{"id":"https://openalex.org/W4416034190","doi":"https://doi.org/10.18653/v1/2025.findings-emnlp.836","title":"The Price of Format: Diversity Collapse in LLMs","display_name":"The Price of Format: Diversity Collapse in LLMs","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4416034190","doi":"https://doi.org/10.18653/v1/2025.findings-emnlp.836"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2025.findings-emnlp.836","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.836","pdf_url":"https://aclanthology.org/2025.findings-emnlp.836.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: EMNLP 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-emnlp.836.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5116335669","display_name":"Longfei Yun","orcid":null},"institutions":[{"id":"https://openalex.org/I36258959","display_name":"University of California San Diego","ror":"https://ror.org/0168r3w48","country_code":"US","type":"education","lineage":["https://openalex.org/I36258959"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Longfei Yun","raw_affiliation_strings":["University of California , San Diego"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of California , San Diego","institution_ids":["https://openalex.org/I36258959"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102648310","display_name":"Chenyang An","orcid":null},"institutions":[{"id":"https://openalex.org/I36258959","display_name":"University of California San Diego","ror":"https://ror.org/0168r3w48","country_code":"US","type":"education","lineage":["https://openalex.org/I36258959"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Chenyang An","raw_affiliation_strings":["University of California , San Diego"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of California , San Diego","institution_ids":["https://openalex.org/I36258959"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100384121","display_name":"Zilong Wang","orcid":"https://orcid.org/0009-0009-8981-9541"},"institutions":[{"id":"https://openalex.org/I36258959","display_name":"University of California San Diego","ror":"https://ror.org/0168r3w48","country_code":"US","type":"education","lineage":["https://openalex.org/I36258959"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Zilong Wang","raw_affiliation_strings":["University of California , San Diego"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of California , San Diego","institution_ids":["https://openalex.org/I36258959"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5001458884","display_name":"Letian Peng","orcid":"https://orcid.org/0000-0001-5039-3496"},"institutions":[{"id":"https://openalex.org/I36258959","display_name":"University of California San Diego","ror":"https://ror.org/0168r3w48","country_code":"US","type":"education","lineage":["https://openalex.org/I36258959"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Letian Peng","raw_affiliation_strings":["University of California , San Diego"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of California , San Diego","institution_ids":["https://openalex.org/I36258959"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5119753714","display_name":"Jingbo Shang","orcid":null},"institutions":[{"id":"https://openalex.org/I36258959","display_name":"University of California San Diego","ror":"https://ror.org/0168r3w48","country_code":"US","type":"education","lineage":["https://openalex.org/I36258959"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Jingbo Shang","raw_affiliation_strings":["University of California , San Diego"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of California , San Diego","institution_ids":["https://openalex.org/I36258959"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I36258959"],"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":"15454","last_page":"15468"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T14330","display_name":"Library Science and Information Systems","score":0.08269999921321869,"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"}},"topics":[{"id":"https://openalex.org/T14330","display_name":"Library Science and Information Systems","score":0.08269999921321869,"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/T10843","display_name":"Diversity and Career in Medicine","score":0.05389999970793724,"subfield":{"id":"https://openalex.org/subfields/3318","display_name":"Gender Studies"},"field":{"id":"https://openalex.org/fields/33","display_name":"Social Sciences"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},{"id":"https://openalex.org/T12478","display_name":"Wikis in Education and Collaboration","score":0.05119999870657921,"subfield":{"id":"https://openalex.org/subfields/3315","display_name":"Communication"},"field":{"id":"https://openalex.org/fields/33","display_name":"Social Sciences"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/diversity","display_name":"Diversity (politics)","score":0.7495999932289124},{"id":"https://openalex.org/keywords/consistency","display_name":"Consistency (knowledge bases)","score":0.6843000054359436},{"id":"https://openalex.org/keywords/task","display_name":"Task (project management)","score":0.580299973487854},{"id":"https://openalex.org/keywords/disk-formatting","display_name":"Disk formatting","score":0.5591999888420105},{"id":"https://openalex.org/keywords/phenomenon","display_name":"Phenomenon","score":0.5393000245094299},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.517300009727478}],"concepts":[{"id":"https://openalex.org/C2781316041","wikidata":"https://www.wikidata.org/wiki/Q1230584","display_name":"Diversity (politics)","level":2,"score":0.7495999932289124},{"id":"https://openalex.org/C2776436953","wikidata":"https://www.wikidata.org/wiki/Q5163215","display_name":"Consistency (knowledge bases)","level":2,"score":0.6843000054359436},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.580299973487854},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5792999863624573},{"id":"https://openalex.org/C88006597","wikidata":"https://www.wikidata.org/wiki/Q690117","display_name":"Disk formatting","level":2,"score":0.5591999888420105},{"id":"https://openalex.org/C50335755","wikidata":"https://www.wikidata.org/wiki/Q483247","display_name":"Phenomenon","level":2,"score":0.5393000245094299},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.517300009727478},{"id":"https://openalex.org/C61797465","wikidata":"https://www.wikidata.org/wiki/Q1188986","display_name":"Term (time)","level":2,"score":0.5066999793052673},{"id":"https://openalex.org/C11012388","wikidata":"https://www.wikidata.org/wiki/Q170658","display_name":"Creativity","level":2,"score":0.4862000048160553},{"id":"https://openalex.org/C180747234","wikidata":"https://www.wikidata.org/wiki/Q23373","display_name":"Cognitive psychology","level":1,"score":0.4124000072479248},{"id":"https://openalex.org/C204323151","wikidata":"https://www.wikidata.org/wiki/Q905424","display_name":"Range (aeronautics)","level":2,"score":0.33629998564720154},{"id":"https://openalex.org/C2781256819","wikidata":"https://www.wikidata.org/wiki/Q16828835","display_name":"Antecedent (behavioral psychology)","level":2,"score":0.3179999887943268},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.31790000200271606},{"id":"https://openalex.org/C37279795","wikidata":"https://www.wikidata.org/wiki/Q2492305","display_name":"Consistency model","level":3,"score":0.29319998621940613},{"id":"https://openalex.org/C162324750","wikidata":"https://www.wikidata.org/wiki/Q8134","display_name":"Economics","level":0,"score":0.2903999984264374},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.2890999913215637},{"id":"https://openalex.org/C149782125","wikidata":"https://www.wikidata.org/wiki/Q160039","display_name":"Econometrics","level":1,"score":0.25940001010894775}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.findings-emnlp.836","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.836","pdf_url":"https://aclanthology.org/2025.findings-emnlp.836.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: EMNLP 2025","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.findings-emnlp.836","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.836","pdf_url":"https://aclanthology.org/2025.findings-emnlp.836.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: EMNLP 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/W4416034190.pdf","grobid_xml":"https://content.openalex.org/works/W4416034190.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Instruction-tuned":[0],"large":[1],"language":[2],"models":[3],"(LLMs)":[4],"employ":[5],"structured":[6,97],"templates,":[7],"such":[8,27],"as":[9],"role":[10],"markers":[11],"and":[12,49,60,73,99,111,121,185],"special":[13],"tokens,":[14,154],"to":[15],"enforce":[16],"format":[17,117],"consistency":[18,118],"during":[19],"inference.However,":[20],"we":[21,33,88],"identify":[22],"a":[23,31,94],"critical":[24],"limitation":[25],"of":[26,96,152],"formatting:":[28],"it":[29],"induces":[30],"phenomenon":[32],"term":[34],"diversity":[35,66,143],"collapse,":[36],"where":[37],"the":[38,81,90,148,159,179],"model":[39,92],"generates":[40],"semantically":[41],"similar":[42],"outputs":[43],"for":[44,125,171,181],"open-ended":[45],"inputs,":[46],"undermining":[47],"creativity":[48],"variability.We":[50],"systematically":[51],"evaluate":[52,101],"this":[53],"effect":[54],"across":[55,103],"tasks":[56,127,137],"like":[57],"story":[58],"completion":[59],"free-form":[61],"generation,":[62],"finding":[63],"that":[64,116,165],"(1)":[65],"collapse":[67],"persists":[68],"even":[69],"under":[70],"high-temperature":[71],"sampling,":[72],"(2)":[74],"structural":[75,153],"tokens":[76],"in":[77],"templates":[78],"significantly":[79],"constrain":[80],"model's":[82],"output":[83,112,142,176],"space.To":[84],"contextualize":[85],"these":[86],"findings,":[87],"fine-tune":[89],"same":[91],"using":[93],"range":[95],"prompts":[98],"then":[100],"them":[102],"three":[104],"axes:":[105],"downstream":[106],"task":[107],"performance,":[108],"alignment":[109],"behavior,":[110],"diversity.Our":[113],"analysis":[114],"shows":[115],"between":[119],"fine-tuning":[120],"inference":[122],"is":[123,144],"crucial":[124],"structure-sensitive":[126],"(e.g.,":[128,138],"GSM8K,":[129],"IFEval),":[130],"but":[131],"has":[132],"marginal":[133],"influence":[134],"on":[135],"knowledge-heavy":[136],"MMLU,":[139],"WebQuestions).In":[140],"contrast,":[141],"primarily":[145],"governed":[146],"by":[147],"presence":[149],"or":[150],"absence":[151],"with":[155],"minimal":[156],"formatting":[157],"yielding":[158],"most":[160],"diverse":[161],"outputs.These":[162],"findings":[163],"reveal":[164],"current":[166],"prompting":[167],"conventions,":[168],"while":[169],"beneficial":[170],"alignment,":[172],"may":[173],"inadvertently":[174],"suppress":[175],"diversity,":[177],"underscoring":[178],"need":[180],"diversity-aware":[182],"prompt":[183],"design":[184],"instruction":[186],"tuning.":[187]},"counts_by_year":[{"year":2026,"cited_by_count":1}],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-11-08T00:00:00"}
