{"id":"https://openalex.org/W4407186773","doi":"https://doi.org/10.52202/085713-2446","title":"Activation-Informed Merging of Large Language Models","display_name":"Activation-Informed Merging of Large Language Models","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4407186773","doi":"https://doi.org/10.52202/085713-2446"},"language":"en","primary_location":{"id":"doi:10.52202/085713-2446","doi":"https://doi.org/10.52202/085713-2446","is_oa":false,"landing_page_url":"https://doi.org/10.52202/085713-2446","pdf_url":null,"source":{"id":"https://openalex.org/S4363606243","display_name":"neural information processing systems","issn_l":"1049-5258","issn":["1049-5258"],"is_oa":false,"is_in_doaj":false,"is_core":true,"listed_in":["cwts-core","norway-2"],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Advances in Neural Information Processing Systems 38","raw_type":"proceedings-article"},"volume":null,"type":"conference-paper","is_traditional":true,"indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2502.02421","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5001647082","display_name":"Amin Heyrani Nobari","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I63966007","display_name":"Massachusetts Institute of Technology","ror":"https://ror.org/042nb2s44","country_code":"US","type":"education","lineage":["https://openalex.org/I63966007"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Amin Heyrani Nobari","raw_affiliation_strings":["MIT"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"MIT","institution_ids":["https://openalex.org/I63966007"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5120351636","display_name":"Kaveh Alim","orcid":"https://orcid.org/0009-0008-8013-7926","observed_orcids":["https://orcid.org/0009-0008-8013-7926"]},"institutions":[{"id":"https://openalex.org/I63966007","display_name":"Massachusetts Institute of Technology","ror":"https://ror.org/042nb2s44","country_code":"US","type":"education","lineage":["https://openalex.org/I63966007"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Kaveh Alimohammadi","raw_affiliation_strings":["MIT"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"MIT","institution_ids":["https://openalex.org/I63966007"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5041601588","display_name":"Ali ArjomandBigdeli","orcid":"https://orcid.org/0009-0003-7758-6839","observed_orcids":["https://orcid.org/0009-0003-7758-6839"]},"institutions":[{"id":"https://openalex.org/I59553526","display_name":"Stony Brook University","ror":"https://ror.org/05qghxh33","country_code":"US","type":"education","lineage":["https://openalex.org/I59553526"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Ali ArjomandBigdeli","raw_affiliation_strings":["State University of New York at Stony Brook"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"State University of New York at Stony Brook","institution_ids":["https://openalex.org/I59553526"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5022716943","display_name":"Akash Srivastava","orcid":"https://orcid.org/0000-0002-5959-3926","observed_orcids":["https://orcid.org/0000-0002-5959-3926"]},"institutions":[{"id":"https://openalex.org/I63966007","display_name":"Massachusetts Institute of Technology","ror":"https://ror.org/042nb2s44","country_code":"US","type":"education","lineage":["https://openalex.org/I63966007"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Akash Srivastava","raw_affiliation_strings":["MIT IBM"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"MIT IBM","institution_ids":["https://openalex.org/I63966007"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5040705509","display_name":"Faez Ahmed","orcid":"https://orcid.org/0000-0002-5227-2628","observed_orcids":["https://orcid.org/0000-0002-5227-2628"]},"institutions":[{"id":"https://openalex.org/I63966007","display_name":"Massachusetts Institute of Technology","ror":"https://ror.org/042nb2s44","country_code":"US","type":"education","lineage":["https://openalex.org/I63966007"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Faez Ahmed","raw_affiliation_strings":["MIT"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"MIT","institution_ids":["https://openalex.org/I63966007"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5005748450","display_name":"Navid Azizan","orcid":"https://orcid.org/0000-0002-4299-2963","observed_orcids":["https://orcid.org/0000-0002-4299-2963"]},"institutions":[{"id":"https://openalex.org/I63966007","display_name":"Massachusetts Institute of Technology","ror":"https://ror.org/042nb2s44","country_code":"US","type":"education","lineage":["https://openalex.org/I63966007"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Navid Azizan","raw_affiliation_strings":["MIT"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"MIT","institution_ids":["https://openalex.org/I63966007"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":10.6721,"has_fulltext":true,"cited_by_count":29,"citation_normalized_percentile":{"value":0.98687082,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":99,"max":100},"biblio":{"volume":null,"issue":null,"first_page":"81124","last_page":"81147"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.6514000296592712,"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":0.6514000296592712,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.2842000126838684,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.03929999843239784,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/model-merging","display_name":"model merging","score":0.8069999814033508},{"id":"https://openalex.org/keywords/large-language-models","display_name":"large language models","score":0.9829999804496765},{"id":"https://openalex.org/keywords/activation-space","display_name":"activation space","score":0.902999997138977},{"id":"https://openalex.org/keywords/continual-learning","display_name":"continual learning","score":0.9039999842643738},{"id":"https://openalex.org/keywords/model-compression","display_name":"model compression","score":0.9480000138282776},{"id":"https://openalex.org/keywords/calibration-set","display_name":"calibration set","score":0.5979999899864197}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.45722949504852295}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.52202/085713-2446","doi":"https://doi.org/10.52202/085713-2446","is_oa":false,"landing_page_url":"https://doi.org/10.52202/085713-2446","pdf_url":null,"source":{"id":"https://openalex.org/S4363606243","display_name":"neural information processing systems","issn_l":"1049-5258","issn":["1049-5258"],"is_oa":false,"is_in_doaj":false,"is_core":true,"listed_in":["cwts-core","norway-2"],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Advances in Neural Information Processing Systems 38","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2502.02421","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2502.02421","pdf_url":"https://arxiv.org/pdf/2502.02421","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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2502.02421","doi":"https://doi.org/10.48550/arxiv.2502.02421","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2502.02421","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:2502.02421","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2502.02421","pdf_url":"https://arxiv.org/pdf/2502.02421","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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4407186773.pdf","grobid_xml":"https://content.openalex.org/works/W4407186773.grobid-xml"},"referenced_works_count":10,"referenced_works":["https://openalex.org/W3083410900","https://openalex.org/W3177813494","https://openalex.org/W3195140297","https://openalex.org/W3210904070","https://openalex.org/W4303449138","https://openalex.org/W4403445272","https://openalex.org/W4406796387","https://openalex.org/W4410371768","https://openalex.org/W4415063999","https://openalex.org/W4415598828"],"related_works":["https://openalex.org/W4417298669","https://openalex.org/W4407759166","https://openalex.org/W7134986073","https://openalex.org/W4407245141","https://openalex.org/W4416789455","https://openalex.org/W6966962122","https://openalex.org/W6966972612","https://openalex.org/W4415317260","https://openalex.org/W4416034030","https://openalex.org/W7134188619"],"abstract_inverted_index":{"Model":[0],"merging,":[1],"a":[2,18,38,63,96,144],"method":[3],"that":[4,40,67,110,125],"combines":[5],"the":[6,42,45,51,82,114,127,135],"parameters":[7],"and":[8,57,92],"embeddings":[9],"of":[10,48,116],"multiple":[11,120],"fine-tuned":[12],"large":[13],"language":[14],"models":[15,118],"(LLMs),":[16],"offers":[17],"promising":[19],"approach":[20],"to":[21,54,70,77,143],"enhance":[22],"model":[23,93,136],"performance":[24,56,115],"across":[25,119],"various":[26],"tasks":[27],"while":[28],"maintaining":[29],"computational":[30],"efficiency.":[31],"This":[32],"paper":[33],"introduces":[34],"Activation-Informed":[35],"Merging":[36],"(AIM),":[37],"technique":[39],"integrates":[41],"information":[43,129],"from":[44,81,88],"activation":[46],"space":[47],"LLMs":[49],"into":[50],"merging":[52,73,137],"process":[53],"improve":[55],"robustness.":[58],"AIM":[59,100,111],"is":[60,68],"designed":[61],"as":[62],"flexible,":[64],"complementary":[65],"solution":[66],"applicable":[69],"any":[71],"existing":[72],"method.":[74],"It":[75],"aims":[76],"preserve":[78],"critical":[79],"weights":[80,104],"base":[83],"model,":[84],"drawing":[85],"on":[86],"principles":[87],"continual":[89],"learning":[90],"(CL)":[91],"compression.":[94],"Utilizing":[95],"task-agnostic":[97],"calibration":[98],"set,":[99],"selectively":[101],"prioritizes":[102],"essential":[103],"during":[105],"merging.":[106],"We":[107],"empirically":[108],"demonstrate":[109],"significantly":[112],"enhances":[113],"merged":[117],"benchmarks.":[121],"Our":[122],"findings":[123],"suggest":[124],"considering":[126],"activation-space":[128],"can":[130],"provide":[131],"substantial":[132],"advancements":[133],"in":[134,147],"strategies":[138],"for":[139],"LLMs,":[140],"with":[141],"up":[142],"40%":[145],"increase":[146],"benchmark":[148],"performance.":[149]},"counts_by_year":[{"year":2026,"cited_by_count":21},{"year":2025,"cited_by_count":8}],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2025-02-06T00:00:00"}