{"id":"https://openalex.org/W4412888416","doi":"https://doi.org/10.18653/v1/2025.findings-acl.475","title":"Debate, Reflect, and Distill: Multi-Agent Feedback with Tree-Structured Preference Optimization for Efficient Language Model Enhancement","display_name":"Debate, Reflect, and Distill: Multi-Agent Feedback with Tree-Structured Preference Optimization for Efficient Language Model Enhancement","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4412888416","doi":"https://doi.org/10.18653/v1/2025.findings-acl.475"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.findings-acl.475","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.475","pdf_url":"https://aclanthology.org/2025.findings-acl.475.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":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.findings-acl.475.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5101860752","display_name":"Xiaofeng Zhou","orcid":"https://orcid.org/0000-0002-6381-0225"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xiaofeng Zhou","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5113891245","display_name":"Heyan Huang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Heyan Huang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5081165986","display_name":"Lizi Liao","orcid":"https://orcid.org/0000-0002-9973-3305"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lizi Liao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":3,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":3.5175,"has_fulltext":true,"cited_by_count":2,"citation_normalized_percentile":{"value":0.93061615,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":98,"max":99},"biblio":{"volume":null,"issue":null,"first_page":"9122","last_page":"9137"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T12031","display_name":"Speech and dialogue systems","score":0.9435999989509583,"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/T12031","display_name":"Speech and dialogue systems","score":0.9435999989509583,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.9363999962806702,"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":0.9294000267982483,"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/preference","display_name":"Preference","score":0.777782678604126},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.6858084797859192},{"id":"https://openalex.org/keywords/tree","display_name":"Tree (set theory)","score":0.5364241003990173},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.4954451322555542},{"id":"https://openalex.org/keywords/natural-language-processing","display_name":"Natural language processing","score":0.43669408559799194},{"id":"https://openalex.org/keywords/machine-learning","display_name":"Machine learning","score":0.36360740661621094},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.1337169110774994},{"id":"https://openalex.org/keywords/statistics","display_name":"Statistics","score":0.08115240931510925}],"concepts":[{"id":"https://openalex.org/C2781249084","wikidata":"https://www.wikidata.org/wiki/Q908656","display_name":"Preference","level":2,"score":0.777782678604126},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6858084797859192},{"id":"https://openalex.org/C113174947","wikidata":"https://www.wikidata.org/wiki/Q2859736","display_name":"Tree (set theory)","level":2,"score":0.5364241003990173},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4954451322555542},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.43669408559799194},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.36360740661621094},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.1337169110774994},{"id":"https://openalex.org/C105795698","wikidata":"https://www.wikidata.org/wiki/Q12483","display_name":"Statistics","level":1,"score":0.08115240931510925},{"id":"https://openalex.org/C134306372","wikidata":"https://www.wikidata.org/wiki/Q7754","display_name":"Mathematical analysis","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.findings-acl.475","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.475","pdf_url":"https://aclanthology.org/2025.findings-acl.475.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.475","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.475","pdf_url":"https://aclanthology.org/2025.findings-acl.475.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":[{"id":"https://openalex.org/G2563629645","display_name":null,"funder_award_id":"U21B2009","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"}],"funders":[{"id":"https://openalex.org/F4320320671","display_name":"National Research Foundation","ror":"https://ror.org/05s0g1g46"},{"id":"https://openalex.org/F4320320709","display_name":"National Research Foundation Singapore","ror":"https://ror.org/03cpyc314"},{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"},{"id":"https://openalex.org/F4320322725","display_name":"China Scholarship Council","ror":"https://ror.org/04atp4p48"},{"id":"https://openalex.org/F4320328656","display_name":"Singapore Management University","ror":"https://ror.org/050qmg959"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4412888416.pdf","grobid_xml":"https://content.openalex.org/works/W4412888416.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W2961085424","https://openalex.org/W4306674287","https://openalex.org/W4387369504","https://openalex.org/W4394896187","https://openalex.org/W3170094116","https://openalex.org/W4386462264","https://openalex.org/W3107602296","https://openalex.org/W4364306694","https://openalex.org/W4312192474","https://openalex.org/W4283697347"],"abstract_inverted_index":{"Large":[0],"Language":[1],"Models":[2],"(LLMs)":[3],"continue":[4],"to":[5,48,85,96],"set":[6],"new":[7],"standards":[8],"in":[9],"knowledge-intensive":[10],"and":[11,51,62,73,125],"complex":[12],"reasoning":[13],"tasks,":[14],"yet":[15],"their":[16],"high":[17],"computational":[18],"demands":[19],"limit":[20],"widespread":[21],"adoption.While":[22],"distilling":[23],"large":[24,132],"models":[25,72],"into":[26,104],"smaller":[27,71],"ones":[28],"offers":[29],"a":[30,59,105,131],"sustainable":[31],"solution,":[32],"current":[33],"techniques-such":[34],"as":[35],"static":[36],"knowledge":[37],"distillation,":[38],"resource-intensive":[39],"reinforcement":[40],"learning":[41],"from":[42],"human":[43],"feedback,":[44],"or":[45],"limited":[46],"selfreflection-struggle":[47],"yield":[49],"substantial":[50],"lasting":[52],"performance":[53],"gains.In":[54],"this":[55],"paper,":[56],"we":[57,89],"present":[58],"novel":[60],"Debate":[61],"Reflect":[63],"(D&R)":[64],"framework":[65],"that":[66,117],"orchestrates":[67],"multi-turn":[68],"debates":[69],"between":[70],"stronger":[74],"teacher":[75],"models,":[76],"eliciting":[77],"actionable":[78],"feedback":[79],"(e.g.,":[80],"error":[81],"analysis,":[82],"corrective":[83],"strategies)":[84],"guide":[86],"student":[87],"models.Further,":[88],"introduce":[90],"Treestructured":[91],"Direct":[92],"Preference":[93],"Optimization":[94],"(T-DPO)":[95],"efficiently":[97],"leverage":[98],"these":[99],"debate":[100],"logs,":[101],"organizing":[102],"interactions":[103],"hierarchical":[106],"format":[107],"for":[108],"effective":[109],"training.Empirical":[110],"evaluations":[111],"across":[112],"diverse":[113],"NLP":[114],"benchmarks":[115],"demonstrate":[116],"our":[118],"approach":[119],"significantly":[120],"improves":[121],"smaller-model":[122],"accuracy,":[123],"robustness,":[124],"generalization,":[126],"outperforming":[127],"conventional":[128],"baselines":[129],"by":[130],"margin.":[133]},"counts_by_year":[{"year":2026,"cited_by_count":2}],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2025-10-10T00:00:00"}