{"id":"https://openalex.org/W4411541877","doi":"https://doi.org/10.1145/3715275.3732204","title":"Detecting Prefix Bias in LLM-based Reward Models","display_name":"Detecting Prefix Bias in LLM-based Reward Models","publication_year":2025,"publication_date":"2025-06-23","ids":{"openalex":"https://openalex.org/W4411541877","doi":"https://doi.org/10.1145/3715275.3732204"},"language":"en","primary_location":{"id":"doi:10.1145/3715275.3732204","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3715275.3732204","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3715275.3732204","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2025 ACM Conference on Fairness, Accountability, and Transparency","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://dl.acm.org/doi/pdf/10.1145/3715275.3732204","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5101824596","display_name":"Ashwin Kumar","orcid":"https://orcid.org/0009-0003-8782-1388"},"institutions":[{"id":"https://openalex.org/I204465549","display_name":"Washington University in St. Louis","ror":"https://ror.org/01yc7t268","country_code":"US","type":"education","lineage":["https://openalex.org/I204465549"]}],"countries":["US"],"is_corresponding":true,"raw_author_name":"Ashwin Kumar","raw_affiliation_strings":["Washington University in St Louis, St Louis, USA"],"raw_orcid":"https://orcid.org/0009-0003-8782-1388","affiliations":[{"raw_affiliation_string":"Washington University in St Louis, St Louis, USA","institution_ids":["https://openalex.org/I204465549"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5067013133","display_name":"Yuzi He","orcid":"https://orcid.org/0000-0003-2083-1349"},"institutions":[{"id":"https://openalex.org/I4210114444","display_name":"Meta (United States)","ror":"https://ror.org/01zbnvs85","country_code":"US","type":"company","lineage":["https://openalex.org/I4210114444"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Yuzi He","raw_affiliation_strings":["Meta Platforms, Inc., Menlo Park, USA"],"raw_orcid":"https://orcid.org/0000-0003-2083-1349","affiliations":[{"raw_affiliation_string":"Meta Platforms, Inc., Menlo Park, USA","institution_ids":["https://openalex.org/I4210114444"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5001344368","display_name":"Aram Markosyan","orcid":"https://orcid.org/0000-0002-3213-8333"},"institutions":[{"id":"https://openalex.org/I4210114444","display_name":"Meta (United States)","ror":"https://ror.org/01zbnvs85","country_code":"US","type":"company","lineage":["https://openalex.org/I4210114444"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Aram H Markosyan","raw_affiliation_strings":["Meta Platforms, Inc., Menlo Park, USA"],"raw_orcid":"https://orcid.org/0000-0002-3213-8333","affiliations":[{"raw_affiliation_string":"Meta Platforms, Inc., Menlo Park, USA","institution_ids":["https://openalex.org/I4210114444"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5015164018","display_name":"Bobbie Chern","orcid":null},"institutions":[{"id":"https://openalex.org/I4210114444","display_name":"Meta (United States)","ror":"https://ror.org/01zbnvs85","country_code":"US","type":"company","lineage":["https://openalex.org/I4210114444"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Bobbie Chern","raw_affiliation_strings":["Meta Platforms, Inc, Sunnyvale, USA"],"raw_orcid":"https://orcid.org/0009-0008-2313-4390","affiliations":[{"raw_affiliation_string":"Meta Platforms, Inc, Sunnyvale, USA","institution_ids":["https://openalex.org/I4210114444"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5051152384","display_name":"Imanol Arrieta-Ibarra","orcid":"https://orcid.org/0000-0003-4402-2618"},"institutions":[{"id":"https://openalex.org/I142600864","display_name":"College of San Mateo","ror":"https://ror.org/01gwn6z70","country_code":"US","type":"education","lineage":["https://openalex.org/I142600864"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Imanol Arrieta-Ibarra","raw_affiliation_strings":["Independent, San Mateo, USA"],"raw_orcid":"https://orcid.org/0000-0003-4402-2618","affiliations":[{"raw_affiliation_string":"Independent, San Mateo, USA","institution_ids":["https://openalex.org/I142600864"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":5,"corresponding_author_ids":["https://openalex.org/A5101824596"],"corresponding_institution_ids":["https://openalex.org/I204465549"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.06723995,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"3196","last_page":"3206"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.9998999834060669,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.9998999834060669,"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/T13629","display_name":"Text Readability and Simplification","score":0.9984999895095825,"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/T10028","display_name":"Topic Modeling","score":0.9959999918937683,"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/prefix","display_name":"Prefix","score":0.7611581087112427},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.7018080949783325},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.38751035928726196}],"concepts":[{"id":"https://openalex.org/C141603448","wikidata":"https://www.wikidata.org/wiki/Q134830","display_name":"Prefix","level":2,"score":0.7611581087112427},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7018080949783325},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.38751035928726196},{"id":"https://openalex.org/C41895202","wikidata":"https://www.wikidata.org/wiki/Q8162","display_name":"Linguistics","level":1,"score":0.0},{"id":"https://openalex.org/C138885662","wikidata":"https://www.wikidata.org/wiki/Q5891","display_name":"Philosophy","level":0,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3715275.3732204","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3715275.3732204","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3715275.3732204","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2025 ACM Conference on Fairness, Accountability, and Transparency","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.1145/3715275.3732204","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3715275.3732204","pdf_url":"https://dl.acm.org/doi/pdf/10.1145/3715275.3732204","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2025 ACM Conference on Fairness, Accountability, and Transparency","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4411541877.pdf","grobid_xml":"https://content.openalex.org/works/W4411541877.grobid-xml"},"referenced_works_count":6,"referenced_works":["https://openalex.org/W3133702157","https://openalex.org/W3174685870","https://openalex.org/W3213241618","https://openalex.org/W4386302153","https://openalex.org/W4387821331","https://openalex.org/W4411120625"],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W2142481367","https://openalex.org/W3196321793","https://openalex.org/W3080705045","https://openalex.org/W2385527937","https://openalex.org/W2005880840","https://openalex.org/W2507465767","https://openalex.org/W4385305499"],"abstract_inverted_index":{"Reinforcement":[0],"Learning":[1],"with":[2],"Human":[3],"Feedback":[4],"(RLHF)":[5],"has":[6],"emerged":[7],"as":[8],"a":[9,115],"key":[10],"paradigm":[11],"for":[12,34,138],"task-specific":[13],"fine-tuning":[14],"of":[15,30,105,108,130],"language":[16],"models":[17,40,69,83],"using":[18],"human":[19],"preference":[20,25,82,94],"data.While":[21],"numerous":[22],"publicly":[23],"available":[24],"datasets":[26,95],"provide":[27],"pairwise":[28],"comparisons":[29],"responses,":[31],"the":[32,37,109,128,135,153],"potential":[33],"biases":[35,80],"in":[36,57,64,81,126,144,158],"resulting":[38],"reward":[39,68,97,149],"remains":[41],"underexplored.In":[42],"this":[43,103],"work,":[44],"we":[45,113],"introduce":[46],"novel":[47],"methods":[48],"to":[49,77,102,119,152],"detect":[50],"and":[51,86,96,142,147],"evaluate":[52],"prefix":[53,131],"bias-a":[54],"systematic":[55],"shift":[56],"model":[58,98,111],"preferences":[59],"triggered":[60],"by":[61],"minor":[62],"variations":[63],"query":[65],"prefixes-in":[66],"LLM-based":[67],"trained":[70],"on":[71,156],"such":[72],"datasets.We":[73],"leverage":[74],"these":[75,121],"metrics":[76],"reveal":[78],"significant":[79],"across":[84],"racial":[85],"gender":[87],"dimensions.Our":[88],"comprehensive":[89],"evaluation":[90,143],"spans":[91],"diverse":[92],"open-source":[93],"architectures,":[99],"demonstrating":[100],"susceptibility":[101],"kind":[104],"bias":[106],"regardless":[107],"underlying":[110],"architecture.Furthermore,":[112],"propose":[114],"data":[116],"augmentation":[117],"strategy":[118],"mitigate":[120],"biases,":[122],"showing":[123],"its":[124],"effectiveness":[125],"reducing":[127],"impact":[129],"bias.Our":[132],"findings":[133],"highlight":[134],"critical":[136],"need":[137],"bias-aware":[139],"dataset":[140],"design":[141],"developing":[145],"fair":[146],"reliable":[148],"models,":[150],"contributing":[151],"broader":[154],"discourse":[155],"fairness":[157],"AI.":[159]},"counts_by_year":[],"updated_date":"2026-03-11T06:11:40.159057","created_date":"2025-10-10T00:00:00"}