{"id":"https://openalex.org/W7138314312","doi":"https://doi.org/10.1609/aaai.v40i24.39120","title":"Prune&Comp: Free Lunch for Layer-Pruned LLMs via Iterative Pruning with Magnitude Compensation","display_name":"Prune&Comp: Free Lunch for Layer-Pruned LLMs via Iterative Pruning with Magnitude Compensation","publication_year":2026,"publication_date":"2026-03-14","ids":{"openalex":"https://openalex.org/W7138314312","doi":"https://doi.org/10.1609/aaai.v40i24.39120"},"language":null,"primary_location":{"id":"doi:10.1609/aaai.v40i24.39120","is_oa":true,"landing_page_url":"https://doi.org/10.1609/aaai.v40i24.39120","pdf_url":"https://ojs.aaai.org/index.php/AAAI/article/download/39120/43082","source":{"id":"https://openalex.org/S4210191458","display_name":"Proceedings of the AAAI Conference on Artificial Intelligence","issn_l":"2159-5399","issn":["2159-5399","2374-3468"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/P4310320058","host_organization_name":"Association for the Advancement of Artificial Intelligence","host_organization_lineage":["https://openalex.org/P4310320058"],"host_organization_lineage_names":["Association for the Advancement of Artificial Intelligence"],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the AAAI Conference on Artificial Intelligence","raw_type":"journal-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"diamond","oa_url":"https://ojs.aaai.org/index.php/AAAI/article/download/39120/43082","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5129681927","display_name":"Xinrui Chen","orcid":null},"institutions":[{"id":"https://openalex.org/I3131625388","display_name":"University Town of Shenzhen","ror":"https://ror.org/05f5j6225","country_code":"CN","type":"education","lineage":["https://openalex.org/I3131625388"]},{"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":true,"raw_author_name":"Xinrui Chen","raw_affiliation_strings":["Shenzhen International Graduate School, Tsinghua University"],"affiliations":[{"raw_affiliation_string":"Shenzhen International Graduate School, Tsinghua University","institution_ids":["https://openalex.org/I3131625388","https://openalex.org/I99065089"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129727440","display_name":"Hongxing Zhang","orcid":null},"institutions":[{"id":"https://openalex.org/I186272606","display_name":"Guangdong University of Foreign Studies","ror":"https://ror.org/00fhc9y79","country_code":"CN","type":"education","lineage":["https://openalex.org/I186272606"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Hongxing Zhang","raw_affiliation_strings":["School of Information Science and Technology, Guangdong University of Foreign Studies"],"affiliations":[{"raw_affiliation_string":"School of Information Science and Technology, Guangdong University of Foreign Studies","institution_ids":["https://openalex.org/I186272606"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5113367655","display_name":"Fanyi Zeng","orcid":null},"institutions":[{"id":"https://openalex.org/I3131625388","display_name":"University Town of Shenzhen","ror":"https://ror.org/05f5j6225","country_code":"CN","type":"education","lineage":["https://openalex.org/I3131625388"]},{"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":"Fanyi Zeng","raw_affiliation_strings":["Shenzhen International Graduate School, Tsinghua University"],"affiliations":[{"raw_affiliation_string":"Shenzhen International Graduate School, Tsinghua University","institution_ids":["https://openalex.org/I3131625388","https://openalex.org/I99065089"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129722869","display_name":"Yongxian Wei","orcid":null},"institutions":[{"id":"https://openalex.org/I3131625388","display_name":"University Town of Shenzhen","ror":"https://ror.org/05f5j6225","country_code":"CN","type":"education","lineage":["https://openalex.org/I3131625388"]},{"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":"Yongxian Wei","raw_affiliation_strings":["Shenzhen International Graduate School, Tsinghua University"],"affiliations":[{"raw_affiliation_string":"Shenzhen International Graduate School, Tsinghua University","institution_ids":["https://openalex.org/I3131625388","https://openalex.org/I99065089"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129718863","display_name":"Yizhi Wang","orcid":null},"institutions":[{"id":"https://openalex.org/I3131625388","display_name":"University Town of Shenzhen","ror":"https://ror.org/05f5j6225","country_code":"CN","type":"education","lineage":["https://openalex.org/I3131625388"]},{"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":"Yizhi Wang","raw_affiliation_strings":["Shenzhen International Graduate School, Tsinghua University"],"affiliations":[{"raw_affiliation_string":"Shenzhen International Graduate School, Tsinghua University","institution_ids":["https://openalex.org/I3131625388","https://openalex.org/I99065089"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5111335051","display_name":"Xitong Ling","orcid":null},"institutions":[{"id":"https://openalex.org/I3131625388","display_name":"University Town of Shenzhen","ror":"https://ror.org/05f5j6225","country_code":"CN","type":"education","lineage":["https://openalex.org/I3131625388"]},{"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":"Xitong Ling","raw_affiliation_strings":["Shenzhen International Graduate School, Tsinghua University"],"affiliations":[{"raw_affiliation_string":"Shenzhen International Graduate School, Tsinghua University","institution_ids":["https://openalex.org/I3131625388","https://openalex.org/I99065089"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129708789","display_name":"Guanghao Li","orcid":null},"institutions":[{"id":"https://openalex.org/I3131625388","display_name":"University Town of Shenzhen","ror":"https://ror.org/05f5j6225","country_code":"CN","type":"education","lineage":["https://openalex.org/I3131625388"]},{"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":"Guanghao Li","raw_affiliation_strings":["Shenzhen International Graduate School, Tsinghua University"],"affiliations":[{"raw_affiliation_string":"Shenzhen International Graduate School, Tsinghua University","institution_ids":["https://openalex.org/I3131625388","https://openalex.org/I99065089"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5129678500","display_name":"Chun Yuan","orcid":null},"institutions":[{"id":"https://openalex.org/I3131625388","display_name":"University Town of Shenzhen","ror":"https://ror.org/05f5j6225","country_code":"CN","type":"education","lineage":["https://openalex.org/I3131625388"]},{"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":"Chun Yuan","raw_affiliation_strings":["Shenzhen International Graduate School, Tsinghua University"],"affiliations":[{"raw_affiliation_string":"Shenzhen International Graduate School, Tsinghua University","institution_ids":["https://openalex.org/I3131625388","https://openalex.org/I99065089"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":8,"corresponding_author_ids":["https://openalex.org/A5129681927"],"corresponding_institution_ids":["https://openalex.org/I3131625388","https://openalex.org/I99065089"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.5267666,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":"40","issue":"24","first_page":"20316","last_page":"20324"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.6888999938964844,"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":0.6888999938964844,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.1160999983549118,"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":0.0430000014603138,"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":0.7856000065803528},{"id":"https://openalex.org/keywords/magnitude","display_name":"Magnitude (astronomy)","score":0.6970000267028809},{"id":"https://openalex.org/keywords/layer","display_name":"Layer (electronics)","score":0.5960999727249146},{"id":"https://openalex.org/keywords/iterative-method","display_name":"Iterative method","score":0.5626999735832214},{"id":"https://openalex.org/keywords/compensation","display_name":"Compensation (psychology)","score":0.5616000294685364},{"id":"https://openalex.org/keywords/stability","display_name":"Stability (learning theory)","score":0.5522000193595886},{"id":"https://openalex.org/keywords/acceleration","display_name":"Acceleration","score":0.5497000217437744},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.3758000135421753}],"concepts":[{"id":"https://openalex.org/C108010975","wikidata":"https://www.wikidata.org/wiki/Q500094","display_name":"Pruning","level":2,"score":0.7856000065803528},{"id":"https://openalex.org/C126691448","wikidata":"https://www.wikidata.org/wiki/Q2028919","display_name":"Magnitude (astronomy)","level":2,"score":0.6970000267028809},{"id":"https://openalex.org/C2779227376","wikidata":"https://www.wikidata.org/wiki/Q6505497","display_name":"Layer (electronics)","level":2,"score":0.5960999727249146},{"id":"https://openalex.org/C159694833","wikidata":"https://www.wikidata.org/wiki/Q2321565","display_name":"Iterative method","level":2,"score":0.5626999735832214},{"id":"https://openalex.org/C2780023022","wikidata":"https://www.wikidata.org/wiki/Q1338171","display_name":"Compensation (psychology)","level":2,"score":0.5616000294685364},{"id":"https://openalex.org/C112972136","wikidata":"https://www.wikidata.org/wiki/Q7595718","display_name":"Stability (learning theory)","level":2,"score":0.5522000193595886},{"id":"https://openalex.org/C117896860","wikidata":"https://www.wikidata.org/wiki/Q11376","display_name":"Acceleration","level":2,"score":0.5497000217437744},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5038999915122986},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.4729999899864197},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.39149999618530273},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.3758000135421753},{"id":"https://openalex.org/C2780586882","wikidata":"https://www.wikidata.org/wiki/Q7520643","display_name":"Simple (philosophy)","level":2,"score":0.3716999888420105},{"id":"https://openalex.org/C2777303404","wikidata":"https://www.wikidata.org/wiki/Q759757","display_name":"Convergence (economics)","level":2,"score":0.36559998989105225},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.35740000009536743},{"id":"https://openalex.org/C47446073","wikidata":"https://www.wikidata.org/wiki/Q5165890","display_name":"Control theory (sociology)","level":3,"score":0.33320000767707825},{"id":"https://openalex.org/C77618280","wikidata":"https://www.wikidata.org/wiki/Q1155772","display_name":"Scheme (mathematics)","level":2,"score":0.2930999994277954},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.288100004196167},{"id":"https://openalex.org/C143587482","wikidata":"https://www.wikidata.org/wiki/Q1543216","display_name":"Iterative and incremental development","level":2,"score":0.28690001368522644},{"id":"https://openalex.org/C2779982483","wikidata":"https://www.wikidata.org/wiki/Q6094420","display_name":"Iterative refinement","level":2,"score":0.28600001335144043},{"id":"https://openalex.org/C2779679103","wikidata":"https://www.wikidata.org/wiki/Q5251805","display_name":"Degradation (telecommunications)","level":2,"score":0.27239999175071716},{"id":"https://openalex.org/C204323151","wikidata":"https://www.wikidata.org/wiki/Q905424","display_name":"Range (aeronautics)","level":2,"score":0.2644999921321869},{"id":"https://openalex.org/C105795698","wikidata":"https://www.wikidata.org/wiki/Q12483","display_name":"Statistics","level":1,"score":0.259799987077713},{"id":"https://openalex.org/C61797465","wikidata":"https://www.wikidata.org/wiki/Q1188986","display_name":"Term (time)","level":2,"score":0.258899986743927}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1609/aaai.v40i24.39120","is_oa":true,"landing_page_url":"https://doi.org/10.1609/aaai.v40i24.39120","pdf_url":"https://ojs.aaai.org/index.php/AAAI/article/download/39120/43082","source":{"id":"https://openalex.org/S4210191458","display_name":"Proceedings of the AAAI Conference on Artificial Intelligence","issn_l":"2159-5399","issn":["2159-5399","2374-3468"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/P4310320058","host_organization_name":"Association for the Advancement of Artificial Intelligence","host_organization_lineage":["https://openalex.org/P4310320058"],"host_organization_lineage_names":["Association for the Advancement of Artificial Intelligence"],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the AAAI Conference on Artificial Intelligence","raw_type":"journal-article"}],"best_oa_location":{"id":"doi:10.1609/aaai.v40i24.39120","is_oa":true,"landing_page_url":"https://doi.org/10.1609/aaai.v40i24.39120","pdf_url":"https://ojs.aaai.org/index.php/AAAI/article/download/39120/43082","source":{"id":"https://openalex.org/S4210191458","display_name":"Proceedings of the AAAI Conference on Artificial Intelligence","issn_l":"2159-5399","issn":["2159-5399","2374-3468"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/P4310320058","host_organization_name":"Association for the Advancement of Artificial Intelligence","host_organization_lineage":["https://openalex.org/P4310320058"],"host_organization_lineage_names":["Association for the Advancement of Artificial Intelligence"],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the AAAI Conference on Artificial Intelligence","raw_type":"journal-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7138314312.pdf","grobid_xml":"https://content.openalex.org/works/W7138314312.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Layer":[0],"pruning":[1,17,63,123],"is":[2],"a":[3,29,38,58,74],"viable":[4],"technique":[5],"for":[6],"compressing":[7],"large":[8],"language":[9],"models":[10],"while":[11],"achieving":[12],"acceleration":[13],"proportional":[14],"to":[15,43,55,69,144],"the":[16,81,93,100,106,135,150,158],"ratio.":[18],"In":[19],"this":[20],"work,":[21],"we":[22,78],"identify":[23],"that":[24,37,65],"removing":[25],"any":[26],"layer":[27,48,62,85,122],"induces":[28],"magnitude":[30,67,82],"gap":[31,83],"in":[32,46,73,104],"hidden":[33],"states,":[34],"and":[35,87,146],"demonstrate":[36,99],"simple":[39],"compensation":[40,68],"operation":[41],"leads":[42],"superior":[44],"performance":[45,152],"iterative":[47,61,109,115],"pruning.":[49,110],"This":[50],"key":[51],"observation":[52],"motivates":[53],"us":[54],"propose":[56],"Prune&Comp,":[57],"novel,":[59],"plug-and-play":[60],"scheme":[64],"leverages":[66],"mitigate":[70],"such":[71],"gaps":[72],"training-free":[75],"manner.":[76],"Specifically,":[77],"first":[79],"estimate":[80],"of":[84,102,108,130,149],"removal":[86],"then":[88],"eliminate":[89],"it":[90],"by":[91,160],"rescaling":[92],"remaining":[94],"weights":[95],"offline.":[96],"We":[97],"further":[98],"advantages":[101],"Prune&Comp":[103,118,139],"improving":[105],"stability":[107],"When":[111],"integrated":[112],"with":[113,134],"an":[114],"prune-and-compensate":[116],"loop,":[117],"consistently":[119],"enhances":[120],"existing":[121],"metrics.":[124],"For":[125],"instance,":[126],"when":[127],"5":[128],"layers":[129],"LLaMA-3-8B":[131],"are":[132],"pruned":[133],"prevalent":[136],"Taylor+":[137],"metric,":[138],"reduces":[140],"PPL":[141],"from":[142],"512.78":[143],"16.34":[145],"retains":[147],"90.57%":[148],"original":[151],"across":[153],"9":[154],"question-answering":[155],"tasks,":[156],"outperforming":[157],"baseline":[159],"24.72%.":[161]},"counts_by_year":[],"updated_date":"2026-03-20T20:47:17.329874","created_date":"2026-03-18T00:00:00"}