{"id":"https://openalex.org/W4416036017","doi":"https://doi.org/10.18653/v1/2025.emnlp-main.1056","title":"GAP: a Global Adaptive Pruning Method for Large Language Models","display_name":"GAP: a Global Adaptive Pruning Method for Large Language Models","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4416036017","doi":"https://doi.org/10.18653/v1/2025.emnlp-main.1056"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2025.emnlp-main.1056","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.emnlp-main.1056","pdf_url":"https://aclanthology.org/2025.emnlp-main.1056.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":"Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.emnlp-main.1056.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5049225898","display_name":"Zhihua Ban","orcid":"https://orcid.org/0000-0002-3209-4916"},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Zhihua Ban","raw_affiliation_strings":["CVTE Research"],"affiliations":[{"raw_affiliation_string":"CVTE Research","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5103042414","display_name":"Haotian Ma","orcid":"https://orcid.org/0000-0003-2114-5586"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Haotian Ma","raw_affiliation_strings":["CVTE Research"],"affiliations":[{"raw_affiliation_string":"CVTE Research","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5086412442","display_name":"Siheng Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Siheng Zhang","raw_affiliation_strings":["CVTE Research"],"affiliations":[{"raw_affiliation_string":"CVTE Research","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100739950","display_name":"Shengyu Liu","orcid":"https://orcid.org/0000-0002-4028-7793"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shengyu Liu","raw_affiliation_strings":["CVTE Research"],"affiliations":[{"raw_affiliation_string":"CVTE Research","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5022146103","display_name":"Xichen Chen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xichen Chen","raw_affiliation_strings":["CVTE Research"],"affiliations":[{"raw_affiliation_string":"CVTE Research","institution_ids":[]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5019934281","display_name":"Ming Yang","orcid":"https://orcid.org/0000-0003-1691-6817"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ming Yang","raw_affiliation_strings":["CVTE Research"],"affiliations":[{"raw_affiliation_string":"CVTE Research","institution_ids":[]}]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":6,"corresponding_author_ids":["https://openalex.org/A5049225898"],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.18397108,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"20909","last_page":"20914"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.38350000977516174,"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":0.38350000977516174,"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/T10201","display_name":"Speech Recognition and Synthesis","score":0.1225999966263771,"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":0.09049999713897705,"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/pruning","display_name":"Pruning","score":0.4453999996185303},{"id":"https://openalex.org/keywords/feature","display_name":"Feature (linguistics)","score":0.32120001316070557},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.2793000042438507},{"id":"https://openalex.org/keywords/identification","display_name":"Identification (biology)","score":0.257099986076355},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.2556999921798706}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6014000177383423},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5843999981880188},{"id":"https://openalex.org/C108010975","wikidata":"https://www.wikidata.org/wiki/Q500094","display_name":"Pruning","level":2,"score":0.4453999996185303},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.32120001316070557},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.30250000953674316},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.2953999936580658},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.2793000042438507},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.2775999903678894},{"id":"https://openalex.org/C116834253","wikidata":"https://www.wikidata.org/wiki/Q2039217","display_name":"Identification (biology)","level":2,"score":0.257099986076355},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.2556999921798706},{"id":"https://openalex.org/C527412718","wikidata":"https://www.wikidata.org/wiki/Q855395","display_name":"Interpretation (philosophy)","level":2,"score":0.2540000081062317},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.2522999942302704},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.25200000405311584}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.emnlp-main.1056","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.emnlp-main.1056","pdf_url":"https://aclanthology.org/2025.emnlp-main.1056.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":"Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.emnlp-main.1056","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.emnlp-main.1056","pdf_url":"https://aclanthology.org/2025.emnlp-main.1056.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":"Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4416036017.pdf","grobid_xml":"https://content.openalex.org/works/W4416036017.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"The":[0],"deployment":[1],"of":[2,36],"Large":[3],"Language":[4],"Models":[5],"(LLMs)":[6],"faces":[7],"significant":[8,112],"challenges":[9],"due":[10],"to":[11,83],"high":[12,105],"computational":[13],"costs,":[14],"driving":[15],"the":[16,33,61,126,130],"demand":[17],"for":[18],"effective":[19],"pruning":[20,23,57,62,106,116],"techniques.Existing":[21],"structured":[22],"methods":[24,103],"employ":[25],"uniform":[26],"compression":[27,87],"rates":[28,107,117],"across":[29],"network":[30,38],"layers,":[31],"neglecting":[32],"varying":[34],"importance":[35],"different":[37],"depths.To":[39],"address":[40],"this":[41],"limitation,":[42],"we":[43],"propose":[44],"a":[45,65,71],"novel":[46],"optimization":[47,67],"framework":[48,59],"that":[49],"directly":[50],"minimizes":[51],"global":[52],"capability":[53],"loss":[54],"through":[55],"layer-adaptive":[56],"rates.The":[58],"formulates":[60],"task":[63],"as":[64],"combinatorial":[66],"problem":[68],"constrained":[69],"by":[70],"total":[72],"parameter":[73],"budget,":[74],"and":[75,110],"an":[76],"efficient":[77],"dynamic":[78],"programming":[79],"solution":[80],"is":[81,93,121],"derived":[82],"determine":[84],"optimal":[85],"layerwise":[86],"rates.Experiments":[88],"demonstrate":[89],"that,":[90],"when":[91],"tuning":[92,120],"not":[94],"included,":[95,122],"our":[96,123],"approach":[97],"achieves":[98,125],"comparable":[99],"performance":[100,128],"with":[101],"state-of-the-art":[102],"at":[104,114],"(37-50%":[108],"reduction),":[109],"shows":[111],"advantages":[113],"low":[115],"(13-25%":[118],"reduction).When":[119],"method":[124],"best":[127],"among":[129],"compared":[131],"methods.":[132]},"counts_by_year":[],"updated_date":"2026-03-11T06:11:40.159057","created_date":"2025-11-08T00:00:00"}