{"id":"https://openalex.org/W7164022717","doi":"https://doi.org/10.1007/s11263-026-02865-5","title":"Structural Pruning of Large Vision Language Models: A Comprehensive Study on Pruning Dynamics, Recovery, and Data Efficiency","display_name":"Structural Pruning of Large Vision Language Models: A Comprehensive Study on Pruning Dynamics, Recovery, and Data Efficiency","publication_year":2026,"publication_date":"2026-06-01","ids":{"openalex":"https://openalex.org/W7164022717","doi":"https://doi.org/10.1007/s11263-026-02865-5"},"language":"en","primary_location":{"id":"doi:10.1007/s11263-026-02865-5","is_oa":true,"landing_page_url":"https://doi.org/10.1007/s11263-026-02865-5","pdf_url":"https://link.springer.com/content/pdf/10.1007/s11263-026-02865-5.pdf","source":{"id":"https://openalex.org/S25538012","display_name":"International Journal of Computer Vision","issn_l":"0920-5691","issn":["0920-5691","1573-1405"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319900","host_organization_name":"Springer Science+Business Media","host_organization_lineage":["https://openalex.org/P4310319900","https://openalex.org/P4310319965"],"host_organization_lineage_names":["Springer Science+Business Media","Springer Nature"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"International Journal of Computer Vision","raw_type":"journal-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"hybrid","oa_url":"https://link.springer.com/content/pdf/10.1007/s11263-026-02865-5.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5138254669","display_name":"Yiran Huang","orcid":"https://orcid.org/0009-0004-6937-5668"},"institutions":[{"id":"https://openalex.org/I3018134672","display_name":"Helmholtz Zentrum M\u00fcnchen","ror":"https://ror.org/00cfam450","country_code":"DE","type":"facility","lineage":["https://openalex.org/I1305996414","https://openalex.org/I3018134672"]},{"id":"https://openalex.org/I4403386549","display_name":"Munich Center for Machine Learning","ror":"https://ror.org/02nfy3535","country_code":null,"type":"education","lineage":["https://openalex.org/I4403386549","https://openalex.org/I62916508","https://openalex.org/I8204097"]},{"id":"https://openalex.org/I62916508","display_name":"Technical University of Munich","ror":"https://ror.org/02kkvpp62","country_code":"DE","type":"education","lineage":["https://openalex.org/I62916508"]}],"countries":["DE"],"is_corresponding":true,"raw_author_name":"Yiran Huang","raw_affiliation_strings":["Helmholtz Munich, Munich, Germany","Munich Center for Machine Learning, Munich, Germany","Technical University of Munich, Munich, Germany"],"raw_orcid":"https://orcid.org/0009-0004-6937-5668","affiliations":[{"raw_affiliation_string":"Helmholtz Munich, Munich, Germany","institution_ids":["https://openalex.org/I3018134672"]},{"raw_affiliation_string":"Munich Center for Machine Learning, Munich, Germany","institution_ids":["https://openalex.org/I4403386549"]},{"raw_affiliation_string":"Technical University of Munich, Munich, Germany","institution_ids":["https://openalex.org/I62916508"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5039022463","display_name":"Lukas Thede","orcid":null},"institutions":[{"id":"https://openalex.org/I3018134672","display_name":"Helmholtz Zentrum M\u00fcnchen","ror":"https://ror.org/00cfam450","country_code":"DE","type":"facility","lineage":["https://openalex.org/I1305996414","https://openalex.org/I3018134672"]},{"id":"https://openalex.org/I8087733","display_name":"University of T\u00fcbingen","ror":"https://ror.org/03a1kwz48","country_code":"DE","type":"education","lineage":["https://openalex.org/I8087733"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Lukas Thede","raw_affiliation_strings":["Helmholtz Munich, Munich, Germany","University of T\u00fcbingen, T\u00fcbingen AI Center, T\u00fcbingen, Germany"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Helmholtz Munich, Munich, Germany","institution_ids":["https://openalex.org/I3018134672"]},{"raw_affiliation_string":"University of T\u00fcbingen, T\u00fcbingen AI Center, T\u00fcbingen, Germany","institution_ids":["https://openalex.org/I8087733"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138265321","display_name":"Massimiliano Mancini","orcid":null},"institutions":[{"id":"https://openalex.org/I193223587","display_name":"University of Trento","ror":"https://ror.org/05trd4x28","country_code":"IT","type":"education","lineage":["https://openalex.org/I193223587"]}],"countries":["IT"],"is_corresponding":false,"raw_author_name":"Massimiliano Mancini","raw_affiliation_strings":["University of Trento, Trento, Italy"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Trento, Trento, Italy","institution_ids":["https://openalex.org/I193223587"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138225612","display_name":"Wenjia Xu","orcid":null},"institutions":[{"id":"https://openalex.org/I139759216","display_name":"Beijing University of Posts and Telecommunications","ror":"https://ror.org/04w9fbh59","country_code":"CN","type":"education","lineage":["https://openalex.org/I139759216"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Wenjia Xu","raw_affiliation_strings":["Beijing University of Posts and Telecommunications, Beijing, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Beijing University of Posts and Telecommunications, Beijing, China","institution_ids":["https://openalex.org/I139759216"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5138266577","display_name":"Zeynep Akata","orcid":null},"institutions":[{"id":"https://openalex.org/I3018134672","display_name":"Helmholtz Zentrum M\u00fcnchen","ror":"https://ror.org/00cfam450","country_code":"DE","type":"facility","lineage":["https://openalex.org/I1305996414","https://openalex.org/I3018134672"]},{"id":"https://openalex.org/I4403386549","display_name":"Munich Center for Machine Learning","ror":"https://ror.org/02nfy3535","country_code":null,"type":"education","lineage":["https://openalex.org/I4403386549","https://openalex.org/I62916508","https://openalex.org/I8204097"]},{"id":"https://openalex.org/I62916508","display_name":"Technical University of Munich","ror":"https://ror.org/02kkvpp62","country_code":"DE","type":"education","lineage":["https://openalex.org/I62916508"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Zeynep Akata","raw_affiliation_strings":["Helmholtz Munich, Munich, Germany","Munich Center for Machine Learning, Munich, Germany","Technical University of Munich, Munich, Germany"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Helmholtz Munich, Munich, Germany","institution_ids":["https://openalex.org/I3018134672"]},{"raw_affiliation_string":"Munich Center for Machine Learning, Munich, Germany","institution_ids":["https://openalex.org/I4403386549"]},{"raw_affiliation_string":"Technical University of Munich, Munich, Germany","institution_ids":["https://openalex.org/I62916508"]}]}],"institutions":[],"countries_distinct_count":3,"institutions_distinct_count":6,"corresponding_author_ids":["https://openalex.org/A5138254669"],"corresponding_institution_ids":["https://openalex.org/I3018134672","https://openalex.org/I4403386549","https://openalex.org/I62916508"],"apc_list":{"value":3690,"currency":"USD","value_usd":3690},"apc_paid":{"value":3690,"currency":"USD","value_usd":3690},"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.68540358,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":"134","issue":"6","first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11714","display_name":"Multimodal Machine Learning Applications","score":0.659600019454956,"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"}},"topics":[{"id":"https://openalex.org/T11714","display_name":"Multimodal Machine Learning Applications","score":0.659600019454956,"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"}},{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.1177000030875206,"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"}},{"id":"https://openalex.org/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.042899999767541885,"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":0.8294000029563904},{"id":"https://openalex.org/keywords/flexibility","display_name":"Flexibility (engineering)","score":0.527400016784668},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.4853000044822693},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.4115999937057495},{"id":"https://openalex.org/keywords/distillation","display_name":"Distillation","score":0.3982999920845032},{"id":"https://openalex.org/keywords/exploit","display_name":"Exploit","score":0.39660000801086426},{"id":"https://openalex.org/keywords/reduction","display_name":"Reduction (mathematics)","score":0.37459999322891235}],"concepts":[{"id":"https://openalex.org/C108010975","wikidata":"https://www.wikidata.org/wiki/Q500094","display_name":"Pruning","level":2,"score":0.8294000029563904},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8277000188827515},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.588100016117096},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5856000185012817},{"id":"https://openalex.org/C2780598303","wikidata":"https://www.wikidata.org/wiki/Q65921492","display_name":"Flexibility (engineering)","level":2,"score":0.527400016784668},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.4853000044822693},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.4115999937057495},{"id":"https://openalex.org/C204030448","wikidata":"https://www.wikidata.org/wiki/Q101017","display_name":"Distillation","level":2,"score":0.3982999920845032},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.39660000801086426},{"id":"https://openalex.org/C111335779","wikidata":"https://www.wikidata.org/wiki/Q3454686","display_name":"Reduction (mathematics)","level":2,"score":0.37459999322891235},{"id":"https://openalex.org/C162307627","wikidata":"https://www.wikidata.org/wiki/Q204833","display_name":"Enhanced Data Rates for GSM Evolution","level":2,"score":0.358599990606308},{"id":"https://openalex.org/C120936955","wikidata":"https://www.wikidata.org/wiki/Q2155640","display_name":"Empirical research","level":2,"score":0.34860000014305115},{"id":"https://openalex.org/C115051666","wikidata":"https://www.wikidata.org/wiki/Q6522493","display_name":"Ranging","level":2,"score":0.3450999855995178},{"id":"https://openalex.org/C22367795","wikidata":"https://www.wikidata.org/wiki/Q7625208","display_name":"Structured prediction","level":2,"score":0.30169999599456787},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.30169999599456787},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.2930999994277954},{"id":"https://openalex.org/C136197465","wikidata":"https://www.wikidata.org/wiki/Q1729295","display_name":"Variety (cybernetics)","level":2,"score":0.26489999890327454}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.1007/s11263-026-02865-5","is_oa":true,"landing_page_url":"https://doi.org/10.1007/s11263-026-02865-5","pdf_url":"https://link.springer.com/content/pdf/10.1007/s11263-026-02865-5.pdf","source":{"id":"https://openalex.org/S25538012","display_name":"International Journal of Computer Vision","issn_l":"0920-5691","issn":["0920-5691","1573-1405"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319900","host_organization_name":"Springer Science+Business Media","host_organization_lineage":["https://openalex.org/P4310319900","https://openalex.org/P4310319965"],"host_organization_lineage_names":["Springer Science+Business Media","Springer Nature"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"International Journal of Computer Vision","raw_type":"journal-article"},{"id":"pmh:oai:iris.unitn.it:11572/491370","is_oa":true,"landing_page_url":"https://link.springer.com/article/10.1007/s11263-026-02865-5","pdf_url":null,"source":{"id":"https://openalex.org/S4306401913","display_name":"Institutional Research Information System (Universit\u00e0 degli Studi di Trento)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I193223587","host_organization_name":"University of Trento","host_organization_lineage":["https://openalex.org/I193223587"],"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":"info:eu-repo/semantics/article"}],"best_oa_location":{"id":"doi:10.1007/s11263-026-02865-5","is_oa":true,"landing_page_url":"https://doi.org/10.1007/s11263-026-02865-5","pdf_url":"https://link.springer.com/content/pdf/10.1007/s11263-026-02865-5.pdf","source":{"id":"https://openalex.org/S25538012","display_name":"International Journal of Computer Vision","issn_l":"0920-5691","issn":["0920-5691","1573-1405"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319900","host_organization_name":"Springer Science+Business Media","host_organization_lineage":["https://openalex.org/P4310319900","https://openalex.org/P4310319965"],"host_organization_lineage_names":["Springer Science+Business Media","Springer Nature"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"International Journal of Computer Vision","raw_type":"journal-article"},"sustainable_development_goals":[],"awards":[],"funders":[{"id":"https://openalex.org/F4320323383","display_name":"Technische Universit\u00e4t M\u00fcnchen","ror":"https://ror.org/02kkvpp62"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7164022717.pdf","grobid_xml":"https://content.openalex.org/works/W7164022717.grobid-xml"},"referenced_works_count":25,"referenced_works":["https://openalex.org/W2307512708","https://openalex.org/W2946794439","https://openalex.org/W2963518342","https://openalex.org/W2970454332","https://openalex.org/W3034368386","https://openalex.org/W3105966348","https://openalex.org/W3120043490","https://openalex.org/W3136363192","https://openalex.org/W3174544005","https://openalex.org/W4385571011","https://openalex.org/W4386083031","https://openalex.org/W4389523832","https://openalex.org/W4400121375","https://openalex.org/W4402713111","https://openalex.org/W4402716027","https://openalex.org/W4402716477","https://openalex.org/W4402727764","https://openalex.org/W4402772638","https://openalex.org/W4403793633","https://openalex.org/W4404356490","https://openalex.org/W4412887764","https://openalex.org/W4416034326","https://openalex.org/W7117972625","https://openalex.org/W7160088235","https://openalex.org/W7160114284"],"related_works":[],"abstract_inverted_index":{"Abstract":[0],"While":[1],"Large":[2],"Vision":[3],"Language":[4],"Models":[5],"(LVLMs)":[6],"demonstrate":[7],"impressive":[8],"capabilities,":[9],"their":[10],"substantial":[11],"computational":[12,124],"and":[13,41,75,78,84,89,156],"memory":[14],"requirements":[15],"pose":[16],"deployment":[17],"challenges":[18],"on":[19,87,190],"resource-constrained":[20],"edge":[21],"devices.":[22],"Current":[23],"parameter":[24],"reduction":[25],"techniques":[26],"primarily":[27],"involve":[28],"training":[29,100],"LVLMs":[30,52,210],"from":[31,196],"small":[32,104,147],"language":[33,59],"models,":[34],"but":[35],"these":[36],"methods":[37],"offer":[38],"limited":[39,127],"flexibility":[40],"remain":[42],"computationally":[43],"intensive.":[44],"We":[45],"study":[46,189,202],"a":[47,103,151],"complementary":[48],"route:":[49],"compressing":[50],"existing":[51],"by":[53,63],"applying":[54],"structured":[55],"pruning":[56,72,115,164],"to":[57,198,208],"the":[58,95,107,136,141,176,184],"model":[60],"backbone,":[61],"followed":[62],"lightweight":[64],"recovery":[65,99,137,161,168],"training.":[66],"Specifically,":[67],"we":[68,93],"investigate":[69],"two":[70],"structural":[71],"paradigms:":[73],"layerwise":[74],"widthwise":[76,114],"pruning,":[77],"pair":[79],"them":[80],"with":[81,101],"supervised":[82,154],"finetuning":[83,132,139,155],"knowledge":[85],"distillation":[86,158],"logits":[88],"hidden":[90],"states.":[91],"Additionally,":[92],"assess":[94],"feasibility":[96],"of":[97,106,153,175,183],"conducting":[98],"only":[102,140],"fraction":[105],"available":[108],"data.":[109,133,217],"Our":[110],"results":[111],"show":[112],"that":[113],"generally":[116],"maintains":[117],"better":[118],"performance":[119],"in":[120],"low-resource":[121],"scenarios,":[122],"where":[123],"resources":[125,214],"are":[126],"or":[128,215],"there":[129],"is":[130,144],"insufficient":[131],"As":[134],"for":[135,206],"training,":[138],"multimodal":[142],"projector":[143],"sufficient":[145,216],"at":[146],"compression":[148],"levels.":[149,165],"Furthermore,":[150],"combination":[152],"hidden-state":[157],"yields":[159],"optimal":[160],"across":[162],"various":[163],"Notably,":[166],"effective":[167],"can":[169],"be":[170],"achieved":[171],"using":[172],"just":[173],"5%":[174],"original":[177,185],"data,":[178],"while":[179],"retaining":[180],"over":[181],"95%":[182],"performance.":[186],"Through":[187],"empirical":[188],"three":[191],"representative":[192],"LVLM":[193],"families":[194],"ranging":[195],"3B":[197],"7B":[199],"parameters,":[200],"this":[201],"offers":[203],"actionable":[204],"insights":[205],"practitioners":[207],"compress":[209],"without":[211],"extensive":[212],"computation":[213]},"counts_by_year":[],"updated_date":"2026-07-29T09:40:50.615796","created_date":"2026-06-10T00:00:00"}