{"id":"https://openalex.org/W4416034899","doi":"https://doi.org/10.18653/v1/2025.findings-emnlp.317","title":"Constructing Your Model\u2019s Value Distinction: Towards LLM Alignment with Anchor Words Tuning","display_name":"Constructing Your Model\u2019s Value Distinction: Towards LLM Alignment with Anchor Words Tuning","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4416034899","doi":"https://doi.org/10.18653/v1/2025.findings-emnlp.317"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2025.findings-emnlp.317","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.317","pdf_url":"https://aclanthology.org/2025.findings-emnlp.317.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: EMNLP 2025","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.findings-emnlp.317.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5021693730","display_name":"Zhen Yang","orcid":"https://orcid.org/0000-0003-0670-4538"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhen Yang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101796595","display_name":"Ping Jian","orcid":"https://orcid.org/0000-0001-7236-2922"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ping Jian","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102010141","display_name":"Chengzhi Li","orcid":"https://orcid.org/0000-0002-2373-5638"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chengzhi Li","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5010390467","display_name":"Chenxu Wang","orcid":"https://orcid.org/0009-0009-0751-0242"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chenxu Wang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100319963","display_name":"Xinyue Zhang","orcid":"https://orcid.org/0000-0002-4833-6481"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xinyue Zhang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5076564877","display_name":"Wenpeng L\u00fc","orcid":"https://orcid.org/0000-0002-1840-3540"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wenpeng Lu","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":null,"has_fulltext":true,"cited_by_count":1,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"5932","last_page":"5948"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.4526999890804291,"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":0.4526999890804291,"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":0.09210000187158585,"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/T10215","display_name":"Semantic Web and Ontologies","score":0.05889999866485596,"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/value","display_name":"Value (mathematics)","score":0.5414999723434448},{"id":"https://openalex.org/keywords/feature","display_name":"Feature (linguistics)","score":0.3052999973297119},{"id":"https://openalex.org/keywords/term","display_name":"Term (time)","score":0.26759999990463257},{"id":"https://openalex.org/keywords/class","display_name":"Class (philosophy)","score":0.26179999113082886},{"id":"https://openalex.org/keywords/noise","display_name":"Noise (video)","score":0.260699987411499}],"concepts":[{"id":"https://openalex.org/C2776291640","wikidata":"https://www.wikidata.org/wiki/Q2912517","display_name":"Value (mathematics)","level":2,"score":0.5414999723434448},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.508899986743927},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.40619999170303345},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3862999975681305},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.3488999903202057},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.3052999973297119},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.28049999475479126},{"id":"https://openalex.org/C61797465","wikidata":"https://www.wikidata.org/wiki/Q1188986","display_name":"Term (time)","level":2,"score":0.26759999990463257},{"id":"https://openalex.org/C2777212361","wikidata":"https://www.wikidata.org/wiki/Q5127848","display_name":"Class (philosophy)","level":2,"score":0.26179999113082886},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.260699987411499},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.25440001487731934}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.findings-emnlp.317","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.317","pdf_url":"https://aclanthology.org/2025.findings-emnlp.317.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: EMNLP 2025","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.findings-emnlp.317","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-emnlp.317","pdf_url":"https://aclanthology.org/2025.findings-emnlp.317.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: EMNLP 2025","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G917634060","display_name":null,"funder_award_id":"62376130","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"}],"funders":[{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4416034899.pdf","grobid_xml":"https://content.openalex.org/works/W4416034899.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"With":[0],"the":[1,38,42,48,78,92,101,151,169,175,190],"widespread":[2],"applications":[3],"of":[4,36,80,130],"large":[5],"language":[6],"models":[7],"(LLMs),":[8],"aligning":[9],"LLMs":[10,24,31,67,104,124],"with":[11],"human":[12,55],"values":[13],"has":[14],"emerged":[15],"as":[16,47,184],"a":[17,61,70,108,127,163],"critical":[18],"challenge.For":[19],"alignment,":[20,65,197],"we":[21,85,161],"always":[22],"expect":[23],"to":[25,33,100,140,205],"be":[26,34],"honest,":[27],"positive,":[28],"harmless,":[29],"etc.And":[30],"appear":[32],"capable":[35],"generating":[37],"desired":[39],"outputs":[40],"after":[41,64,105,134],"alignment":[43,106,142,207],"tuning":[44,50],"process,":[45],"such":[46,146],"preference":[49],"via":[51],"reinforcement":[52],"learning":[53],"from":[54,91],"feedback":[56],"(RLHF).However,":[57],"it":[58,136],"also":[59,137,199],"raises":[60],"question":[62,90],"about":[63],"do":[66,125],"genuinely":[68],"obtain":[69,126],"value":[71,128,147],"distinction":[72,129],"between":[73,112,178],"positives":[74,131],"and":[75,114,132,180],"negatives,":[76],"beyond":[77],"generation":[79,119],"positive":[81,113,179],"outputs?In":[82],"this":[83,89],"work,":[84],"start":[86],"by":[87,143,158,173],"investigating":[88],"token":[93],"distribution":[94],"perspective.Our":[95],"findings":[96],"reveal":[97],"that":[98,123,167,189],"compared":[99,204],"unaligned":[102],"versions,":[103],"exhibit":[107],"larger":[109],"logits":[110,176],"gap":[111],"negative":[115,181],"tokens":[116,182],"at":[117],"each":[118],"step,":[120],"which":[121],"suggests":[122],"negatives":[133],"alignment.Meanwhile,":[135],"motivates":[138],"us":[139],"achieve":[141],"directly":[144],"constructing":[145],"distinction,":[148],"thus":[149],"alleviating":[150],"excessive":[152],"reliance":[153],"on":[154],"computational":[155,202],"resources":[156,203],"required":[157],"training-time":[159,206],"alignment.Specifically,":[160],"propose":[162],"representation":[164,172],"editing":[165],"method":[166,192],"intervenes":[168],"last":[170],"hidden":[171],"amplifying":[174],"difference":[177],"(defined":[183],"anchor":[185],"words).Experimental":[186],"results":[187],"demonstrate":[188],"proposed":[191],"not":[193],"only":[194],"achieves":[195],"effective":[196],"but":[198],"requires":[200],"fewer":[201],"methods":[208],"1":[209],".":[210]},"counts_by_year":[{"year":2025,"cited_by_count":1}],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-11-08T00:00:00"}