{"id":"https://openalex.org/W7166815635","doi":"https://doi.org/10.18653/v1/2026.findings-acl.2109","title":"Joint Optimization of Training Data and Policy in RLHF","display_name":"Joint Optimization of Training Data and Policy in RLHF","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7166815635","doi":"https://doi.org/10.18653/v1/2026.findings-acl.2109"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2026.findings-acl.2109","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.2109","pdf_url":"https://aclanthology.org/2026.findings-acl.2109.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 2026","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2026.findings-acl.2109.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5010912592","display_name":"Zhuohao Yu","orcid":"https://orcid.org/0009-0006-0037-0396"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhuohao Yu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139739083","display_name":"Jiali Zeng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jiali Zeng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139766583","display_name":"Weizheng Gu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Weizheng Gu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139827499","display_name":"Mengyuan Sun","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Mengyuan Sun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139851704","display_name":"Yidong Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yidong Wang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5024849044","display_name":"Fandong Meng","orcid":"https://orcid.org/0000-0002-8158-2377"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Fandong Meng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139746769","display_name":"Jie Zhou","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jie Zhou","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139790263","display_name":"Shikun Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shikun Zhang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5129797514","display_name":"Wei Ye","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wei Ye","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.83095723,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"42498","last_page":"42514"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11994","display_name":"Human Resource Development and Performance Evaluation","score":0.06279999762773514,"subfield":{"id":"https://openalex.org/subfields/3202","display_name":"Applied Psychology"},"field":{"id":"https://openalex.org/fields/32","display_name":"Psychology"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},"topics":[{"id":"https://openalex.org/T11994","display_name":"Human Resource Development and Performance Evaluation","score":0.06279999762773514,"subfield":{"id":"https://openalex.org/subfields/3202","display_name":"Applied Psychology"},"field":{"id":"https://openalex.org/fields/32","display_name":"Psychology"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},{"id":"https://openalex.org/T11902","display_name":"Intelligent Tutoring Systems and Adaptive Learning","score":0.06040000170469284,"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/T12401","display_name":"Scheduling and Timetabling Solutions","score":0.049400001764297485,"subfield":{"id":"https://openalex.org/subfields/1803","display_name":"Management Science and Operations Research"},"field":{"id":"https://openalex.org/fields/18","display_name":"Decision Sciences"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/joint","display_name":"Joint (building)","score":0.6061999797821045},{"id":"https://openalex.org/keywords/training","display_name":"Training (meteorology)","score":0.5659999847412109},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.271699994802475},{"id":"https://openalex.org/keywords/training-set","display_name":"Training set","score":0.26339998841285706},{"id":"https://openalex.org/keywords/process","display_name":"Process (computing)","score":0.2547000050544739}],"concepts":[{"id":"https://openalex.org/C18555067","wikidata":"https://www.wikidata.org/wiki/Q8375051","display_name":"Joint (building)","level":2,"score":0.6061999797821045},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5896000266075134},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.5659999847412109},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.31520000100135803},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.271699994802475},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.26339998841285706},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.2547000050544739},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.25189998745918274},{"id":"https://openalex.org/C18762648","wikidata":"https://www.wikidata.org/wiki/Q42213","display_name":"Work (physics)","level":2,"score":0.24390000104904175},{"id":"https://openalex.org/C147764199","wikidata":"https://www.wikidata.org/wiki/Q6865248","display_name":"Minification","level":2,"score":0.2395000010728836}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2026.findings-acl.2109","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.2109","pdf_url":"https://aclanthology.org/2026.findings-acl.2109.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 2026","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.findings-acl.2109","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.2109","pdf_url":"https://aclanthology.org/2026.findings-acl.2109.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 2026","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7166815635.pdf","grobid_xml":"https://content.openalex.org/works/W7166815635.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Traditional":[0],"reinforcement":[1],"learning":[2,17],"from":[3],"human":[4],"feedback":[5],"(RLHF)":[6],"optimizes":[7,50],"policies":[8,96],"on":[9,132],"fixed":[10],"training":[11,38,45,125],"inputs,":[12],"limiting":[13],"the":[14,30,55,68,73],"diversity":[15],"of":[16,23,37],"signals.We":[18],"propose":[19],"JODP":[20,49],"(Joint":[21],"Optimization":[22],"Data":[24],"and":[25,127,143],"Policy),":[26],"a":[27,81,91,107],"framework":[28],"where":[29,94],"evolving":[31],"policy":[32,56,103],"model":[33,151],"generates":[34,57],"improved":[35],"variants":[36],"problems":[39,46],"to":[40,66,84,110,148],"enhance":[41],"its":[42],"own":[43],"learning.While":[44],"remain":[47],"fixed,":[48],"how":[51],"they":[52],"are":[53,114],"presented:":[54],"specification":[58],"hints":[59],"that":[60],"guide":[61],"rollout":[62],"generation,":[63],"then":[64],"learns":[65],"reproduce":[67],"discovered":[69],"high-reward":[70],"behaviors":[71],"without":[72,87],"hints.This":[74],"\"if":[75],"you":[76],"can":[77],"solve":[78,85],"it":[79,86],"with":[80,140],"hint,":[82],"learn":[83],"one\"":[88],"principle":[89],"creates":[90],"co-evolutionary":[92],"dynamic":[93],"better":[95,98],"discover":[97],"specifications,":[99],"which":[100],"enable":[101],"further":[102],"improvement.JODP":[104],"operates":[105],"as":[106],"plug-and-play":[108],"enhancement":[109],"existing":[111],"algorithms:":[112],"specifications":[113],"selected":[115],"via":[116],"UCB":[117],"bandits":[118],"for":[119],"exploration-exploitation":[120],"balance,":[121],"used":[122],"only":[123],"during":[124],"rollouts,":[126],"discarded":[128],"at":[129],"deployment.Through":[130],"evaluation":[131],"safety":[133],"alignment":[134],"tasks,":[135],"we":[136],"demonstrate":[137],"consistent":[138],"improvements":[139],"GRPO,":[141],"RLOO,":[142],"REINFORCE++,":[144],"allowing":[145],"4B":[146],"models":[147],"approach":[149],"8B":[150],"performance":[152],"using":[153],"less":[154],"than":[155],"1%":[156],"additional":[157],"computational":[158],"overhead.":[159]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2026-07-02T00:00:00"}