{"id":"https://openalex.org/W7138243647","doi":"https://doi.org/10.1609/aaai.v40i40.40696","title":"SDA: Steering-Driven Distribution Alignment for Open LLMs Without Fine-Tuning","display_name":"SDA: Steering-Driven Distribution Alignment for Open LLMs Without Fine-Tuning","publication_year":2026,"publication_date":"2026-03-14","ids":{"openalex":"https://openalex.org/W7138243647","doi":"https://doi.org/10.1609/aaai.v40i40.40696"},"language":null,"primary_location":{"id":"doi:10.1609/aaai.v40i40.40696","is_oa":true,"landing_page_url":"https://doi.org/10.1609/aaai.v40i40.40696","pdf_url":null,"source":{"id":"https://openalex.org/S4210191458","display_name":"Proceedings of the AAAI Conference on Artificial Intelligence","issn_l":"2159-5399","issn":["2159-5399","2374-3468"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/P4310320058","host_organization_name":"Association for the Advancement of Artificial Intelligence","host_organization_lineage":["https://openalex.org/P4310320058"],"host_organization_lineage_names":["Association for the Advancement of Artificial Intelligence"],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the AAAI Conference on Artificial Intelligence","raw_type":"journal-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"diamond","oa_url":"https://doi.org/10.1609/aaai.v40i40.40696","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5129659146","display_name":"Wei Xia","orcid":null},"institutions":[{"id":"https://openalex.org/I20231570","display_name":"Peking University","ror":"https://ror.org/02v51f717","country_code":"CN","type":"education","lineage":["https://openalex.org/I20231570"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Wei Xia","raw_affiliation_strings":["State Key Laboratory of General Artificial Intelligence, School of Intelligence Science and Technology, Peking University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"State Key Laboratory of General Artificial Intelligence, School of Intelligence Science and Technology, Peking University","institution_ids":["https://openalex.org/I20231570"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5129657811","display_name":"Zhi-Hong Deng","orcid":null},"institutions":[{"id":"https://openalex.org/I20231570","display_name":"Peking University","ror":"https://ror.org/02v51f717","country_code":"CN","type":"education","lineage":["https://openalex.org/I20231570"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhi-Hong Deng","raw_affiliation_strings":["State Key Laboratory of General Artificial Intelligence, School of Intelligence Science and Technology, Peking University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"State Key Laboratory of General Artificial Intelligence, School of Intelligence Science and Technology, Peking University","institution_ids":["https://openalex.org/I20231570"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I20231570"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.20483562,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":"40","issue":"40","first_page":"34025","last_page":"34033"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.3637000024318695,"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":0.3637000024318695,"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/T11636","display_name":"Artificial Intelligence in Healthcare and Education","score":0.15559999644756317,"subfield":{"id":"https://openalex.org/subfields/2718","display_name":"Health Informatics"},"field":{"id":"https://openalex.org/fields/27","display_name":"Medicine"},"domain":{"id":"https://openalex.org/domains/4","display_name":"Health Sciences"}},{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.07329999655485153,"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/honesty","display_name":"Honesty","score":0.5544000267982483},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.5293999910354614},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.4796000123023987},{"id":"https://openalex.org/keywords/software-deployment","display_name":"Software deployment","score":0.4490000009536743},{"id":"https://openalex.org/keywords/generalization","display_name":"Generalization","score":0.446399986743927},{"id":"https://openalex.org/keywords/function","display_name":"Function (biology)","score":0.4374000132083893}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6028000116348267},{"id":"https://openalex.org/C2777293324","wikidata":"https://www.wikidata.org/wiki/Q337349","display_name":"Honesty","level":2,"score":0.5544000267982483},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.5293999910354614},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.4796000123023987},{"id":"https://openalex.org/C105339364","wikidata":"https://www.wikidata.org/wiki/Q2297740","display_name":"Software deployment","level":2,"score":0.4490000009536743},{"id":"https://openalex.org/C177148314","wikidata":"https://www.wikidata.org/wiki/Q170084","display_name":"Generalization","level":2,"score":0.446399986743927},{"id":"https://openalex.org/C14036430","wikidata":"https://www.wikidata.org/wiki/Q3736076","display_name":"Function (biology)","level":2,"score":0.4374000132083893},{"id":"https://openalex.org/C2781249084","wikidata":"https://www.wikidata.org/wiki/Q908656","display_name":"Preference","level":2,"score":0.41280001401901245},{"id":"https://openalex.org/C204323151","wikidata":"https://www.wikidata.org/wiki/Q905424","display_name":"Range (aeronautics)","level":2,"score":0.36169999837875366},{"id":"https://openalex.org/C2775924081","wikidata":"https://www.wikidata.org/wiki/Q55608371","display_name":"Control (management)","level":2,"score":0.3125999867916107},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.31200000643730164},{"id":"https://openalex.org/C110121322","wikidata":"https://www.wikidata.org/wiki/Q865811","display_name":"Distribution (mathematics)","level":2,"score":0.3111000061035156},{"id":"https://openalex.org/C2777655017","wikidata":"https://www.wikidata.org/wiki/Q1501161","display_name":"Toolbox","level":2,"score":0.2994999885559082},{"id":"https://openalex.org/C120936955","wikidata":"https://www.wikidata.org/wiki/Q2155640","display_name":"Empirical research","level":2,"score":0.2874999940395355},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.2847999930381775},{"id":"https://openalex.org/C42475967","wikidata":"https://www.wikidata.org/wiki/Q194292","display_name":"Operations research","level":1,"score":0.25029999017715454}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1609/aaai.v40i40.40696","is_oa":true,"landing_page_url":"https://doi.org/10.1609/aaai.v40i40.40696","pdf_url":null,"source":{"id":"https://openalex.org/S4210191458","display_name":"Proceedings of the AAAI Conference on Artificial Intelligence","issn_l":"2159-5399","issn":["2159-5399","2374-3468"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/P4310320058","host_organization_name":"Association for the Advancement of Artificial Intelligence","host_organization_lineage":["https://openalex.org/P4310320058"],"host_organization_lineage_names":["Association for the Advancement of Artificial Intelligence"],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the AAAI Conference on Artificial Intelligence","raw_type":"journal-article"}],"best_oa_location":{"id":"doi:10.1609/aaai.v40i40.40696","is_oa":true,"landing_page_url":"https://doi.org/10.1609/aaai.v40i40.40696","pdf_url":null,"source":{"id":"https://openalex.org/S4210191458","display_name":"Proceedings of the AAAI Conference on Artificial Intelligence","issn_l":"2159-5399","issn":["2159-5399","2374-3468"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/P4310320058","host_organization_name":"Association for the Advancement of Artificial Intelligence","host_organization_lineage":["https://openalex.org/P4310320058"],"host_organization_lineage_names":["Association for the Advancement of Artificial Intelligence"],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the AAAI Conference on Artificial Intelligence","raw_type":"journal-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"With":[0],"the":[1,78,154,205],"rapid":[2],"advancement":[3],"of":[4,128,193],"large":[5],"language":[6],"models":[7,215],"(LLMs),":[8],"their":[9],"deployment":[10],"in":[11,195,198,202],"real-world":[12],"applications":[13],"has":[14],"become":[15],"increasingly":[16],"widespread.":[17],"LLMs":[18,39,170],"are":[19],"expected":[20],"to":[21],"deliver":[22],"robust":[23],"performance":[24,166],"across":[25,167,204,213],"diverse":[26,175,214],"tasks,":[27],"user":[28],"preferences,":[29],"and":[30,56,71,88,112,122,174,185,200,211,216],"practical":[31],"scenarios.":[32,218],"However,":[33],"as":[34],"demands":[35],"grow,":[36],"ensuring":[37],"that":[38,161],"produce":[40],"responses":[41],"aligned":[42],"with":[43,124,140,171],"human":[44,113],"intent":[45],"remains":[46],"a":[47,68,72,86,125],"foundational":[48],"challenge.":[49],"In":[50],"particular,":[51],"aligning":[52],"model":[53,99,110],"behavior":[54,111],"effectively":[55],"efficiently":[57],"during":[58,135],"inference,":[59],"without":[60,115],"costly":[61],"retraining":[62],"or":[63,137],"extensive":[64],"supervision,":[65],"is":[66,119],"both":[67],"critical":[69],"requirement":[70],"non-trivial":[73],"technical":[74],"endeavor.":[75],"To":[76],"address":[77],"challenge,":[79],"we":[80],"propose":[81],"SDA":[82,96,145,162,189],"(Steering-Driven":[83],"Distribution":[84],"Alignment),":[85],"training-free":[87],"model-agnostic":[89],"alignment":[90,105,108,142,165,181],"framework":[91],"designed":[92],"for":[93],"open-source":[94,129,169],"LLMs.":[95,130],"dynamically":[97],"redistributes":[98],"output":[100],"probabilities":[101],"based":[102],"on":[103,178],"user-defined":[104],"instructions,":[106],"enhancing":[107],"between":[109],"intents":[114],"fine-tuning.":[116],"The":[117],"method":[118],"lightweight,":[120],"resource-efficient,":[121],"compatible":[123],"wide":[126],"range":[127],"It":[131],"can":[132],"function":[133],"independently":[134],"inference":[136],"be":[138],"integrated":[139],"training-based":[141],"strategies.":[143],"Moreover,":[144],"supports":[146],"personalized":[147],"preference":[148],"alignment,":[149],"enabling":[150],"flexible":[151],"control":[152],"over":[153],"model\u2019s":[155],"response":[156],"behavior.":[157],"Empirical":[158],"results":[159],"demonstrate":[160],"consistently":[163],"improves":[164],"8":[168],"varying":[172],"scales":[173],"origins,":[176],"evaluated":[177],"three":[179],"key":[180],"dimensions,":[182],"helpfulness,":[183,196],"harmlessness,":[184],"honesty":[186,199],"(3H).":[187],"Specifically,":[188],"achieves":[190],"average":[191],"gains":[192],"64.4%":[194],"30%":[197],"11.5%":[201],"harmlessness":[203],"tested":[206],"models,":[207],"indicating":[208],"its":[209],"effectiveness":[210],"generalization":[212],"application":[217]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2026-03-18T00:00:00"}