{"id":"https://openalex.org/W4414820693","doi":"https://doi.org/10.18653/v1/2026.findings-acl.1633","title":"TrimTokenator: Towards Adaptive Visual Token Pruning for Large Multimodal Models","display_name":"TrimTokenator: Towards Adaptive Visual Token Pruning for Large Multimodal Models","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W4414820693","doi":"https://doi.org/10.18653/v1/2026.findings-acl.1633"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2026.findings-acl.1633","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1633","pdf_url":"https://aclanthology.org/2026.findings-acl.1633.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 2026","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2026.findings-acl.1633.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5100396895","display_name":"Hao Zhang","orcid":"https://orcid.org/0000-0002-3525-2318"},"institutions":[{"id":"https://openalex.org/I4210100255","display_name":"Beijing Academy of Artificial Intelligence","ror":"https://ror.org/016a74861","country_code":"CN","type":"other","lineage":["https://openalex.org/I4210100255"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Hao Zhang","raw_affiliation_strings":["Beijing Academy of Artificial Intelligence (BAAI"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Beijing Academy of Artificial Intelligence (BAAI","institution_ids":["https://openalex.org/I4210100255"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5068656312","display_name":"Mengsi Lyu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lyu Mengsi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5054001546","display_name":"Chaoqun He","orcid":"https://orcid.org/0000-0002-2615-6244"},"institutions":[{"id":"https://openalex.org/I4210100255","display_name":"Beijing Academy of Artificial Intelligence","ror":"https://ror.org/016a74861","country_code":"CN","type":"other","lineage":["https://openalex.org/I4210100255"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Chenrui He","raw_affiliation_strings":["Beijing Academy of Artificial Intelligence (BAAI"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Beijing Academy of Artificial Intelligence (BAAI","institution_ids":["https://openalex.org/I4210100255"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5001005409","display_name":"Yulong Ao","orcid":null},"institutions":[{"id":"https://openalex.org/I4210100255","display_name":"Beijing Academy of Artificial Intelligence","ror":"https://ror.org/016a74861","country_code":"CN","type":"other","lineage":["https://openalex.org/I4210100255"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yulong Ao","raw_affiliation_strings":["Beijing Academy of Artificial Intelligence (BAAI"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Beijing Academy of Artificial Intelligence (BAAI","institution_ids":["https://openalex.org/I4210100255"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5103510896","display_name":"Yonghua Lin","orcid":null},"institutions":[{"id":"https://openalex.org/I4210100255","display_name":"Beijing Academy of Artificial Intelligence","ror":"https://ror.org/016a74861","country_code":"CN","type":"other","lineage":["https://openalex.org/I4210100255"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yonghua Lin","raw_affiliation_strings":["Beijing Academy of Artificial Intelligence (BAAI"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Beijing Academy of Artificial Intelligence (BAAI","institution_ids":["https://openalex.org/I4210100255"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.00719594,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"32633","last_page":"32650"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11714","display_name":"Multimodal Machine Learning Applications","score":0.9965999722480774,"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":0.9965999722480774,"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/T11439","display_name":"Video Analysis and Summarization","score":0.9857000112533569,"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/T10812","display_name":"Human Pose and Action Recognition","score":0.9628000259399414,"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/security-token","display_name":"Security token","score":0.8410000205039978},{"id":"https://openalex.org/keywords/pruning","display_name":"Pruning","score":0.6337000131607056},{"id":"https://openalex.org/keywords/redundancy","display_name":"Redundancy (engineering)","score":0.5656999945640564},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.555400013923645},{"id":"https://openalex.org/keywords/pairwise-comparison","display_name":"Pairwise comparison","score":0.5365999937057495},{"id":"https://openalex.org/keywords/encode","display_name":"ENCODE","score":0.5185999870300293},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.46129998564720154}],"concepts":[{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.8410000205039978},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8154000043869019},{"id":"https://openalex.org/C108010975","wikidata":"https://www.wikidata.org/wiki/Q500094","display_name":"Pruning","level":2,"score":0.6337000131607056},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.569100022315979},{"id":"https://openalex.org/C152124472","wikidata":"https://www.wikidata.org/wiki/Q1204361","display_name":"Redundancy (engineering)","level":2,"score":0.5656999945640564},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.555400013923645},{"id":"https://openalex.org/C184898388","wikidata":"https://www.wikidata.org/wiki/Q1435712","display_name":"Pairwise comparison","level":2,"score":0.5365999937057495},{"id":"https://openalex.org/C66746571","wikidata":"https://www.wikidata.org/wiki/Q1134833","display_name":"ENCODE","level":3,"score":0.5185999870300293},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.46129998564720154},{"id":"https://openalex.org/C41608201","wikidata":"https://www.wikidata.org/wiki/Q980509","display_name":"Embedding","level":2,"score":0.3677999973297119},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.36070001125335693},{"id":"https://openalex.org/C147764199","wikidata":"https://www.wikidata.org/wiki/Q6865248","display_name":"Minification","level":2,"score":0.35600000619888306},{"id":"https://openalex.org/C111335779","wikidata":"https://www.wikidata.org/wiki/Q3454686","display_name":"Reduction (mathematics)","level":2,"score":0.3303999900817871},{"id":"https://openalex.org/C36464697","wikidata":"https://www.wikidata.org/wiki/Q451553","display_name":"Visualization","level":2,"score":0.32170000672340393},{"id":"https://openalex.org/C2777508537","wikidata":"https://www.wikidata.org/wiki/Q7936620","display_name":"Visual reasoning","level":2,"score":0.30140000581741333},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.29840001463890076},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.2770000100135803},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.2653999924659729},{"id":"https://openalex.org/C51823790","wikidata":"https://www.wikidata.org/wiki/Q504353","display_name":"Greedy algorithm","level":2,"score":0.2531999945640564}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.18653/v1/2026.findings-acl.1633","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1633","pdf_url":"https://aclanthology.org/2026.findings-acl.1633.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 2026","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2509.00320","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2509.00320","pdf_url":"https://arxiv.org/pdf/2509.00320","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,"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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2509.00320","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2509.00320","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,"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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.findings-acl.1633","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1633","pdf_url":"https://aclanthology.org/2026.findings-acl.1633.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 2026","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4414820693.pdf","grobid_xml":"https://content.openalex.org/works/W4414820693.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Large":[0],"Multimodal":[1],"Models":[2],"(LMMs)":[3],"have":[4],"achieved":[5],"significant":[6],"success":[7],"across":[8],"various":[9],"tasks.These":[10],"models":[11,181],"usually":[12],"encode":[13],"visual":[14,82,91,98,121,134],"inputs":[15],"into":[16],"dense":[17],"token":[18,36,57,99,116],"sequences,":[19],"which":[20],"are":[21],"then":[22],"concatenated":[23],"with":[24,125,163],"textual":[25,84,126,136],"tokens":[26,85,122,142,177],"and":[27,41,83,86,108,135,157,185],"jointly":[28],"processed":[29],"by":[30,143,178],"a":[31,50,97,113,164,189],"language":[32],"model.However,":[33],"the":[34,78,130,133,140,150],"increased":[35],"count":[37],"substantially":[38],"raises":[39],"computational":[40],"memory":[42],"costs":[43],"during":[44],"inference.Token":[45],"pruning":[46,58,88,100,117],"has":[47],"emerged":[48],"as":[49,183],"promising":[51],"approach":[52],"to":[53,70,153],"address":[54],"this":[55,74],"issue.Existing":[56],"methods":[59],"often":[60],"rely":[61],"on":[62,90,93,180],"costly":[63],"calibration":[64],"or":[65],"suboptimal":[66],"importance":[67],"metrics,":[68],"leading":[69],"redundant":[71],"retained":[72,141],"tokens.In":[73],"paper,":[75],"we":[76,95],"analyze":[77],"redundancy":[79],"differences":[80],"between":[81,132],"propose":[87,96],"exclusively":[89],"tokens.Based":[92],"this,":[94],"strategy":[101,118],"that":[102,119,169],"explicitly":[103],"preserves":[104],"both":[105],"cross-modal":[106],"alignment":[107,131],"intra-modal":[109],"informational":[110],"diversity.We":[111],"introduce":[112],"mutual":[114],"information-based":[115],"removes":[120],"semantically":[123],"misaligned":[124],"tokens,":[127],"effectively":[128],"preserving":[129],"modalities.We":[137],"further":[138],"refine":[139],"maximizing":[144],"their":[145],"expected":[146],"pairwise":[147],"distances":[148],"in":[149,188,192],"latent":[151],"space":[152],"enhance":[154],"representational":[155],"quality":[156],"reduce":[158],"redundancy.which":[159],"is":[160],"solved":[161],"efficiently":[162],"greedy":[165],"algorithm.Extensive":[166],"experiments":[167],"demonstrate":[168],"our":[170],"method":[171],"maintains":[172],"strong":[173],"performance":[174],"while":[175],"reducing":[176],"88.9%":[179],"such":[182],"LLaVA-1.5-7B":[184],"LLaVA-NEXT-7B,":[186],"resulting":[187],"56.7%":[190],"improvement":[191],"inference":[193],"speed.":[194]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2025-10-10T00:00:00"}