{"id":"https://openalex.org/W4414359487","doi":"https://doi.org/10.24963/ijcai.2025/931","title":"Token-Level Accept or Reject: A Micro Alignment Approach for Large Language Models","display_name":"Token-Level Accept or Reject: A Micro Alignment Approach for Large Language Models","publication_year":2025,"publication_date":"2025-09-01","ids":{"openalex":"https://openalex.org/W4414359487","doi":"https://doi.org/10.24963/ijcai.2025/931"},"language":"en","primary_location":{"id":"doi:10.24963/ijcai.2025/931","is_oa":false,"landing_page_url":"https://doi.org/10.24963/ijcai.2025/931","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the Thirty-Fourth International Joint Conference on Artificial Intelligence","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":false,"oa_status":"closed","oa_url":null,"any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5100354819","display_name":"Yang Zhang","orcid":"https://orcid.org/0000-0003-4658-565X"},"institutions":[{"id":"https://openalex.org/I14243506","display_name":"Hong Kong Polytechnic University","ror":"https://ror.org/0030zas98","country_code":"HK","type":"education","lineage":["https://openalex.org/I14243506"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Yang Zhang","raw_affiliation_strings":["Hong Kong Polytechnic University, Hong Kong SAR, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Hong Kong Polytechnic University, Hong Kong SAR, China","institution_ids":["https://openalex.org/I14243506"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5110299446","display_name":"Yu Yu","orcid":null},"institutions":[{"id":"https://openalex.org/I4210150405","display_name":"Hongzhiwei Technology (China)","ror":"https://ror.org/04bapa014","country_code":"CN","type":"company","lineage":["https://openalex.org/I4210150405"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yu Yu","raw_affiliation_strings":["MemTensor (Shanghai) Technology Co., Ltd, Shanghai, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"MemTensor (Shanghai) Technology Co., Ltd, Shanghai, China","institution_ids":["https://openalex.org/I4210150405"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5057816960","display_name":"Bo Tang","orcid":"https://orcid.org/0000-0001-7129-0250"},"institutions":[{"id":"https://openalex.org/I308837","display_name":"Suzhou University of Science and Technology","ror":"https://ror.org/04en8wb91","country_code":"CN","type":"education","lineage":["https://openalex.org/I308837"]},{"id":"https://openalex.org/I4210125878","display_name":"Suzhou Research Institute","ror":"https://ror.org/03ebk0c60","country_code":"CN","type":"facility","lineage":["https://openalex.org/I4210125878"]},{"id":"https://openalex.org/I4210150405","display_name":"Hongzhiwei Technology (China)","ror":"https://ror.org/04bapa014","country_code":"CN","type":"company","lineage":["https://openalex.org/I4210150405"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Bo Tang","raw_affiliation_strings":["MemTensor (Shanghai) Technology Co., Ltd, Shanghai, China","University of Science and Technology of China, Suzhou Institute for Advanced Research, Suzhou, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"MemTensor (Shanghai) Technology Co., Ltd, Shanghai, China","institution_ids":["https://openalex.org/I4210150405"]},{"raw_affiliation_string":"University of Science and Technology of China, Suzhou Institute for Advanced Research, Suzhou, China","institution_ids":["https://openalex.org/I308837","https://openalex.org/I4210125878"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5103205239","display_name":"Limin Zhu","orcid":"https://orcid.org/0000-0003-3194-6731"},"institutions":[{"id":"https://openalex.org/I126520041","display_name":"University of Science and Technology of China","ror":"https://ror.org/04c4dkn09","country_code":"CN","type":"education","lineage":["https://openalex.org/I126520041","https://openalex.org/I19820366"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yu Zhu","raw_affiliation_strings":["University of Science and Technology of China, Hefei, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Science and Technology of China, Hefei, China","institution_ids":["https://openalex.org/I126520041"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5057734833","display_name":"Chuxiong Sun","orcid":null},"institutions":[{"id":"https://openalex.org/I4210136246","display_name":"China Telecom (China)","ror":"https://ror.org/03jgnzt20","country_code":"CN","type":"company","lineage":["https://openalex.org/I4210136246"]},{"id":"https://openalex.org/I4387153335","display_name":"China Telecom","ror":"https://ror.org/05p67dv18","country_code":null,"type":"company","lineage":["https://openalex.org/I4387153335"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Chuxiong Sun","raw_affiliation_strings":["China Telecom Corporation Limited Beijing Research Institute, Beijing, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"China Telecom Corporation Limited Beijing Research Institute, Beijing, China","institution_ids":["https://openalex.org/I4210136246","https://openalex.org/I4387153335"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5052103501","display_name":"Wenqiang Wei","orcid":"https://orcid.org/0000-0003-2078-9056"},"institutions":[{"id":"https://openalex.org/I4210150405","display_name":"Hongzhiwei Technology (China)","ror":"https://ror.org/04bapa014","country_code":"CN","type":"company","lineage":["https://openalex.org/I4210150405"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Wenqiang Wei","raw_affiliation_strings":["MemTensor (Shanghai) Technology Co., Ltd, Shanghai, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"MemTensor (Shanghai) Technology Co., Ltd, Shanghai, China","institution_ids":["https://openalex.org/I4210150405"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5066864248","display_name":"Jie Hu","orcid":null},"institutions":[{"id":"https://openalex.org/I4210136246","display_name":"China Telecom (China)","ror":"https://ror.org/03jgnzt20","country_code":"CN","type":"company","lineage":["https://openalex.org/I4210136246"]},{"id":"https://openalex.org/I4387153335","display_name":"China Telecom","ror":"https://ror.org/05p67dv18","country_code":null,"type":"company","lineage":["https://openalex.org/I4387153335"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jie Hu","raw_affiliation_strings":["China Telecom Corporation Limited Beijing Research Institute, Beijing, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"China Telecom Corporation Limited Beijing Research Institute, Beijing, China","institution_ids":["https://openalex.org/I4210136246","https://openalex.org/I4387153335"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5044995870","display_name":"Zheng Xie","orcid":"https://orcid.org/0000-0001-8260-3259"},"institutions":[{"id":"https://openalex.org/I200845125","display_name":"Nanjing University of Information Science and Technology","ror":"https://ror.org/02y0rxk19","country_code":"CN","type":"education","lineage":["https://openalex.org/I200845125"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zipeng Xie","raw_affiliation_strings":["Nanjing University of Information Science and Technology, Nanjing, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Nanjing University of Information Science and Technology, Nanjing, China","institution_ids":["https://openalex.org/I200845125"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5113013369","display_name":"Zhiyu Li","orcid":"https://orcid.org/0000-0003-3771-0199"},"institutions":[{"id":"https://openalex.org/I4210150405","display_name":"Hongzhiwei Technology (China)","ror":"https://ror.org/04bapa014","country_code":"CN","type":"company","lineage":["https://openalex.org/I4210150405"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhiyu Li","raw_affiliation_strings":["MemTensor (Shanghai) Technology Co., Ltd, Shanghai, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"MemTensor (Shanghai) Technology Co., Ltd, Shanghai, China","institution_ids":["https://openalex.org/I4210150405"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5090294171","display_name":"Feiyu Xiong","orcid":"https://orcid.org/0000-0002-1456-2202"},"institutions":[{"id":"https://openalex.org/I4210150405","display_name":"Hongzhiwei Technology (China)","ror":"https://ror.org/04bapa014","country_code":"CN","type":"company","lineage":["https://openalex.org/I4210150405"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Feiyu Xiong","raw_affiliation_strings":["MemTensor (Shanghai) Technology Co., Ltd, Shanghai, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"MemTensor (Shanghai) Technology Co., Ltd, Shanghai, China","institution_ids":["https://openalex.org/I4210150405"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5044341587","display_name":"Edward Chung","orcid":"https://orcid.org/0000-0001-6969-7764"},"institutions":[{"id":"https://openalex.org/I14243506","display_name":"Hong Kong Polytechnic University","ror":"https://ror.org/0030zas98","country_code":"HK","type":"education","lineage":["https://openalex.org/I14243506"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Edward Chung","raw_affiliation_strings":["Hong Kong Polytechnic University, Hong Kong SAR, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Hong Kong Polytechnic University, Hong Kong SAR, China","institution_ids":["https://openalex.org/I14243506"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":8,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"8375","last_page":"8383"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.9675999879837036,"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/T10028","display_name":"Topic Modeling","score":0.9675999879837036,"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":0.9649999737739563,"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/process","display_name":"Process (computing)","score":0.6151999831199646},{"id":"https://openalex.org/keywords/code","display_name":"Code (set theory)","score":0.5175999999046326},{"id":"https://openalex.org/keywords/computational-model","display_name":"Computational model","score":0.4797999858856201},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.4596000015735626},{"id":"https://openalex.org/keywords/binary-number","display_name":"Binary number","score":0.4302999973297119},{"id":"https://openalex.org/keywords/preference","display_name":"Preference","score":0.39820000529289246},{"id":"https://openalex.org/keywords/natural-language","display_name":"Natural language","score":0.34049999713897705}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7319999933242798},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.6151999831199646},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.5175999999046326},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.48829999566078186},{"id":"https://openalex.org/C66024118","wikidata":"https://www.wikidata.org/wiki/Q1122506","display_name":"Computational model","level":2,"score":0.4797999858856201},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.4596000015735626},{"id":"https://openalex.org/C48372109","wikidata":"https://www.wikidata.org/wiki/Q3913","display_name":"Binary number","level":2,"score":0.4302999973297119},{"id":"https://openalex.org/C2781249084","wikidata":"https://www.wikidata.org/wiki/Q908656","display_name":"Preference","level":2,"score":0.39820000529289246},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.38190001249313354},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.3709000051021576},{"id":"https://openalex.org/C195324797","wikidata":"https://www.wikidata.org/wiki/Q33742","display_name":"Natural language","level":2,"score":0.34049999713897705},{"id":"https://openalex.org/C43126263","wikidata":"https://www.wikidata.org/wiki/Q128751","display_name":"Source code","level":2,"score":0.32670000195503235},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.3000999987125397},{"id":"https://openalex.org/C2993724205","wikidata":"https://www.wikidata.org/wiki/Q315","display_name":"Human language","level":2,"score":0.2994000017642975},{"id":"https://openalex.org/C155092808","wikidata":"https://www.wikidata.org/wiki/Q182557","display_name":"Computational linguistics","level":2,"score":0.29600000381469727},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.2727000117301941},{"id":"https://openalex.org/C63435697","wikidata":"https://www.wikidata.org/wiki/Q864135","display_name":"Binary code","level":3,"score":0.2563000023365021},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.2515999972820282}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.24963/ijcai.2025/931","is_oa":false,"landing_page_url":"https://doi.org/10.24963/ijcai.2025/931","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the Thirty-Fourth International Joint Conference on Artificial Intelligence","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"With":[0],"the":[1,67,72,101,138],"rapid":[2],"development":[3],"of":[4,42,66,100],"Large":[5],"Language":[6],"Models":[7],"(LLMs),":[8],"aligning":[9],"these":[10],"models":[11,140],"with":[12,40],"human":[13],"preferences":[14],"and":[15,22,49,109,129,137],"values":[16],"is":[17],"critical":[18],"to":[19,63],"ensuring":[20],"ethical":[21],"safe":[23],"applications.":[24],"However,":[25],"existing":[26],"alignment":[27,73,120],"techniques":[28],"such":[29],"as":[30,98],"RLHF":[31],"or":[32,96],"DPO":[33],"often":[34],"require":[35],"direct":[36],"fine-tuning":[37],"on":[38],"LLMs":[39,108],"billions":[41],"parameters,":[43],"resulting":[44],"in":[45,119],"substantial":[46],"computational":[47,124],"costs":[48],"inefficiencies.":[50],"To":[51],"address":[52],"this,":[53],"we":[54],"propose":[55],"Micro":[56],"token-level":[57,81],"Accept-Reject":[58],"Aligning":[59],"(MARA)":[60],"approach":[61],"designed":[62],"operate":[64],"independently":[65],"language":[68],"models.":[69],"MARA":[70,115],"simplifies":[71],"process":[74],"by":[75],"decomposing":[76],"sentence-level":[77],"preference":[78],"learning":[79],"into":[80],"binary":[82],"classification,":[83],"where":[84],"a":[85],"compact":[86],"three-layer":[87],"fully-connected":[88],"network":[89],"determines":[90],"whether":[91],"candidate":[92],"tokens":[93],"are":[94,132,141],"\u201cAccepted\u201d":[95],"\u201cRejected\u201d":[97],"part":[99],"response.":[102],"Extensive":[103],"experiments":[104],"across":[105],"seven":[106],"different":[107],"three":[110],"open-source":[111],"datasets":[112],"show":[113],"that":[114],"achieves":[116],"significant":[117],"improvements":[118],"performance":[121],"while":[122],"reducing":[123],"costs.":[125],"The":[126],"source":[127],"code":[128],"implementation":[130],"details":[131],"publicly":[133],"available":[134],"at":[135,143],"https://github.com/IAAR-Shanghai/MARA,":[136],"trained":[139],"released":[142],"https://huggingface.co/IAAR-Shanghai/MARA_AGENTS.":[144]},"counts_by_year":[],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}