{"id":"https://openalex.org/W4392575397","doi":"https://doi.org/10.48550/arxiv.2403.03329","title":"Guardrail Baselines for Unlearning in LLMs","display_name":"Guardrail Baselines for Unlearning in LLMs","publication_year":2024,"publication_date":"2024-03-05","ids":{"openalex":"https://openalex.org/W4392575397","doi":"https://doi.org/10.48550/arxiv.2403.03329"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2403.03329","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2403.03329","pdf_url":"https://arxiv.org/pdf/2403.03329","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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2403.03329","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5011471473","display_name":"Pratiksha Thaker","orcid":"https://orcid.org/0000-0001-9977-5081","observed_orcids":["https://orcid.org/0000-0001-9977-5081"]},"institutions":[{"id":"https://openalex.org/I74973139","display_name":"Carnegie Mellon University","ror":"https://ror.org/05x2bcf33","country_code":"US","type":"education","lineage":["https://openalex.org/I74973139"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Thaker, Pratiksha","raw_affiliation_strings":["Department of Machine Learning Carnegie Mellon University Pittsburgh , PA 15213 , USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Machine Learning Carnegie Mellon University Pittsburgh , PA 15213 , USA","institution_ids":["https://openalex.org/I74973139"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5019615608","display_name":"Yash Maurya","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I74973139","display_name":"Carnegie Mellon University","ror":"https://ror.org/05x2bcf33","country_code":"US","type":"education","lineage":["https://openalex.org/I74973139"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Maurya, Yash","raw_affiliation_strings":["Department of Machine Learning Carnegie Mellon University Pittsburgh , PA 15213 , USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Machine Learning Carnegie Mellon University Pittsburgh , PA 15213 , USA","institution_ids":["https://openalex.org/I74973139"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Hu, Shengyuan","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I74973139","display_name":"Carnegie Mellon University","ror":"https://ror.org/05x2bcf33","country_code":"US","type":"education","lineage":["https://openalex.org/I74973139"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Hu, Shengyuan","raw_affiliation_strings":["Department of Machine Learning Carnegie Mellon University Pittsburgh , PA 15213 , USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Machine Learning Carnegie Mellon University Pittsburgh , PA 15213 , USA","institution_ids":["https://openalex.org/I74973139"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Wu, Zhiwei Steven","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wu, Zhiwei Steven","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":null,"display_name":"Smith, Virginia","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I74973139","display_name":"Carnegie Mellon University","ror":"https://ror.org/05x2bcf33","country_code":"US","type":"education","lineage":["https://openalex.org/I74973139"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Smith, Virginia","raw_affiliation_strings":["Department of Machine Learning Carnegie Mellon University Pittsburgh , PA 15213 , USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Machine Learning Carnegie Mellon University Pittsburgh , PA 15213 , USA","institution_ids":["https://openalex.org/I74973139"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":6,"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/T10764","display_name":"Privacy-Preserving Technologies in Data","score":0.8657000064849854,"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/T10764","display_name":"Privacy-Preserving Technologies in Data","score":0.8657000064849854,"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":0.05739999935030937,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.018699999898672104,"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/unlearning","display_name":"unlearning","score":0.8040000200271606},{"id":"https://openalex.org/keywords/large-language-models","display_name":"large language models","score":0.9070000052452087},{"id":"https://openalex.org/keywords/guardrails","display_name":"guardrails","score":0.8960000276565552},{"id":"https://openalex.org/keywords/prompting","display_name":"prompting","score":0.9279999732971191},{"id":"https://openalex.org/keywords/filtering","display_name":"filtering","score":0.8050000071525574},{"id":"https://openalex.org/keywords/fine-tuning","display_name":"fine-tuning","score":0.8270000219345093},{"id":"https://openalex.org/keywords/evaluation-indicators","display_name":"evaluation metrics","score":0.781000018119812},{"id":"https://openalex.org/keywords/benchmarking","display_name":"benchmarking","score":0.6399999856948853}],"concepts":[{"id":"https://openalex.org/C95457728","wikidata":"https://www.wikidata.org/wiki/Q309","display_name":"History","level":0,"score":0.37837857007980347},{"id":"https://openalex.org/C17744445","wikidata":"https://www.wikidata.org/wiki/Q36442","display_name":"Political science","level":0,"score":0.3425683379173279}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2403.03329","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2403.03329","pdf_url":"https://arxiv.org/pdf/2403.03329","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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2403.03329","doi":"https://doi.org/10.48550/arxiv.2403.03329","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2403.03329","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2403.03329","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2403.03329","pdf_url":"https://arxiv.org/pdf/2403.03329","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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"sustainable_development_goals_aurora":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4392575397.pdf","grobid_xml":"https://content.openalex.org/works/W4392575397.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W2748952813","https://openalex.org/W2949263084","https://openalex.org/W2743539335","https://openalex.org/W594353338","https://openalex.org/W2922049016","https://openalex.org/W4390697879","https://openalex.org/W2070214669","https://openalex.org/W2724734218","https://openalex.org/W4382466601","https://openalex.org/W2955504986"],"abstract_inverted_index":{"Recent":[0],"work":[1,100],"has":[2],"demonstrated":[3],"that":[4,45,63,84,107],"finetuning":[5,18,35,77],"is":[6],"a":[7,27],"promising":[8],"approach":[9],"to":[10,36,59,94],"'unlearn'":[11],"concepts":[12],"from":[13],"large":[14],"language":[15],"models.":[16],"However,":[17],"can":[19,54,108],"be":[20],"expensive,":[21],"as":[22,50,87],"it":[23],"requires":[24],"both":[25],"generating":[26],"set":[28],"of":[29,34,73,97,113],"examples":[30],"and":[31,52,117,129],"running":[32],"iterations":[33],"update":[37],"the":[38,71,95,102,111],"model.":[39],"In":[40],"this":[41],"work,":[42],"we":[43,80],"show":[44],"simple":[46],"guardrail-based":[47],"approaches":[48],"such":[49,86],"prompting":[51,88],"filtering":[53,90],"achieve":[55],"unlearning":[56],"results":[57],"comparable":[58],"finetuning.":[60],"We":[61],"recommend":[62],"researchers":[64],"investigate":[65],"these":[66],"lightweight":[67],"baselines":[68],"when":[69],"evaluating":[70],"performance":[72],"more":[74],"computationally":[75],"intensive":[76],"methods.":[78],"While":[79],"do":[81],"not":[82],"claim":[83],"methods":[85],"or":[89],"are":[91],"universal":[92],"solutions":[93],"problem":[96],"unlearning,":[98],"our":[99],"suggests":[101],"need":[103],"for":[104],"evaluation":[105],"metrics":[106,128],"better":[109],"separate":[110],"power":[112],"guardrails":[114,121],"vs.":[115],"finetuning,":[116],"highlights":[118],"scenarios":[119],"where":[120],"expose":[122],"possible":[123],"unintended":[124],"behavior":[125],"in":[126],"existing":[127],"benchmarks.":[130]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2025,"cited_by_count":4},{"year":2024,"cited_by_count":1}],"updated_date":"2026-10-16T09:59:22.235173","created_date":"2025-10-10T00:00:00"}