{"id":"https://openalex.org/W7155056132","doi":"https://doi.org/10.1007/s00521-026-12052-9","title":"2T-FT: Two-Token Fine-tuning improves zero-shot performance with minimal training","display_name":"2T-FT: Two-Token Fine-tuning improves zero-shot performance with minimal training","publication_year":2026,"publication_date":"2026-04-21","ids":{"openalex":"https://openalex.org/W7155056132","doi":"https://doi.org/10.1007/s00521-026-12052-9"},"language":"en","primary_location":{"id":"doi:10.1007/s00521-026-12052-9","is_oa":true,"landing_page_url":"https://doi.org/10.1007/s00521-026-12052-9","pdf_url":"https://link.springer.com/content/pdf/10.1007/s00521-026-12052-9.pdf","source":{"id":"https://openalex.org/S147897268","display_name":"Neural Computing and Applications","issn_l":"0941-0643","issn":["0941-0643","1433-3058"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319900","host_organization_name":"Springer Science+Business Media","host_organization_lineage":["https://openalex.org/P4310319900","https://openalex.org/P4310319965"],"host_organization_lineage_names":["Springer Science+Business Media","Springer Nature"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Neural Computing and Applications","raw_type":"journal-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"hybrid","oa_url":"https://link.springer.com/content/pdf/10.1007/s00521-026-12052-9.pdf","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5038741018","display_name":"Paulo S. Neto","orcid":null},"institutions":[{"id":"https://openalex.org/I126460647","display_name":"Universidade Federal do Rio Grande","ror":"https://ror.org/05hpfkn88","country_code":"BR","type":"education","lineage":["https://openalex.org/I126460647"]}],"countries":["BR"],"is_corresponding":false,"raw_author_name":"Paulo S. Neto","raw_affiliation_strings":["Universidade Federal do Rio Grande (FURG), Rio Grande, RS, Brasil"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Universidade Federal do Rio Grande (FURG), Rio Grande, RS, Brasil","institution_ids":["https://openalex.org/I126460647"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134154194","display_name":"Jardel D. S. Dyonisio","orcid":null},"institutions":[{"id":"https://openalex.org/I126460647","display_name":"Universidade Federal do Rio Grande","ror":"https://ror.org/05hpfkn88","country_code":"BR","type":"education","lineage":["https://openalex.org/I126460647"]}],"countries":["BR"],"is_corresponding":false,"raw_author_name":"Jardel D. S. Dyonisio","raw_affiliation_strings":["Universidade Federal do Rio Grande (FURG), Rio Grande, RS, Brasil"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Universidade Federal do Rio Grande (FURG), Rio Grande, RS, Brasil","institution_ids":["https://openalex.org/I126460647"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134148472","display_name":"Jo\u00e3o F. S. S. Lemos","orcid":null},"institutions":[{"id":"https://openalex.org/I126460647","display_name":"Universidade Federal do Rio Grande","ror":"https://ror.org/05hpfkn88","country_code":"BR","type":"education","lineage":["https://openalex.org/I126460647"]}],"countries":["BR"],"is_corresponding":false,"raw_author_name":"Jo\u00e3o F. S. S. Lemos","raw_affiliation_strings":["Universidade Federal do Rio Grande (FURG), Rio Grande, RS, Brasil"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Universidade Federal do Rio Grande (FURG), Rio Grande, RS, Brasil","institution_ids":["https://openalex.org/I126460647"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5036650570","display_name":"Felipe K\u00fchne","orcid":null},"institutions":[{"id":"https://openalex.org/I126460647","display_name":"Universidade Federal do Rio Grande","ror":"https://ror.org/05hpfkn88","country_code":"BR","type":"education","lineage":["https://openalex.org/I126460647"]}],"countries":["BR"],"is_corresponding":false,"raw_author_name":"Felipe K\u00fchne","raw_affiliation_strings":["Universidade Federal do Rio Grande (FURG), Rio Grande, RS, Brasil"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Universidade Federal do Rio Grande (FURG), Rio Grande, RS, Brasil","institution_ids":["https://openalex.org/I126460647"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5123671326","display_name":"Rodrigo S. Guerra","orcid":null},"institutions":[{"id":"https://openalex.org/I126460647","display_name":"Universidade Federal do Rio Grande","ror":"https://ror.org/05hpfkn88","country_code":"BR","type":"education","lineage":["https://openalex.org/I126460647"]}],"countries":["BR"],"is_corresponding":true,"raw_author_name":"Rodrigo S. Guerra","raw_affiliation_strings":["Universidade Federal do Rio Grande (FURG), Rio Grande, RS, Brasil"],"raw_orcid":"https://orcid.org/0000-0003-4011-0901","affiliations":[{"raw_affiliation_string":"Universidade Federal do Rio Grande (FURG), Rio Grande, RS, Brasil","institution_ids":["https://openalex.org/I126460647"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5134107206","display_name":"Paulo L. J. Drews","orcid":null},"institutions":[{"id":"https://openalex.org/I126460647","display_name":"Universidade Federal do Rio Grande","ror":"https://ror.org/05hpfkn88","country_code":"BR","type":"education","lineage":["https://openalex.org/I126460647"]}],"countries":["BR"],"is_corresponding":false,"raw_author_name":"Paulo L. J. Drews","raw_affiliation_strings":["Universidade Federal do Rio Grande (FURG), Rio Grande, RS, Brasil"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Universidade Federal do Rio Grande (FURG), Rio Grande, RS, Brasil","institution_ids":["https://openalex.org/I126460647"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":["https://openalex.org/A5123671326"],"corresponding_institution_ids":["https://openalex.org/I126460647"],"apc_list":{"value":2990,"currency":"USD","value_usd":2990},"apc_paid":{"value":2990,"currency":"USD","value_usd":2990},"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.48511579,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":"38","issue":"9","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":0.2822999954223633,"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":0.2822999954223633,"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/T14347","display_name":"Big Data and Digital Economy","score":0.07620000094175339,"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"}},{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.07199999690055847,"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/decoding-methods","display_name":"Decoding methods","score":0.6915000081062317},{"id":"https://openalex.org/keywords/property","display_name":"Property (philosophy)","score":0.5738999843597412},{"id":"https://openalex.org/keywords/exploit","display_name":"Exploit","score":0.5404000282287598},{"id":"https://openalex.org/keywords/code","display_name":"Code (set theory)","score":0.48030000925064087},{"id":"https://openalex.org/keywords/range","display_name":"Range (aeronautics)","score":0.4453999996185303},{"id":"https://openalex.org/keywords/performance-improvement","display_name":"Performance improvement","score":0.41530001163482666},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.33149999380111694}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.832099974155426},{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.6915000081062317},{"id":"https://openalex.org/C189950617","wikidata":"https://www.wikidata.org/wiki/Q937228","display_name":"Property (philosophy)","level":2,"score":0.5738999843597412},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.5404000282287598},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.48030000925064087},{"id":"https://openalex.org/C204323151","wikidata":"https://www.wikidata.org/wiki/Q905424","display_name":"Range (aeronautics)","level":2,"score":0.4453999996185303},{"id":"https://openalex.org/C2778915421","wikidata":"https://www.wikidata.org/wiki/Q3643177","display_name":"Performance improvement","level":2,"score":0.41530001163482666},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3352999985218048},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.33149999380111694},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.3070000112056732},{"id":"https://openalex.org/C18762648","wikidata":"https://www.wikidata.org/wiki/Q42213","display_name":"Work (physics)","level":2,"score":0.3066999912261963},{"id":"https://openalex.org/C51823790","wikidata":"https://www.wikidata.org/wiki/Q504353","display_name":"Greedy algorithm","level":2,"score":0.29120001196861267},{"id":"https://openalex.org/C127162648","wikidata":"https://www.wikidata.org/wiki/Q16858953","display_name":"Channel (broadcasting)","level":2,"score":0.29100000858306885},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.2800999879837036},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.27469998598098755},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.2745000123977661},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.2653999924659729},{"id":"https://openalex.org/C2780586882","wikidata":"https://www.wikidata.org/wiki/Q7520643","display_name":"Simple (philosophy)","level":2,"score":0.26339998841285706},{"id":"https://openalex.org/C94375191","wikidata":"https://www.wikidata.org/wiki/Q11205","display_name":"Arithmetic","level":1,"score":0.2615000009536743},{"id":"https://openalex.org/C68597687","wikidata":"https://www.wikidata.org/wiki/Q362601","display_name":"Computational Science and Engineering","level":2,"score":0.25060001015663147}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1007/s00521-026-12052-9","is_oa":true,"landing_page_url":"https://doi.org/10.1007/s00521-026-12052-9","pdf_url":"https://link.springer.com/content/pdf/10.1007/s00521-026-12052-9.pdf","source":{"id":"https://openalex.org/S147897268","display_name":"Neural Computing and Applications","issn_l":"0941-0643","issn":["0941-0643","1433-3058"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319900","host_organization_name":"Springer Science+Business Media","host_organization_lineage":["https://openalex.org/P4310319900","https://openalex.org/P4310319965"],"host_organization_lineage_names":["Springer Science+Business Media","Springer Nature"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Neural Computing and Applications","raw_type":"journal-article"}],"best_oa_location":{"id":"doi:10.1007/s00521-026-12052-9","is_oa":true,"landing_page_url":"https://doi.org/10.1007/s00521-026-12052-9","pdf_url":"https://link.springer.com/content/pdf/10.1007/s00521-026-12052-9.pdf","source":{"id":"https://openalex.org/S147897268","display_name":"Neural Computing and Applications","issn_l":"0941-0643","issn":["0941-0643","1433-3058"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319900","host_organization_name":"Springer Science+Business Media","host_organization_lineage":["https://openalex.org/P4310319900","https://openalex.org/P4310319965"],"host_organization_lineage_names":["Springer Science+Business Media","Springer Nature"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Neural Computing and Applications","raw_type":"journal-article"},"sustainable_development_goals":[],"awards":[],"funders":[{"id":"https://openalex.org/F4320322987","display_name":"Universidade Federal do Rio Grande do Sul","ror":"https://ror.org/041yk2d64"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7155056132.pdf","grobid_xml":"https://content.openalex.org/works/W7155056132.grobid-xml"},"referenced_works_count":18,"referenced_works":["https://openalex.org/W2251935656","https://openalex.org/W2890894339","https://openalex.org/W2970750583","https://openalex.org/W2998617917","https://openalex.org/W3011411500","https://openalex.org/W3034446185","https://openalex.org/W3034457116","https://openalex.org/W3098324846","https://openalex.org/W3170403598","https://openalex.org/W4385569970","https://openalex.org/W4385571157","https://openalex.org/W4385571219","https://openalex.org/W4385571831","https://openalex.org/W4402671620","https://openalex.org/W4409363794","https://openalex.org/W4415078012","https://openalex.org/W4415797312","https://openalex.org/W6948966457"],"related_works":[],"abstract_inverted_index":{"Abstract":[0],"Chain":[1],"of":[2,13,22,36,66,90],"Thought":[3],"(CoT)":[4],"prompting":[5,69],"has":[6,47],"been":[7],"shown":[8,48],"to":[9,111,163],"improve":[10],"the":[11,34,41,64,103,147,156,164],"performance":[12,123],"large":[14],"language":[15],"models":[16],"(LLMs)":[17],"in":[18,57,63],"a":[19,76,112,132,151],"wide":[20],"range":[21],"tasks,":[23],"including":[24],"arithmetic,":[25],"common-sense,":[26],"and":[27,140],"symbolic":[28],"reasoning.":[29],"However,":[30],"this":[31,72,82],"improvement":[32,134],"requires":[33],"development":[35],"effective":[37],"CoT":[38,50,93,108],"prompts.":[39],"On":[40],"other":[42],"hand,":[43],"more":[44],"recent":[45],"work":[46],"that":[49,80,97],"reasoning":[51],"paths":[52,109],"are":[53],"often":[54],"inherently":[55],"present":[56],"top-k":[58],"alternative":[59],"decoding":[60,153],"sequences,":[61],"even":[62],"absence":[65],"any":[67],"specific":[68,88],"technique.":[70],"In":[71],"study,":[73],"we":[74,130],"propose":[75],"new":[77],"fine-tuning":[78,98,166],"method":[79],"exploits":[81],"property":[83],"by":[84,159],"targeting":[85],"only":[86,99],"two":[87,100],"tokens":[89,101],"these":[91],"pre-existing":[92],"responses.":[94],"We":[95],"demonstrate":[96],"using":[102],"model\u2019s":[104],"own":[105],"implicitly":[106],"generated":[107],"leads":[110],"significant":[113],"efficiency":[114],"gain,":[115],"reducing":[116,155],"training":[117],"time":[118,158],"while":[119],"still":[120],"achieving":[121],"meaningful":[122],"improvements.":[124],"When":[125],"evaluated":[126],"on":[127,135,138,142,146],"arithmetic":[128],"datasets,":[129],"achieved":[131],"22.7%":[133],"MultiArith,":[136],"9.0%":[137],"GSM8K,":[139],"2.3%":[141],"SVAMP":[143],"when":[144],"validated":[145],"Phi-2":[148],"model":[149],"from":[150],"greedy":[152],"perspective,":[154],"processing":[157],"over":[160],"90%":[161],"compared":[162],"LoRA":[165],"method.":[167],"Code":[168],"is":[169],"publicly":[170],"available":[171],"at:":[172],"https://github.com/paulosantosneto/2tft":[173],".":[174]},"counts_by_year":[],"updated_date":"2026-07-23T08:03:31.855105","created_date":"2026-04-22T00:00:00"}