{"id":"https://openalex.org/W4412887774","doi":"https://doi.org/10.18653/v1/2025.findings-acl.1030","title":"Confidence Improves Self-Consistency in LLMs","display_name":"Confidence Improves Self-Consistency in LLMs","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4412887774","doi":"https://doi.org/10.18653/v1/2025.findings-acl.1030"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.findings-acl.1030","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.1030","pdf_url":"https://aclanthology.org/2025.findings-acl.1030.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: ACL 2025","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.findings-acl.1030.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5092596481","display_name":"Amir Taubenfeld","orcid":null},"institutions":[{"id":"https://openalex.org/I78577930","display_name":"Columbia University","ror":"https://ror.org/00hj8s172","country_code":"US","type":"education","lineage":["https://openalex.org/I78577930"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Amir Taubenfeld","raw_affiliation_strings":["Columbia University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Columbia University","institution_ids":["https://openalex.org/I78577930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5092279590","display_name":"Tom Sheffer","orcid":null},"institutions":[{"id":"https://openalex.org/I78577930","display_name":"Columbia University","ror":"https://ror.org/00hj8s172","country_code":"US","type":"education","lineage":["https://openalex.org/I78577930"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Tom Sheffer","raw_affiliation_strings":["Columbia University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Columbia University","institution_ids":["https://openalex.org/I78577930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5078910984","display_name":"E. O. Ofek","orcid":"https://orcid.org/0000-0002-6786-8774"},"institutions":[{"id":"https://openalex.org/I1291425158","display_name":"Google (United States)","ror":"https://ror.org/00njsd438","country_code":"US","type":"company","lineage":["https://openalex.org/I1291425158","https://openalex.org/I4210128969"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Eran Ofek","raw_affiliation_strings":["Google Research ,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Google Research ,","institution_ids":["https://openalex.org/I1291425158"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5056266191","display_name":"Amir Feder","orcid":"https://orcid.org/0000-0001-5472-1135"},"institutions":[{"id":"https://openalex.org/I78577930","display_name":"Columbia University","ror":"https://ror.org/00hj8s172","country_code":"US","type":"education","lineage":["https://openalex.org/I78577930"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Amir Feder","raw_affiliation_strings":["Columbia University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Columbia University","institution_ids":["https://openalex.org/I78577930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5074374901","display_name":"Ariel Goldstein","orcid":"https://orcid.org/0000-0003-4017-3363"},"institutions":[{"id":"https://openalex.org/I197251160","display_name":"Hebrew University of Jerusalem","ror":"https://ror.org/03qxff017","country_code":"IL","type":"education","lineage":["https://openalex.org/I197251160"]}],"countries":["IL"],"is_corresponding":false,"raw_author_name":"Ariel Goldstein","raw_affiliation_strings":["The Hebrew University of Jerusalem ,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The Hebrew University of Jerusalem ,","institution_ids":["https://openalex.org/I197251160"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5087260801","display_name":"Zorik Gekhman","orcid":null},"institutions":[{"id":"https://openalex.org/I1291425158","display_name":"Google (United States)","ror":"https://ror.org/00njsd438","country_code":"US","type":"company","lineage":["https://openalex.org/I1291425158","https://openalex.org/I4210128969"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Zorik Gekhman","raw_affiliation_strings":["Google Research ,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Google Research ,","institution_ids":["https://openalex.org/I1291425158"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5037714222","display_name":"Gal Yona","orcid":null},"institutions":[{"id":"https://openalex.org/I1291425158","display_name":"Google (United States)","ror":"https://ror.org/00njsd438","country_code":"US","type":"company","lineage":["https://openalex.org/I1291425158","https://openalex.org/I4210128969"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Gal Yona","raw_affiliation_strings":["Google Research ,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Google Research ,","institution_ids":["https://openalex.org/I1291425158"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":3,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":11.3628,"has_fulltext":true,"cited_by_count":8,"citation_normalized_percentile":{"value":0.98653151,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":95,"max":99},"biblio":{"volume":null,"issue":null,"first_page":"20090","last_page":"20111"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10215","display_name":"Semantic Web and Ontologies","score":0.45680001378059387,"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/T10215","display_name":"Semantic Web and Ontologies","score":0.45680001378059387,"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/T11396","display_name":"Artificial Intelligence in Healthcare","score":0.45089998841285706,"subfield":{"id":"https://openalex.org/subfields/3605","display_name":"Health Information Management"},"field":{"id":"https://openalex.org/fields/36","display_name":"Health Professions"},"domain":{"id":"https://openalex.org/domains/4","display_name":"Health Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/consistency","display_name":"Consistency (knowledge bases)","score":0.6397993564605713},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.48159050941467285},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.15217959880828857}],"concepts":[{"id":"https://openalex.org/C2776436953","wikidata":"https://www.wikidata.org/wiki/Q5163215","display_name":"Consistency (knowledge bases)","level":2,"score":0.6397993564605713},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.48159050941467285},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.15217959880828857}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.findings-acl.1030","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.1030","pdf_url":"https://aclanthology.org/2025.findings-acl.1030.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: ACL 2025","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.findings-acl.1030","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.1030","pdf_url":"https://aclanthology.org/2025.findings-acl.1030.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: ACL 2025","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4412887774.pdf","grobid_xml":"https://content.openalex.org/works/W4412887774.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W2390279801","https://openalex.org/W4391913857","https://openalex.org/W2358668433","https://openalex.org/W4396701345","https://openalex.org/W2376932109","https://openalex.org/W2001405890","https://openalex.org/W4396696052"],"abstract_inverted_index":{"Self-consistency":[0],"decoding":[1],"enhances":[2],"LLMs'":[3],"performance":[4],"on":[5,59,83,106,177],"reasoning":[6,11,101],"tasks":[7],"by":[8,103],"sampling":[9,24],"diverse":[10],"paths":[12,29,102],"and":[13,86,131,158],"selecting":[14],"the":[15,34,37,42,65,73,97,111,135,139,147,166,174],"most":[16,43,140],"frequent":[17,44],"answer.However,":[18],"it":[19,70],"is":[20,30],"computationally":[21],"expensive,":[22],"as":[23,41],"many":[25],"of":[26,100,113,126,168],"these":[27,153],"(lengthy)":[28],"required":[31,98],"to":[32,134,145,173],"increase":[33],"chances":[35],"that":[36,119,161],"correct":[38,74,130],"answer":[39,75],"emerges":[40],"one.To":[45],"address":[46],"this,":[47],"we":[48,109],"introduce":[49,110],"Confidence-Informed":[50],"Self-Consistency":[51],"(CISC).CISC":[52],"performs":[53],"a":[54,77],"weighted":[55],"majority":[56],"vote":[57],"based":[58],"confidence":[60,115,142],"scores":[61],"obtained":[62],"directly":[63],"from":[64],"model.By":[66],"prioritizing":[67],"highconfidence":[68],"paths,":[69],"can":[71,163],"identify":[72],"with":[76],"significantly":[78],"smaller":[79],"sample":[80],"size.When":[81],"tested":[82],"nine":[84],"models":[85],"four":[87],"datasets,":[88],"CISC":[89],"outperforms":[90],"self-consistency":[91],"in":[92,128],"nearly":[93],"all":[94],"configurations,":[95],"reducing":[96],"number":[99],"over":[104],"40%":[105],"average.In":[107],"addition,":[108],"notion":[112],"withinquestion":[114],"evaluation,":[116],"after":[117],"showing":[118],"standard":[120],"evaluation":[121],"methods":[122],"are":[123],"poor":[124],"predictors":[125],"success":[127],"distinguishing":[129],"incorrect":[132],"answers":[133],"same":[136],"question.In":[137],"fact,":[138],"calibrated":[141],"method":[143],"proved":[144],"be":[146],"least":[148],"effective":[149],"for":[150],"CISC.Lastly,":[151],"beyond":[152],"practical":[154],"implications,":[155],"our":[156],"results":[157],"analyses":[159],"show":[160],"LLMs":[162],"effectively":[164],"judge":[165],"correctness":[167],"their":[169],"own":[170],"outputs,":[171],"contributing":[172],"ongoing":[175],"debate":[176],"this":[178],"topic.":[179]},"counts_by_year":[{"year":2026,"cited_by_count":6},{"year":2025,"cited_by_count":2}],"updated_date":"2026-08-24T07:32:12.397491","created_date":"2025-10-10T00:00:00"}