{"id":"https://openalex.org/W4412888444","doi":"https://doi.org/10.18653/v1/2025.findings-acl.359","title":"TAMP: Token-Adaptive Layerwise Pruning in Multimodal Large Language Models","display_name":"TAMP: Token-Adaptive Layerwise Pruning in Multimodal Large Language Models","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4412888444","doi":"https://doi.org/10.18653/v1/2025.findings-acl.359"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.findings-acl.359","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.359","pdf_url":"https://aclanthology.org/2025.findings-acl.359.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":"Findings of the Association for Computational Linguistics: ACL 2025","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.findings-acl.359.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5100415738","display_name":"Jae Wook Lee","orcid":"https://orcid.org/0000-0002-8756-0195"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jaewoo Lee","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5023952633","display_name":"Xuan Ke","orcid":"https://orcid.org/0000-0001-5485-3664"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Keyang Xuan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5086331456","display_name":"Chanakya Ekbote","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chanakya Ekbote","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5010813852","display_name":"Sandeep Polisetty","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sandeep Polisetty","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5029408111","display_name":"Yi R. Fung","orcid":"https://orcid.org/0009-0006-1869-0363"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yi R. Fung","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5110927565","display_name":"Paul Pu Liang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Paul Pu Liang","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":true,"cited_by_count":1,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"6892","last_page":"6908"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.9961000084877014,"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":0.9961000084877014,"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/T12031","display_name":"Speech and dialogue systems","score":0.9940000176429749,"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":0.993399977684021,"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/computer-science","display_name":"Computer science","score":0.7192869782447815},{"id":"https://openalex.org/keywords/pruning","display_name":"Pruning","score":0.7077152729034424},{"id":"https://openalex.org/keywords/security-token","display_name":"Security token","score":0.7007902264595032},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.4467306435108185},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.3888709545135498},{"id":"https://openalex.org/keywords/computer-network","display_name":"Computer network","score":0.13031110167503357}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7192869782447815},{"id":"https://openalex.org/C108010975","wikidata":"https://www.wikidata.org/wiki/Q500094","display_name":"Pruning","level":2,"score":0.7077152729034424},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.7007902264595032},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.4467306435108185},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3888709545135498},{"id":"https://openalex.org/C31258907","wikidata":"https://www.wikidata.org/wiki/Q1301371","display_name":"Computer network","level":1,"score":0.13031110167503357},{"id":"https://openalex.org/C6557445","wikidata":"https://www.wikidata.org/wiki/Q173113","display_name":"Agronomy","level":1,"score":0.0},{"id":"https://openalex.org/C86803240","wikidata":"https://www.wikidata.org/wiki/Q420","display_name":"Biology","level":0,"score":0.0}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.18653/v1/2025.findings-acl.359","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.359","pdf_url":"https://aclanthology.org/2025.findings-acl.359.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":"Findings of the Association for Computational Linguistics: ACL 2025","raw_type":"proceedings-article"},{"id":"pmh:oai:repository.hkust.edu.hk:1783.1-169441","is_oa":false,"landing_page_url":"http://repository.hkust.edu.hk/ir/Record/1783.1-169441","pdf_url":null,"source":{"id":"https://openalex.org/S4306401796","display_name":"Rare & Special e-Zone (The Hong Kong University of Science and Technology)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I200769079","host_organization_name":"Hong Kong University of Science and Technology","host_organization_lineage":["https://openalex.org/I200769079"],"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":"Conference paper"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.findings-acl.359","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.findings-acl.359","pdf_url":"https://aclanthology.org/2025.findings-acl.359.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":"Findings of the Association for Computational Linguistics: ACL 2025","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4412888444.pdf","grobid_xml":"https://content.openalex.org/works/W4412888444.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W4388335561","https://openalex.org/W2970530566","https://openalex.org/W4288261899","https://openalex.org/W4307309205","https://openalex.org/W2967478618","https://openalex.org/W4385009901","https://openalex.org/W4385572700"],"abstract_inverted_index":{"Multimodal":[0,104],"Large":[1],"Language":[2],"Models":[3],"(MLLMs)":[4],"have":[5],"shown":[6],"remarkable":[7],"versatility":[8],"in":[9,29,98],"understanding":[10],"diverse":[11],"multimodal":[12,92,110,146],"data":[13],"and":[14,55,101,133,140],"tasks.However,":[15],"these":[16],"capabilities":[17],"come":[18],"with":[19],"an":[20],"increased":[21],"model":[22,27],"scale.While":[23],"post-training":[24],"pruning":[25,70,159],"reduces":[26],"size":[28],"unimodal":[30],"models,":[31],"its":[32],"application":[33],"to":[34,46,58,116],"MLLMs":[35],"often":[36],"yields":[37],"limited":[38],"success.Our":[39],"analysis":[40],"discovers":[41],"that":[42,150],"conventional":[43],"methods":[44],"fail":[45],"account":[47],"for":[48,73,130],"the":[49],"unique":[50],"token":[51],"attributes":[52],"across":[53,144],"layers":[54],"modalities":[56],"inherent":[57],"MLLMs.Inspired":[59],"by":[60],"this":[61],"observation,":[62],"we":[63],"propose":[64],"TAMP,":[65],"a":[66],"simple":[67],"yet":[68],"effective":[69],"framework":[71],"tailored":[72],"MLLMs,":[74],"featuring":[75],"two":[76,125],"key":[77],"components:":[78],"(1)":[79],"Diversity-Aware":[80],"Sparsity,":[81],"which":[82,107],"adjusts":[83],"sparsity":[84],"ratio":[85],"per":[86],"layer":[87],"based":[88],"on":[89,124],"diversities":[90],"among":[91],"output":[93],"tokens,":[94],"preserving":[95],"more":[96],"parameters":[97],"highdiversity":[99],"layers;":[100],"(2)":[102],"Adaptive":[103],"Input":[105],"Activation,":[106],"identifies":[108],"representative":[109],"input":[111],"tokens":[112],"using":[113],"attention":[114],"scores":[115],"guide":[117],"unstructured":[118],"weight":[119],"pruning.We":[120],"validate":[121],"our":[122,154],"method":[123],"state-of-the-art":[126],"MLLMs:":[127],"LLaVA-NeXT,":[128],"designed":[129],"visionlanguage":[131],"tasks,":[132],"VideoLLaMA2,":[134],"capable":[135],"of":[136,153],"processing":[137],"audio,":[138],"visual,":[139],"language":[141],"modalities.Empirical":[142],"experiments":[143],"various":[145],"evaluation":[147],"benchmarks":[148],"demonstrate":[149],"each":[151],"component":[152],"approach":[155],"substantially":[156],"outperforms":[157],"existing":[158],"techniques.":[160]},"counts_by_year":[{"year":2026,"cited_by_count":1}],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}