{"id":"https://openalex.org/W4416072498","doi":"https://doi.org/10.48550/arxiv.2506.03489","title":"EpiCoDe: Boosting Model Performance Beyond Training with Extrapolation and Contrastive Decoding","display_name":"EpiCoDe: Boosting Model Performance Beyond Training with Extrapolation and Contrastive Decoding","publication_year":2025,"publication_date":"2025-06-04","ids":{"openalex":"https://openalex.org/W4416072498","doi":"https://doi.org/10.48550/arxiv.2506.03489"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2506.03489","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2506.03489","pdf_url":"https://arxiv.org/pdf/2506.03489","source":{"id":"https://openalex.org/S4393918464","display_name":"ArXiv.org","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","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/2506.03489","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5072361963","display_name":"Mingxu Tao","orcid":"https://orcid.org/0009-0007-2326-4980"},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Tao, Mingxu","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5091260655","display_name":"Jie Hu","orcid":"https://orcid.org/0000-0002-1725-6366"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hu, Jie","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5037042612","display_name":"Mingchuan Yang","orcid":"https://orcid.org/0000-0003-1511-4265"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yang, Mingchuan","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5082653046","display_name":"Yunhuai Liu","orcid":"https://orcid.org/0000-0002-1180-8078"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Yunhuai","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5037132097","display_name":"Dongyan Zhao","orcid":"https://orcid.org/0000-0002-0396-6703"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhao, Dongyan","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5102220317","display_name":"Yansong Feng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Feng, Yansong","raw_affiliation_strings":[],"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":6,"corresponding_author_ids":["https://openalex.org/A5072361963"],"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/T10028","display_name":"Topic Modeling","score":0.4253000020980835,"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":0.4253000020980835,"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/T11636","display_name":"Artificial Intelligence in Healthcare and Education","score":0.05950000137090683,"subfield":{"id":"https://openalex.org/subfields/2718","display_name":"Health Informatics"},"field":{"id":"https://openalex.org/fields/27","display_name":"Medicine"},"domain":{"id":"https://openalex.org/domains/4","display_name":"Health Sciences"}},{"id":"https://openalex.org/T11714","display_name":"Multimodal Machine Learning Applications","score":0.056699998676776886,"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/decoding-methods","display_name":"Decoding methods","score":0.6876000165939331},{"id":"https://openalex.org/keywords/boosting","display_name":"Boosting (machine learning)","score":0.652400016784668},{"id":"https://openalex.org/keywords/extrapolation","display_name":"Extrapolation","score":0.5530999898910522},{"id":"https://openalex.org/keywords/training","display_name":"Training (meteorology)","score":0.4643000066280365},{"id":"https://openalex.org/keywords/training-set","display_name":"Training set","score":0.4480000138282776},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.40720000863075256},{"id":"https://openalex.org/keywords/mechanism","display_name":"Mechanism (biology)","score":0.3580999970436096}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8503999710083008},{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.6876000165939331},{"id":"https://openalex.org/C46686674","wikidata":"https://www.wikidata.org/wiki/Q466303","display_name":"Boosting (machine learning)","level":2,"score":0.652400016784668},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5782999992370605},{"id":"https://openalex.org/C132459708","wikidata":"https://www.wikidata.org/wiki/Q744069","display_name":"Extrapolation","level":2,"score":0.5530999898910522},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.5199000239372253},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.4643000066280365},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.4480000138282776},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.40720000863075256},{"id":"https://openalex.org/C89611455","wikidata":"https://www.wikidata.org/wiki/Q6804646","display_name":"Mechanism (biology)","level":2,"score":0.3580999970436096},{"id":"https://openalex.org/C67186912","wikidata":"https://www.wikidata.org/wiki/Q367664","display_name":"Data modeling","level":2,"score":0.3370000123977661},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.3257000148296356},{"id":"https://openalex.org/C75553542","wikidata":"https://www.wikidata.org/wiki/Q178161","display_name":"A priori and a posteriori","level":2,"score":0.3089999854564667},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.28029999136924744},{"id":"https://openalex.org/C175154964","wikidata":"https://www.wikidata.org/wiki/Q380077","display_name":"Task analysis","level":3,"score":0.273499995470047},{"id":"https://openalex.org/C157657479","wikidata":"https://www.wikidata.org/wiki/Q2367247","display_name":"Closed captioning","level":3,"score":0.259799987077713},{"id":"https://openalex.org/C2778915421","wikidata":"https://www.wikidata.org/wiki/Q3643177","display_name":"Performance improvement","level":2,"score":0.251800000667572}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2506.03489","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2506.03489","pdf_url":"https://arxiv.org/pdf/2506.03489","source":{"id":"https://openalex.org/S4393918464","display_name":"ArXiv.org","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2506.03489","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2506.03489","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":"article"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2506.03489","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2506.03489","pdf_url":"https://arxiv.org/pdf/2506.03489","source":{"id":"https://openalex.org/S4393918464","display_name":"ArXiv.org","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"awards":[],"funders":[{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4416072498.pdf","grobid_xml":"https://content.openalex.org/works/W4416072498.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"The":[0],"remarkable":[1],"performance":[2,48],"of":[3,13,22,138],"Large":[4],"language":[5],"models":[6,28],"(LLMs)":[7],"relies":[8],"heavily":[9],"on":[10],"the":[11,19,81,86,89,122,136],"availability":[12],"abundant":[14],"high-quality":[15],"training":[16],"data.":[17],"However,":[18],"high":[20],"cost":[21],"acquiring":[23],"annotated":[24],"data":[25],"often":[26],"prevents":[27],"from":[29],"obtaining":[30],"capabilities":[31],"to":[32,60,74,120],"tackle":[33],"downstream":[34],"tasks.":[35],"In":[36],"this":[37],"paper,":[38],"we":[39],"introduce":[40],"a":[41,62,116],"novel":[42],"method,":[43],"EpiCoDe":[44,103],"that":[45,102],"boosts":[46],"model":[47,58,64],"in":[49,127],"data-scarcity":[50,128],"scenarios":[51],"without":[52],"extra":[53],"training.":[54],"We":[55,113],"first":[56],"employ":[57],"extrapolation":[59],"enhance":[61],"finetuned":[63,91],"with":[65,108],"its":[66],"inferior":[67],"version,":[68],"and":[69,88,110],"then":[70],"adopt":[71],"contrastive":[72,125],"decoding":[73,126],"further":[75,131],"reduce":[76],"predicted":[77],"errors,":[78],"by":[79,85],"comparing":[80],"logit":[82],"scores":[83],"given":[84],"extrapolated":[87],"vanilla":[90],"model.":[92],"Experiments":[93],"across":[94],"three":[95],"tasks":[96],"over":[97],"four":[98],"different":[99],"LLMs":[100],"show":[101],"consistently":[104],"outperforms":[105],"existing":[106],"methods":[107],"significant":[109],"robust":[111],"improvement.":[112],"also":[114],"propose":[115],"new":[117],"theoretical":[118],"framework":[119],"reveal":[121],"mechanism":[123],"behind":[124],"scenarios,":[129],"which":[130],"helps":[132],"us":[133],"better":[134],"understand":[135],"effectiveness":[137],"EpiCoDe.":[139]},"counts_by_year":[],"updated_date":"2026-03-10T16:38:18.471706","created_date":"2025-10-10T00:00:00"}