{"id":"https://openalex.org/W7140326139","doi":"https://doi.org/10.48550/arxiv.2603.23495","title":"VISion On Request: Enhanced VLLM efficiency with sparse, dynamically selected, vision-language interactions","display_name":"VISion On Request: Enhanced VLLM efficiency with sparse, dynamically selected, vision-language interactions","publication_year":2026,"publication_date":"2026-03-24","ids":{"openalex":"https://openalex.org/W7140326139","doi":"https://doi.org/10.48550/arxiv.2603.23495"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2603.23495","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.23495","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":"article"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2603.23495","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5072822225","display_name":"Adrian Bulat","orcid":"https://orcid.org/0000-0002-3185-4979"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Bulat, Adrian","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5008550420","display_name":"Alberto Baldrati","orcid":"https://orcid.org/0000-0002-5012-5800"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Baldrati, Alberto","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5086692515","display_name":"Ioannis Maniadis Metaxas","orcid":"https://orcid.org/0009-0000-9588-6913"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Metaxas, Ioannis Maniadis","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5081481632","display_name":"Yassine Ouali","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ouali, Yassine","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5130609887","display_name":"Georgios Tzimiropoulos","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tzimiropoulos, Georgios","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":5,"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":0.8507999777793884,"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":0.8507999777793884,"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":0.04019999876618385,"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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.026499999687075615,"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/set","display_name":"Set (abstract data type)","score":0.6349999904632568},{"id":"https://openalex.org/keywords/bottleneck","display_name":"Bottleneck","score":0.6345000267028809},{"id":"https://openalex.org/keywords/context","display_name":"Context (archaeology)","score":0.5591999888420105},{"id":"https://openalex.org/keywords/security-token","display_name":"Security token","score":0.4724000096321106},{"id":"https://openalex.org/keywords/matching","display_name":"Matching (statistics)","score":0.4677000045776367},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.4650000035762787},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.43230000138282776},{"id":"https://openalex.org/keywords/computational-model","display_name":"Computational model","score":0.3758000135421753}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8009999990463257},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.6349999904632568},{"id":"https://openalex.org/C2780513914","wikidata":"https://www.wikidata.org/wiki/Q18210350","display_name":"Bottleneck","level":2,"score":0.6345000267028809},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.5591999888420105},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.508400022983551},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.4724000096321106},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.4677000045776367},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.4650000035762787},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.43230000138282776},{"id":"https://openalex.org/C66024118","wikidata":"https://www.wikidata.org/wiki/Q1122506","display_name":"Computational model","level":2,"score":0.3758000135421753},{"id":"https://openalex.org/C79581498","wikidata":"https://www.wikidata.org/wiki/Q1367530","display_name":"Suite","level":2,"score":0.36469998955726624},{"id":"https://openalex.org/C36464697","wikidata":"https://www.wikidata.org/wiki/Q451553","display_name":"Visualization","level":2,"score":0.362199991941452},{"id":"https://openalex.org/C204323151","wikidata":"https://www.wikidata.org/wiki/Q905424","display_name":"Range (aeronautics)","level":2,"score":0.34610000252723694},{"id":"https://openalex.org/C107457646","wikidata":"https://www.wikidata.org/wiki/Q207434","display_name":"Human\u2013computer interaction","level":1,"score":0.32519999146461487},{"id":"https://openalex.org/C2777508537","wikidata":"https://www.wikidata.org/wiki/Q7936620","display_name":"Visual reasoning","level":2,"score":0.32010000944137573},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.31610000133514404},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.31139999628067017},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.3084000051021576},{"id":"https://openalex.org/C158495155","wikidata":"https://www.wikidata.org/wiki/Q2369151","display_name":"Visual search","level":2,"score":0.27070000767707825},{"id":"https://openalex.org/C44291984","wikidata":"https://www.wikidata.org/wiki/Q1074173","display_name":"Question answering","level":2,"score":0.26460000872612},{"id":"https://openalex.org/C77618280","wikidata":"https://www.wikidata.org/wiki/Q1155772","display_name":"Scheme (mathematics)","level":2,"score":0.262800008058548},{"id":"https://openalex.org/C178253425","wikidata":"https://www.wikidata.org/wiki/Q162668","display_name":"Visual perception","level":3,"score":0.2515999972820282}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2603.23495","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.23495","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":"article"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2603.23495","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.23495","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":"article"},"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":{"Existing":[0],"approaches":[1],"for":[2],"improving":[3],"the":[4,15,67,74,82,87,123,152],"efficiency":[5,71],"of":[6,17,65,90,100,147,154,192],"Large":[7],"Vision-Language":[8],"Models":[9],"(LVLMs)":[10],"are":[11],"largely":[12],"based":[13,169],"on":[14,32,134,144,170],"concept":[16],"visual":[18,62,92,104,124,167,202],"token":[19],"reduction.":[20],"This":[21],"approach,":[22],"however,":[23],"creates":[24],"an":[25],"information":[26],"bottleneck":[27],"that":[28,35,56,164,176,199],"impairs":[29],"performance,":[30],"especially":[31],"challenging":[33,197],"tasks":[34,198],"require":[36,200],"fine-grained":[37],"understanding":[38],"and":[39,78,117,157,194],"reasoning.":[40],"In":[41],"this":[42,46,135],"work,":[43],"we":[44,137],"challenge":[45],"paradigm":[47],"by":[48,72,108,150],"introducing":[49],"VISion":[50],"On":[51],"Request":[52],"(VISOR),":[53],"a":[54,95,114,140,145,160,189],"method":[55],"reduces":[57,179],"inference":[58],"cost":[59,181],"without":[60],"discarding":[61],"information.":[63],"Instead":[64],"compressing":[66],"image,":[68],"VISOR":[69,177],"improves":[70],"sparsifying":[73],"interaction":[75],"between":[76,111],"image":[77],"text":[79],"tokens.":[80],"Specifically,":[81],"language":[83],"model":[84],"attends":[85],"to":[86],"full":[88],"set":[89,99],"high-resolution":[91,129],"tokens":[93],"through":[94],"small,":[96],"strategically":[97],"placed":[98],"attention":[101],"layers:":[102],"general":[103],"context":[105],"is":[106],"provided":[107],"efficient":[109],"cross-attention":[110],"text-image,":[112],"while":[113,182],"few":[115],"well-placed":[116],"dynamically":[118,165],"selected":[119],"self-attention":[120,155],"layers":[121],"refine":[122],"representations":[125],"themselves,":[126],"enabling":[127],"complex,":[128],"reasoning":[130],"when":[131],"needed.":[132],"Based":[133],"principle,":[136],"first":[138],"train":[139],"single":[141],"universal":[142],"network":[143],"range":[146],"computational":[148,180],"budgets":[149],"varying":[151],"number":[153],"layers,":[156],"then":[158],"introduce":[159],"lightweight":[161],"policy":[162],"mechanism":[163],"allocates":[166],"computation":[168],"per-sample":[171],"complexity.":[172],"Extensive":[173],"experiments":[174],"show":[175],"drastically":[178],"matching":[183],"or":[184],"exceeding":[185],"state-of-the-art":[186],"results":[187],"across":[188],"diverse":[190],"suite":[191],"benchmarks,":[193],"excels":[195],"in":[196],"detailed":[201],"understanding.":[203]},"counts_by_year":[],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2026-03-26T00:00:00"}