{"id":"https://openalex.org/W4412944934","doi":"https://doi.org/10.18653/v1/2025.acl-long.1055","title":"LED-Merging: Mitigating Safety-Utility Conflicts in Model Merging with Location-Election-Disjoint","display_name":"LED-Merging: Mitigating Safety-Utility Conflicts in Model Merging with Location-Election-Disjoint","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4412944934","doi":"https://doi.org/10.18653/v1/2025.acl-long.1055"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.acl-long.1055","is_oa":false,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.1055","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["arxiv","crossref"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2502.16770","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5100348018","display_name":"Qianli Ma","orcid":"https://orcid.org/0000-0003-4504-1933"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Qianli Ma","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5020653216","display_name":"Dongrui Liu","orcid":"https://orcid.org/0000-0003-0087-1124"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Dongrui Liu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100428466","display_name":"Qian Chen","orcid":"https://orcid.org/0000-0002-3873-8022"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Qian Chen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100689116","display_name":"Linfeng Zhang","orcid":"https://orcid.org/0000-0002-7740-7971"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Linfeng Zhang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5100903630","display_name":"Shao Jing","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jing Shao","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":false,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"21749","last_page":"21767"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11942","display_name":"Transportation and Mobility Innovations","score":0.9983000159263611,"subfield":{"id":"https://openalex.org/subfields/2203","display_name":"Automotive Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T11942","display_name":"Transportation and Mobility Innovations","score":0.9983000159263611,"subfield":{"id":"https://openalex.org/subfields/2203","display_name":"Automotive Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T11450","display_name":"Model-Driven Software Engineering Techniques","score":0.9190000295639038,"subfield":{"id":"https://openalex.org/subfields/1712","display_name":"Software"},"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/T10456","display_name":"Multi-Agent Systems and Negotiation","score":0.9157000184059143,"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/disjoint-sets","display_name":"Disjoint sets","score":0.7415703535079956},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.6462439298629761},{"id":"https://openalex.org/keywords/computer-security","display_name":"Computer security","score":0.49769285321235657},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.06502622365951538}],"concepts":[{"id":"https://openalex.org/C45340560","wikidata":"https://www.wikidata.org/wiki/Q215382","display_name":"Disjoint sets","level":2,"score":0.7415703535079956},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6462439298629761},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.49769285321235657},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.06502622365951538},{"id":"https://openalex.org/C114614502","wikidata":"https://www.wikidata.org/wiki/Q76592","display_name":"Combinatorics","level":1,"score":0.0}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.18653/v1/2025.acl-long.1055","is_oa":false,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.1055","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2502.16770","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2502.16770","pdf_url":"https://arxiv.org/pdf/2502.16770","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"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"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2502.16770","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2502.16770","pdf_url":"https://arxiv.org/pdf/2502.16770","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"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":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W4256429076","https://openalex.org/W1971174658","https://openalex.org/W2099195351","https://openalex.org/W2348092930","https://openalex.org/W2012063272","https://openalex.org/W2390279801","https://openalex.org/W3129272755"],"abstract_inverted_index":{"Fine-tuning":[0],"pre-trained":[1],"Large":[2],"Language":[3],"Models":[4],"(LLMs)":[5],"for":[6,138],"specialized":[7],"tasks":[8],"incurs":[9],"substantial":[10],"computational":[11],"and":[12,53,84,96,132],"data":[13],"costs.":[14],"While":[15],"model":[16],"merging":[17],"offers":[18],"a":[19,66,107,134],"training-free":[20,136],"solution":[21],"to":[22,48],"integrate":[23],"multiple":[24],"task-specific":[25,71],"models,":[26],"existing":[27],"methods":[28],"suffer":[29],"from":[30],"safety-utility":[31,130],"conflicts":[32,131],"where":[33],"enhanced":[34],"general":[35],"capabilities":[36],"degrade":[37],"safety":[38],"safeguards.":[39],"We":[40],"identify":[41],"two":[42],"root":[43],"causes:":[44],"$\\textbf{neuron":[45],"misidentification}$":[46],"due":[47],"simplistic":[49],"parameter":[50,89],"magnitude-based":[51],"selection,":[52],"$\\textbf{cross-task":[54],"neuron":[55],"interference}$":[56],"during":[57],"merging.":[58],"To":[59],"address":[60],"these":[61],"challenges,":[62],"we":[63],"propose":[64],"$\\textbf{LED-Merging}$,":[65],"three-stage":[67],"framework":[68],"that":[69,99],"$\\textbf{L}$ocates":[70],"neurons":[72,79],"via":[73],"gradient-based":[74],"attribution,":[75],"dynamically":[76],"$\\textbf{E}$lects":[77],"critical":[78],"through":[80,88],"multi-model":[81],"importance":[82],"fusion,":[83],"$\\textbf{D}$isjoints":[85],"conflicting":[86],"updates":[87],"isolation.":[90],"Extensive":[91],"experiments":[92],"on":[93,110,112,126],"Llama-3-8B,":[94],"Mistral-7B,":[95],"Llama2-13B":[97],"demonstrate":[98],"LED-Merging":[100,128],"effectively":[101],"reduces":[102],"harmful":[103],"response":[104],"rates,":[105],"showing":[106],"31.4\\%":[108],"decrease":[109],"Llama-3-8B-Instruct":[111],"HarmBench,":[113],"while":[114],"simultaneously":[115],"preserving":[116],"95\\%":[117],"of":[118],"utility":[119],"performance,":[120],"such":[121],"as":[122],"achieving":[123],"52.39\\%":[124],"accuracy":[125],"GSM8K.":[127],"resolves":[129],"provides":[133],"lightweight,":[135],"paradigm":[137],"constructing":[139],"reliable":[140],"multi-task":[141],"LLMs.":[142],"Code":[143],"is":[144],"available":[145],"at":[146],"$\\href{https://github.com/MqLeet/LED-Merging}{GitHub}$.":[147]},"counts_by_year":[],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}