{"id":"https://openalex.org/W4304731097","doi":"https://doi.org/10.48550/arxiv.2210.04492","title":"Unified Detoxifying and Debiasing in Language Generation via Inference-time Adaptive Optimization","display_name":"Unified Detoxifying and Debiasing in Language Generation via Inference-time Adaptive Optimization","publication_year":2022,"publication_date":"2022-10-10","ids":{"openalex":"https://openalex.org/W4304731097","doi":"https://doi.org/10.48550/arxiv.2210.04492"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2210.04492","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2210.04492","pdf_url":"https://arxiv.org/pdf/2210.04492","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by-sa","license_id":"https://openalex.org/licenses/cc-by-sa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2210.04492","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5042856545","display_name":"Zonghan Yang","orcid":"https://orcid.org/0000-0001-5774-5298","observed_orcids":["https://orcid.org/0000-0001-5774-5298"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yang, Zonghan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5040961637","display_name":"Xiaoyuan Yi","orcid":"https://orcid.org/0000-0003-2710-1613","observed_orcids":["https://orcid.org/0000-0003-2710-1613"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yi, Xiaoyuan","raw_affiliation_strings":["Microsoft Research Asia ;"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Microsoft Research Asia ;","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100432795","display_name":"Peng Li","orcid":"https://orcid.org/0000-0003-4981-0496","observed_orcids":["https://orcid.org/0000-0003-4981-0496"]},"institutions":[{"id":"https://openalex.org/I4391012619","display_name":"Shanghai Artificial Intelligence Laboratory","ror":"https://ror.org/03wkvpx79","country_code":"CN","type":"facility","lineage":["https://openalex.org/I4391012619"]},{"id":"https://openalex.org/I99065089","display_name":"Tsinghua University","ror":"https://ror.org/03cve4549","country_code":"CN","type":"education","lineage":["https://openalex.org/I99065089"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Li, Peng","raw_affiliation_strings":["Institute for AI Industry Research , Tsinghua University , Beijing , China;","Shanghai Artificial Intelligence Laboratory , Shanghai , China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Institute for AI Industry Research , Tsinghua University , Beijing , China;","institution_ids":["https://openalex.org/I99065089"]},{"raw_affiliation_string":"Shanghai Artificial Intelligence Laboratory , Shanghai , China","institution_ids":["https://openalex.org/I4391012619"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Liu, Yang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Yang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5044651577","display_name":"Xing Xie","orcid":"https://orcid.org/0000-0002-8608-8482","observed_orcids":["https://orcid.org/0000-0002-8608-8482","https://orcid.org/0000-0001-5139-4634","https://orcid.org/0000-0001-7615-9673"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xie, Xing","raw_affiliation_strings":["Microsoft Research Asia ;"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Microsoft Research Asia ;","institution_ids":[]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":23,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T12262","display_name":"Hate Speech and Cyberbullying Detection","score":0.5435000061988831,"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/T12262","display_name":"Hate Speech and Cyberbullying Detection","score":0.5435000061988831,"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/T10883","display_name":"Ethics and Social Impacts of AI","score":0.32499998807907104,"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":0.06870000064373016,"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/detoxification","display_name":"detoxification","score":0.6600000262260437},{"id":"https://openalex.org/keywords/bias-mitigation","display_name":"bias mitigation","score":0.9850000143051147},{"id":"https://openalex.org/keywords/pre-trained-language-models","display_name":"pre-trained language models","score":0.7519999742507935},{"id":"https://openalex.org/keywords/inference-time-optimization","display_name":"inference-time optimization","score":0.6460000276565552},{"id":"https://openalex.org/keywords/parameter-efficient-fine-tuning","display_name":"parameter-efficient fine-tuning","score":0.8920000195503235},{"id":"https://openalex.org/keywords/toxicity","display_name":"toxicity","score":0.34200000762939453},{"id":"https://openalex.org/keywords/social-bias","display_name":"social bias","score":0.9100000262260437}],"concepts":[{"id":"https://openalex.org/C2779458634","wikidata":"https://www.wikidata.org/wiki/Q24963715","display_name":"Debiasing","level":2,"score":0.9725879430770874},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6543384194374084},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6107686161994934},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.3687446117401123},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.24322965741157532},{"id":"https://openalex.org/C15744967","wikidata":"https://www.wikidata.org/wiki/Q9418","display_name":"Psychology","level":0,"score":0.10599470138549805},{"id":"https://openalex.org/C188147891","wikidata":"https://www.wikidata.org/wiki/Q147638","display_name":"Cognitive science","level":1,"score":0.09581038355827332}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2210.04492","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2210.04492","pdf_url":"https://arxiv.org/pdf/2210.04492","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by-sa","license_id":"https://openalex.org/licenses/cc-by-sa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2210.04492","doi":"https://doi.org/10.48550/arxiv.2210.04492","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2210.04492","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2210.04492","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2210.04492","pdf_url":"https://arxiv.org/pdf/2210.04492","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by-sa","license_id":"https://openalex.org/licenses/cc-by-sa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[{"id":"https://openalex.org/G1111291411","display_name":null,"funder_award_id":"62276152","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G1681167796","display_name":null,"funder_award_id":"62236011","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G7879634455","display_name":null,"funder_award_id":"61925601","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/W4304731097.pdf","grobid_xml":"https://content.openalex.org/works/W4304731097.grobid-xml"},"referenced_works_count":38,"referenced_works":["https://openalex.org/W2798812533","https://openalex.org/W2885765530","https://openalex.org/W2942370121","https://openalex.org/W2950866572","https://openalex.org/W2963524349","https://openalex.org/W2971307358","https://openalex.org/W2975003218","https://openalex.org/W2981852735","https://openalex.org/W2982399380","https://openalex.org/W2988937804","https://openalex.org/W3015919249","https://openalex.org/W3017961061","https://openalex.org/W3019416653","https://openalex.org/W3035591180","https://openalex.org/W3085190015","https://openalex.org/W3088599783","https://openalex.org/W3099635335","https://openalex.org/W3100355408","https://openalex.org/W3101767999","https://openalex.org/W3117655171","https://openalex.org/W3155742828","https://openalex.org/W3172314079","https://openalex.org/W3174685870","https://openalex.org/W3195577433","https://openalex.org/W3199369541","https://openalex.org/W3199408738","https://openalex.org/W4221144473","https://openalex.org/W4224912576","https://openalex.org/W4226092640","https://openalex.org/W4226112058","https://openalex.org/W4226146865","https://openalex.org/W4287211008","https://openalex.org/W4287887510","https://openalex.org/W4294808066","https://openalex.org/W4319862478","https://openalex.org/W4385572928","https://openalex.org/W7133207027","https://openalex.org/W7133224479"],"related_works":["https://openalex.org/W4285116834","https://openalex.org/W4389524389","https://openalex.org/W3134354193","https://openalex.org/W4401042465","https://openalex.org/W4402683184","https://openalex.org/W4411446580","https://openalex.org/W4408355569","https://openalex.org/W7123754617","https://openalex.org/W7160335338","https://openalex.org/W4225929875"],"abstract_inverted_index":{"Warning:":[0],"this":[1],"paper":[2],"contains":[3],"model":[4],"outputs":[5],"exhibiting":[6],"offensiveness":[7],"and":[8,40,50,64,98,155,175,178,182],"biases.":[9,84],"Recently":[10],"pre-trained":[11],"language":[12,20,49],"models":[13,35,74],"(PLMs)":[14],"have":[15],"prospered":[16],"in":[17,44],"various":[18],"natural":[19],"generation":[21,152],"(NLG)":[22],"tasks":[23],"due":[24],"to":[25,28,38,150,168],"their":[26],"ability":[27],"generate":[29],"fairly":[30],"fluent":[31],"text.":[32],"Nevertheless,":[33],"these":[34,105],"are":[36],"observed":[37],"capture":[39],"reproduce":[41],"harmful":[42],"contents":[43],"training":[45,146],"corpora,":[46],"typically":[47],"toxic":[48],"social":[51,83],"biases,":[52],"raising":[53],"severe":[54],"moral":[55],"issues.":[56],"Prior":[57],"works":[58],"on":[59,139],"ethical":[60,190],"NLG":[61],"tackle":[62],"detoxifying":[63,97,176],"debiasing":[65,99,174],"separately,":[66],"which":[67,102],"is":[68],"problematic":[69],"since":[70],"we":[71,90],"find":[72],"debiased":[73],"still":[75],"exhibit":[76],"toxicity":[77],"while":[78],"detoxified":[79],"ones":[80],"even":[81],"exacerbate":[82],"To":[85],"address":[86],"such":[87],"a":[88,120,133,140,185],"challenge,":[89],"propose":[91],"the":[92,110],"first":[93],"unified":[94],"framework":[95,117],"of":[96,131],"called":[100],"UDDIA,":[101],"jointly":[103],"formalizes":[104],"two":[106],"problems":[107],"as":[108,118],"rectifying":[109],"output":[111],"space.":[112],"We":[113],"theoretically":[114],"interpret":[115],"our":[116],"learning":[119],"text":[121],"distribution":[122],"mixing":[123],"weighted":[124],"attributes.":[125],"Besides,":[126],"UDDIA":[127,172],"conducts":[128],"adaptive":[129],"optimization":[130],"only":[132],"few":[134],"parameters":[135],"during":[136],"decoding":[137],"based":[138],"parameter-efficient":[141],"tuning":[142],"schema":[143],"without":[144],"any":[145],"data.":[147],"This":[148],"leads":[149],"minimal":[151],"quality":[153],"loss":[154],"improved":[156],"rectification":[157],"performance":[158],"with":[159],"acceptable":[160],"computational":[161],"cost.":[162],"Experimental":[163],"results":[164],"demonstrate":[165],"that":[166],"compared":[167],"several":[169],"strong":[170],"baselines,":[171],"achieves":[173],"simultaneously":[177],"better":[179],"balances":[180],"efficiency":[181],"effectiveness,":[183],"taking":[184],"further":[186],"step":[187],"towards":[188],"practical":[189],"NLG.":[191]},"counts_by_year":[{"year":2026,"cited_by_count":3},{"year":2025,"cited_by_count":10},{"year":2024,"cited_by_count":4},{"year":2023,"cited_by_count":6}],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2022-10-12T00:00:00"}