{"id":"https://openalex.org/W7166885016","doi":"https://doi.org/10.48550/arxiv.2606.31903","title":"Attend, Transform, or Silence: Operator-Level Visual Skipping for Efficient Multimodal LLM Inference","display_name":"Attend, Transform, or Silence: Operator-Level Visual Skipping for Efficient Multimodal LLM Inference","publication_year":2026,"publication_date":"2026-06-30","ids":{"openalex":"https://openalex.org/W7166885016","doi":"https://doi.org/10.48550/arxiv.2606.31903"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.31903","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.31903","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"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":false,"raw_source_name":null,"raw_type":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2606.31903","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5121657693","display_name":"Zhaoyang Luo","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Luo, Zhaoyang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139822815","display_name":"Runmin Dong","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Dong, Runmin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139832013","display_name":"Miao Yang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yang, Miao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139780793","display_name":"Fan Wei","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wei, Fan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5113769244","display_name":"Yushan Lai","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lai, Yushan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139821840","display_name":"Bin Luo","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Luo, Bin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5139848997","display_name":"Haohuan Fu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Fu, Haohuan","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":false,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"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":0.9506000280380249,"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":0.9506000280380249,"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":0.01140000019222498,"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":0.008500000461935997,"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/inference","display_name":"Inference","score":0.6949999928474426},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.6403999924659729},{"id":"https://openalex.org/keywords/process","display_name":"Process (computing)","score":0.567300021648407},{"id":"https://openalex.org/keywords/perspective","display_name":"Perspective (graphical)","score":0.5465999841690063},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.47450000047683716},{"id":"https://openalex.org/keywords/visualization","display_name":"Visualization","score":0.3756999969482422},{"id":"https://openalex.org/keywords/sequence","display_name":"Sequence (biology)","score":0.3698999881744385}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8044000267982483},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6949999928474426},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.6403999924659729},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.567300021648407},{"id":"https://openalex.org/C12713177","wikidata":"https://www.wikidata.org/wiki/Q1900281","display_name":"Perspective (graphical)","level":2,"score":0.5465999841690063},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5169000029563904},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.47450000047683716},{"id":"https://openalex.org/C36464697","wikidata":"https://www.wikidata.org/wiki/Q451553","display_name":"Visualization","level":2,"score":0.3756999969482422},{"id":"https://openalex.org/C2778112365","wikidata":"https://www.wikidata.org/wiki/Q3511065","display_name":"Sequence (biology)","level":2,"score":0.3698999881744385},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3499999940395355},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.3474000096321106},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.31369999051094055},{"id":"https://openalex.org/C129916263","wikidata":"https://www.wikidata.org/wiki/Q1141183","display_name":"Backward chaining","level":4,"score":0.31200000643730164},{"id":"https://openalex.org/C44291984","wikidata":"https://www.wikidata.org/wiki/Q1074173","display_name":"Question answering","level":2,"score":0.30559998750686646},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.2847000062465668},{"id":"https://openalex.org/C2780910867","wikidata":"https://www.wikidata.org/wiki/Q1952416","display_name":"Multimodality","level":2,"score":0.2687999904155731}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.31903","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.31903","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"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:10.48550/arxiv.2606.31903","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.31903","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"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":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Multimodal":[0],"large":[1,64],"language":[2],"models":[3],"(MLLMs)":[4],"increasingly":[5],"process":[6],"long":[7],"visual-token":[8,24,50,60,102,109],"sequences,":[9],"increasing":[10],"the":[11,107,145],"overall":[12],"inference":[13],"computation.":[14],"Existing":[15],"acceleration":[16],"methods":[17],"usually":[18],"remove":[19],"visual":[20,91],"tokens":[21],"or":[22,37,117],"skip":[23],"updates":[25,61],"in":[26],"entire":[27],"layers,":[28],"but":[29],"these":[30],"coarse":[31],"strategies":[32],"may":[33],"discard":[34],"fine-grained":[35],"evidence":[36],"suppress":[38],"useful":[39,90],"operators":[40,86],"together":[41],"with":[42],"redundant":[43,114],"ones.":[44],"In":[45],"this":[46,74],"paper,":[47],"we":[48,77],"study":[49],"computation":[51,92],"from":[52],"an":[53,100],"answer-observable":[54],"perspective":[55],"and":[56,84,87,96,124],"find":[57],"that":[58,89,105,129],"late":[59],"can":[62],"remain":[63],"while":[65,111,141],"having":[66],"little":[67],"effect":[68],"on":[69,139],"answer-token":[70],"representations.":[71],"Motivated":[72],"by":[73],"answer-silent":[75],"redundancy,":[76],"decompose":[78],"each":[79],"Transformer":[80],"layer":[81],"into":[82],"attention":[83],"FFN":[85],"show":[88,128],"is":[93],"often":[94],"operator-dominant":[95],"layer-dependent.":[97],"We":[98],"propose":[99],"operator-level":[101],"skipping":[103],"framework":[104],"preserves":[106],"full":[108],"sequence":[110],"selectively":[112],"bypassing":[113],"attention,":[115],"FFN,":[116],"both.":[118],"Experiments":[119],"across":[120],"three":[121],"MLLM":[122],"architectures":[123],"10":[125],"VQA":[126],"benchmarks":[127],"our":[130],"method":[131],"achieves":[132],"strong":[133],"efficiency-accuracy":[134],"trade-offs,":[135],"reducing":[136],"\\textbf{33.7\\%}":[137],"TFLOPs":[138],"Qwen3-VL":[140],"retaining":[142],"\\textbf{99.5\\%}":[143],"of":[144],"vanilla":[146],"model":[147],"performance.":[148]},"counts_by_year":[],"updated_date":"2026-07-02T06:18:51.028212","created_date":"2026-07-02T00:00:00"}