{"id":"https://openalex.org/W7151762824","doi":"https://doi.org/10.48550/arxiv.2604.05546","title":"Efficient Inference for Large Vision-Language Models: Bottlenecks, Techniques, and Prospects","display_name":"Efficient Inference for Large Vision-Language Models: Bottlenecks, Techniques, and Prospects","publication_year":2026,"publication_date":"2026-04-07","ids":{"openalex":"https://openalex.org/W7151762824","doi":"https://doi.org/10.48550/arxiv.2604.05546"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2604.05546","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.05546","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","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.2604.05546","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5083513356","display_name":"Jun Zhang","orcid":"https://orcid.org/0000-0003-2020-9322"},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Zhang, Jun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102011950","display_name":"Yuan Ji","orcid":"https://orcid.org/0000-0002-2631-9082"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ji, Yicheng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133215110","display_name":"Feiyang Ren","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ren, Feiyang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133217899","display_name":"Yihang Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Yihang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133162846","display_name":"Bowen Zeng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zeng, Bowen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5127831773","display_name":"Zonghao Chen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen, Zonghao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100452001","display_name":"Ke Chen","orcid":"https://orcid.org/0000-0002-3855-8363"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen, Ke","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5112840212","display_name":"Lidan Shou","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shou, Lidan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133180477","display_name":"Gang Chen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen, Gang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5016813513","display_name":"Huan Li","orcid":"https://orcid.org/0000-0002-5938-8164"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Huan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":10,"corresponding_author_ids":["https://openalex.org/A5083513356"],"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":0.6050999760627747,"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":0.6050999760627747,"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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.16050000488758087,"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":0.026900000870227814,"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/inference","display_name":"Inference","score":0.6495000123977661},{"id":"https://openalex.org/keywords/visual-analytics","display_name":"Visual analytics","score":0.3977000117301941},{"id":"https://openalex.org/keywords/inference-engine","display_name":"Inference engine","score":0.3709999918937683},{"id":"https://openalex.org/keywords/decoding-methods","display_name":"Decoding methods","score":0.3675999939441681},{"id":"https://openalex.org/keywords/snapshot","display_name":"Snapshot (computer storage)","score":0.34360000491142273},{"id":"https://openalex.org/keywords/backward-chaining","display_name":"Backward chaining","score":0.323199987411499},{"id":"https://openalex.org/keywords/overlay","display_name":"Overlay","score":0.2976999878883362},{"id":"https://openalex.org/keywords/visualization","display_name":"Visualization","score":0.2948000133037567}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7663000226020813},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6495000123977661},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4203000068664551},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4027999937534332},{"id":"https://openalex.org/C59732488","wikidata":"https://www.wikidata.org/wiki/Q2528440","display_name":"Visual analytics","level":3,"score":0.3977000117301941},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.3894999921321869},{"id":"https://openalex.org/C46743427","wikidata":"https://www.wikidata.org/wiki/Q1341685","display_name":"Inference engine","level":3,"score":0.3709999918937683},{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.3675999939441681},{"id":"https://openalex.org/C55282118","wikidata":"https://www.wikidata.org/wiki/Q252683","display_name":"Snapshot (computer storage)","level":2,"score":0.34360000491142273},{"id":"https://openalex.org/C129916263","wikidata":"https://www.wikidata.org/wiki/Q1141183","display_name":"Backward chaining","level":4,"score":0.323199987411499},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.30630001425743103},{"id":"https://openalex.org/C136085584","wikidata":"https://www.wikidata.org/wiki/Q910289","display_name":"Overlay","level":2,"score":0.2976999878883362},{"id":"https://openalex.org/C36464697","wikidata":"https://www.wikidata.org/wiki/Q451553","display_name":"Visualization","level":2,"score":0.2948000133037567},{"id":"https://openalex.org/C2779960059","wikidata":"https://www.wikidata.org/wiki/Q7113681","display_name":"Overhead (engineering)","level":2,"score":0.2939999997615814},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.28769999742507935},{"id":"https://openalex.org/C2777904410","wikidata":"https://www.wikidata.org/wiki/Q7397","display_name":"Software","level":2,"score":0.28760001063346863},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.2815999984741211},{"id":"https://openalex.org/C2776459999","wikidata":"https://www.wikidata.org/wiki/Q2119376","display_name":"Fidelity","level":2,"score":0.28110000491142273},{"id":"https://openalex.org/C152124472","wikidata":"https://www.wikidata.org/wiki/Q1204361","display_name":"Redundancy (engineering)","level":2,"score":0.2782999873161316},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.2743000090122223},{"id":"https://openalex.org/C206729178","wikidata":"https://www.wikidata.org/wiki/Q2271896","display_name":"Scheduling (production processes)","level":2,"score":0.2732999920845032},{"id":"https://openalex.org/C43521106","wikidata":"https://www.wikidata.org/wiki/Q2165493","display_name":"Pipeline (software)","level":2,"score":0.262800008058548},{"id":"https://openalex.org/C174348530","wikidata":"https://www.wikidata.org/wiki/Q188635","display_name":"Bridging (networking)","level":2,"score":0.26260000467300415},{"id":"https://openalex.org/C188045654","wikidata":"https://www.wikidata.org/wiki/Q17148339","display_name":"Memory bandwidth","level":2,"score":0.258899986743927}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2604.05546","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.05546","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"article"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2604.05546","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.05546","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","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":{"Large":[0],"Vision-Language":[1],"Models":[2],"(LVLMs)":[3],"enable":[4],"sophisticated":[5],"reasoning":[6],"over":[7],"images":[8],"and":[9,41,62,94,117,139,174],"videos,":[10],"yet":[11],"their":[12],"inference":[13,56],"is":[14,28,183],"hindered":[15],"by":[16,30,145,151],"a":[17,31,47,124],"systemic":[18],"efficiency":[19,51,105],"barrier":[20],"known":[21],"as":[22],"visual":[23,86,137],"token":[24],"dominance.":[25],"This":[26],"overhead":[27],"driven":[29],"multi-regime":[32],"interplay":[33],"between":[34,136],"high-resolution":[35],"feature":[36],"extraction,":[37],"quadratic":[38],"attention":[39],"scaling,":[40],"memory":[42,97,119],"bandwidth":[43],"constraints.":[44],"We":[45],"present":[46],"systematic":[48],"taxonomy":[49],"of":[50,59,91,110,127],"techniques":[52],"structured":[53,125],"around":[54],"the":[55,73,88,95,104,108,134],"lifecycle,":[57],"consisting":[58],"encoding,":[60,87],"prefilling,":[61],"decoding.":[63,101],"Unlike":[64],"prior":[65],"reviews":[66],"focused":[67],"on":[68,159],"isolated":[69,129],"optimizations,":[70],"we":[71],"analyze":[72],"end-to-end":[74],"pipeline":[75],"to":[76,132],"reveal":[77],"how":[78,128],"upstream":[79],"decisions":[80],"dictate":[81],"downstream":[82],"bottlenecks,":[83],"covering":[84],"compute-bound":[85],"intensive":[89],"prefilling":[90],"massive":[92],"contexts,":[93],"''visual":[96],"wall''":[98],"in":[99],"bandwidth-bound":[100],"By":[102],"decoupling":[103],"landscape":[106],"into":[107],"axes":[109],"shaping":[111],"information":[112],"density,":[113],"managing":[114],"long-context":[115],"attention,":[116],"overcoming":[118],"limits,":[120],"this":[121],"work":[122],"provides":[123],"analysis":[126],"optimizations":[130],"compose":[131],"navigate":[133],"trade-off":[135],"fidelity":[138],"system":[140],"efficiency.":[141],"The":[142],"survey":[143],"concludes":[144],"outlining":[146],"four":[147],"future":[148],"frontiers":[149],"supported":[150],"pilot":[152],"empirical":[153],"insights,":[154],"including":[155],"hybrid":[156],"compression":[157],"based":[158],"functional":[160],"unit":[161],"sensitivity,":[162],"modality-aware":[163],"decoding":[164],"with":[165],"relaxed":[166],"verification,":[167],"progressive":[168],"state":[169],"management":[170],"for":[171],"streaming":[172],"continuity,":[173],"stage-disaggregated":[175],"serving":[176],"through":[177],"hardware-algorithm":[178],"co-design.":[179],"Our":[180],"literature":[181],"repository":[182],"at":[184],"https://github.com/SuDIS-ZJU/Efficient-LVLMs-Inference.":[185]},"counts_by_year":[],"updated_date":"2026-04-29T09:16:38.111599","created_date":"2026-04-09T00:00:00"}