/I177725633"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5153479781","display_name":"Hongxia Yang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yang, Hongxia","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":3,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":1,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.8550000190734863,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.8550000190734863,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.09529999643564224,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.03139999881386757,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/model-merging","display_name":"model merging","score":0.8420000076293945},{"id":"https://openalex.org/keywords/expert-training","display_name":"expert training","score":0.4819999933242798},{"id":"https://openalex.org/keywords/scaling-coefficients","display_name":"scaling coefficients","score":0.4560000002384186},{"id":"https://openalex.org/keywords/periodic-scheduling","display_name":"periodic scheduling","score":0.6150000095367432},{"id":"https://openalex.org/keywords/kernel-fusion","display_name":"kernel fusion","score":0.9599999785423279},{"id":"https://openalex.org/keywords/language-models","display_name":"language models","score":0.6110000014305115},{"id":"https://openalex.org/keywords/vision-language-models","display_name":"vision-language models","score":0.8510000109672546},{"id":"https://openalex.org/keywords/training-efficiency","display_name":"training efficiency","score":0.43700000643730164}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7505000233650208},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.7085000276565552},{"id":"https://openalex.org/C2779960059","wikidata":"https://www.wikidata.org/wiki/Q7113681","display_name":"Overhead (engineering)","level":2,"score":0.6804999709129333},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.6308000087738037},{"id":"https://openalex.org/C74193536","wikidata":"https://www.wikidata.org/wiki/Q574844","display_name":"Kernel (algebra)","level":2,"score":0.6230000257492065},{"id":"https://openalex.org/C2780586882","wikidata":"https://www.wikidata.org/wiki/Q7520643","display_name":"Simple (philosophy)","level":2,"score":0.5730999708175659},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.49160000681877136},{"id":"https://openalex.org/C2778755073","wikidata":"https://www.wikidata.org/wiki/Q10858537","display_name":"Scale (ratio)","level":2,"score":0.4749999940395355},{"id":"https://openalex.org/C99844830","wikidata":"https://www.wikidata.org/wiki/Q102441924","display_name":"Scaling","level":2,"score":0.4706000089645386},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.41659998893737793},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.39989998936653137},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.39570000767707825},{"id":"https://openalex.org/C12725497","wikidata":"https://www.wikidata.org/wiki/Q810247","display_name":"Baseline (sea)","level":2,"score":0.3619999885559082},{"id":"https://openalex.org/C22367795","wikidata":"https://www.wikidata.org/wiki/Q7625208","display_name":"Structured prediction","level":2,"score":0.35690000653266907},{"id":"https://openalex.org/C140779682","wikidata":"https://www.wikidata.org/wiki/Q210868","display_name":"Sampling (signal processing)","level":3,"score":0.33889999985694885},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.3285999894142151},{"id":"https://openalex.org/C204323151","wikidata":"https://www.wikidata.org/wiki/Q905424","display_name":"Range (aeronautics)","level":2,"score":0.3249000012874603},{"id":"https://openalex.org/C58328972","wikidata":"https://www.wikidata.org/wiki/Q184609","display_name":"Expert system","level":2,"score":0.31029999256134033},{"id":"https://openalex.org/C175154964","wikidata":"https://www.wikidata.org/wiki/Q380077","display_name":"Task analysis","level":3,"score":0.30239999294281006},{"id":"https://openalex.org/C18555067","wikidata":"https://www.wikidata.org/wiki/Q8375051","display_name":"Joint (building)","level":2,"score":0.299699991941452},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.27959999442100525},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.27219998836517334}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2609.33437","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.33437","pdf_url":"https://arxiv.org/pdf/2609.33437","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.2609.33437","doi":"https://doi.org/10.48550/arxiv.2609.33437","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2609.33437","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:2609.33437","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.33437","pdf_url":"https://arxiv.org/pdf/2609.33437","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/W7214725969.pdf","grobid_xml":"https://content.openalex.org/works/W7214725969.grobid-xml"},"referenced_works_count":15,"referenced_works":["https://openalex.org/W4412888070","https://openalex.org/W4412888383","https://openalex.org/W4415020667","https://openalex.org/W4415020690","https://openalex.org/W4415188187","https://openalex.org/W4415316839","https://openalex.org/W4416036156","https://openalex.org/W7083825832","https://openalex.org/W7127341067","https://openalex.org/W7129261837","https://openalex.org/W7160906719","https://openalex.org/W7161147459","https://openalex.org/W7161706365","https://openalex.org/W7162574072","https://openalex.org/W7163005055"],"related_works":["https://openalex.org/W7214038908","https://openalex.org/W7203785103","https://openalex.org/W7168348223","https://openalex.org/W7078419339","https://openalex.org/W4415311596","https://openalex.org/W7084058976","https://openalex.org/W4403884751","https://openalex.org/W7221786810","https://openalex.org/W7128360454","https://openalex.org/W7165680275"],"abstract_inverted_index":{"Model":[0],"merging":[1,44,58,149],"integrates":[2],"the":[3,144,155],"capabilities":[4],"of":[5],"multiple":[6],"experts":[7],"without":[8],"joint":[9],"retraining,":[10],"but":[11,35],"standard":[12,168],"expert":[13,94],"training":[14,28,48],"optimizes":[15,93],"task":[16],"loss":[17,95,98],"alone":[18],"and":[19,46,75,96,109,119,130,139],"does":[20],"not":[21,39],"guarantee":[22],"good":[23],"performance":[24],"after":[25],"merging.":[26],"Merge-aware":[27],"(MAT)":[29],"aims":[30],"to":[31,123],"improve":[32],"merged":[33,101],"performance,":[34],"existing":[36],"methods":[37,59,150],"do":[38],"fully":[40],"account":[41],"for":[42,158],"common":[43,57],"operations":[45],"add":[47],"cost.":[49],"We":[50,112],"observe":[51],"that,":[52],"from":[53,79],"an":[54],"expert's":[55],"perspective,":[56],"can":[60],"be":[61],"described":[62],"by":[63,104,151],"three":[64],"operations:":[65],"Scale":[66],"reweights":[67],"its":[68],"own":[69],"update,":[70],"Mask":[71],"removes":[72],"selected":[73],"coordinates,":[74],"Perturb":[76],"adds":[77],"updates":[78],"other":[80],"experts.":[81],"Based":[82],"on":[83],"this":[84],"view,":[85],"we":[86],"introduce":[87,114],"SMAT":[88,125,142],"(Simple":[89],"MAT),":[90],"which":[91],"jointly":[92],"expected":[97],"at":[99],"simulated":[100],"parameters":[102],"generated":[103],"sampling":[105],"scaling":[106],"coefficients,":[107],"masks,":[108],"additive":[110],"noise.":[111],"further":[113],"periodic":[115],"scheduling,":[116],"kernel":[117],"fusion,":[118],"parameter":[120],"storage":[121],"switching":[122],"make":[124],"efficient,":[126],"with":[127,161],"one":[128,131],"forward":[129],"backward":[132],"pass":[133],"per":[134],"step.":[135],"Across":[136],"four":[137],"language":[138],"vision-language":[140],"backbones,":[141],"improves":[143],"mean":[145],"score":[146],"across":[147],"five":[148],"1.07-2.16":[152],"points":[153],"over":[154,167],"strongest":[156],"baseline":[157],"each":[159],"backbone,":[160],"less":[162],"than":[163],"2%":[164],"training-time":[165],"overhead":[166],"fine-tuning.":[169]},"counts_by_year":[{"year":2026,"cited_by_count":1}],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2026-09-30T00:00:00"}