{"id":"https://openalex.org/W7154426258","doi":"https://doi.org/10.48550/arxiv.2604.11095","title":"Bottleneck Tokens for Unified Multimodal Retrieval","display_name":"Bottleneck Tokens for Unified Multimodal Retrieval","publication_year":2026,"publication_date":"2026-04-13","ids":{"openalex":"https://openalex.org/W7154426258","doi":"https://doi.org/10.48550/arxiv.2604.11095"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2604.11095","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.11095","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.2604.11095","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5102014331","display_name":"Siyu Sun","orcid":"https://orcid.org/0009-0003-0861-8788"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sun, Siyu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133591971","display_name":"Jing Ren","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ren, Jing","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133583220","display_name":"Zhaohe Liao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liao, Zhaohe","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133603616","display_name":"Dongxiao Mao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Mao, Dongxiao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133603529","display_name":"Xiangyuan Ren","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ren, Xiangyuan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133620306","display_name":"Yiyi Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Yiyi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101925038","display_name":"Haohua Zhao","orcid":"https://orcid.org/0000-0002-2414-4362"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhao, Haohua","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Lin, Weixiong","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lin, Weixiong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5058894276","display_name":"Jiang Shaohua","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shaohua, Jiang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133561919","display_name":"Liqing Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Liqing","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5133577078","display_name":"Yuchao Zheng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zheng, Yuchao","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":0.6158000230789185,"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":0.6158000230789185,"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/T10028","display_name":"Topic Modeling","score":0.1687999963760376,"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":0.03400000184774399,"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":0.5503000020980835},{"id":"https://openalex.org/keywords/set","display_name":"Set (abstract data type)","score":0.5131999850273132},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.48539999127388},{"id":"https://openalex.org/keywords/pooling","display_name":"Pooling","score":0.4772999882698059},{"id":"https://openalex.org/keywords/information-bottleneck-method","display_name":"Information bottleneck method","score":0.4697999954223633},{"id":"https://openalex.org/keywords/generative-grammar","display_name":"Generative grammar","score":0.46619999408721924},{"id":"https://openalex.org/keywords/bottleneck","display_name":"Bottleneck","score":0.4318000078201294},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.41200000047683716},{"id":"https://openalex.org/keywords/embedding","display_name":"Embedding","score":0.3919000029563904}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8389999866485596},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.5503000020980835},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5453000068664551},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.5131999850273132},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.48539999127388},{"id":"https://openalex.org/C70437156","wikidata":"https://www.wikidata.org/wiki/Q7228652","display_name":"Pooling","level":2,"score":0.4772999882698059},{"id":"https://openalex.org/C60008888","wikidata":"https://www.wikidata.org/wiki/Q6031013","display_name":"Information bottleneck method","level":3,"score":0.4697999954223633},{"id":"https://openalex.org/C39890363","wikidata":"https://www.wikidata.org/wiki/Q36108","display_name":"Generative grammar","level":2,"score":0.46619999408721924},{"id":"https://openalex.org/C2780513914","wikidata":"https://www.wikidata.org/wiki/Q18210350","display_name":"Bottleneck","level":2,"score":0.4318000078201294},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.41200000047683716},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.3993000090122223},{"id":"https://openalex.org/C41608201","wikidata":"https://www.wikidata.org/wiki/Q980509","display_name":"Embedding","level":2,"score":0.3919000029563904},{"id":"https://openalex.org/C2777601683","wikidata":"https://www.wikidata.org/wiki/Q6499736","display_name":"Vocabulary","level":2,"score":0.38989999890327454},{"id":"https://openalex.org/C2779960059","wikidata":"https://www.wikidata.org/wiki/Q7113681","display_name":"Overhead (engineering)","level":2,"score":0.3783000111579895},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.36809998750686646},{"id":"https://openalex.org/C167966045","wikidata":"https://www.wikidata.org/wiki/Q5532625","display_name":"Generative model","level":3,"score":0.36809998750686646},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.36000001430511475},{"id":"https://openalex.org/C48103436","wikidata":"https://www.wikidata.org/wiki/Q599031","display_name":"State (computer science)","level":2,"score":0.3393999934196472},{"id":"https://openalex.org/C2777735758","wikidata":"https://www.wikidata.org/wiki/Q817765","display_name":"Path (computing)","level":2,"score":0.32580000162124634},{"id":"https://openalex.org/C19768560","wikidata":"https://www.wikidata.org/wiki/Q320727","display_name":"Dependency (UML)","level":2,"score":0.3255000114440918},{"id":"https://openalex.org/C184337299","wikidata":"https://www.wikidata.org/wiki/Q1437428","display_name":"Semantics (computer science)","level":2,"score":0.3215999901294708},{"id":"https://openalex.org/C44291984","wikidata":"https://www.wikidata.org/wiki/Q1074173","display_name":"Question answering","level":2,"score":0.30880001187324524},{"id":"https://openalex.org/C97931131","wikidata":"https://www.wikidata.org/wiki/Q5282087","display_name":"Discriminative model","level":2,"score":0.29660001397132874},{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.28940001130104065},{"id":"https://openalex.org/C61797465","wikidata":"https://www.wikidata.org/wiki/Q1188986","display_name":"Term (time)","level":2,"score":0.2720000147819519},{"id":"https://openalex.org/C46686674","wikidata":"https://www.wikidata.org/wiki/Q466303","display_name":"Boosting (machine learning)","level":2,"score":0.271699994802475},{"id":"https://openalex.org/C23224414","wikidata":"https://www.wikidata.org/wiki/Q176769","display_name":"Hidden Markov model","level":2,"score":0.27079999446868896},{"id":"https://openalex.org/C195324797","wikidata":"https://www.wikidata.org/wiki/Q33742","display_name":"Natural language","level":2,"score":0.2678999900817871},{"id":"https://openalex.org/C60048249","wikidata":"https://www.wikidata.org/wiki/Q37437","display_name":"Syntax","level":2,"score":0.25859999656677246}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2604.11095","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.11095","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.2604.11095","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.11095","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":[{"display_name":"Quality Education","score":0.5930231809616089,"id":"https://metadata.un.org/sdg/4"}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Adapting":[0],"decoder-only":[1],"multimodal":[2,9],"large":[3],"language":[4],"models":[5],"(MLLMs)":[6],"for":[7,42,140],"unified":[8],"retrieval":[10],"faces":[11],"two":[12,72],"structural":[13],"gaps.":[14],"First,":[15],"existing":[16],"methods":[17,179],"rely":[18],"on":[19,59,196,202],"implicit":[20],"pooling,":[21],"which":[22],"overloads":[23],"the":[24,35,50,113,130,133,147],"hidden":[25],"state":[26],"of":[27,84,188],"a":[28,38,81,90,102,108,154],"standard":[29],"vocabulary":[30],"token":[31],"(e.g.,":[32,200],")":[33],"as":[34,89],"sequence-level":[36],"representation,":[37],"mechanism":[39],"never":[40],"designed":[41],"information":[43,61],"aggregation.":[44],"Second,":[45],"contrastive":[46],"fine-tuning":[47],"specifies":[48],"what":[49],"embedding":[51],"should":[52,62],"match":[53],"but":[54],"provides":[55],"no":[56],"token-level":[57,138],"guidance":[58],"how":[60],"be":[63],"compressed":[64],"into":[65,136],"it.":[66],"We":[67],"address":[68],"both":[69],"gaps":[70],"with":[71,107,158,193],"complementary":[73],"components.":[74],"Architecturally,":[75],"we":[76,97],"introduce":[77],"Bottleneck":[78],"Tokens":[79],"(BToks),":[80],"small":[82],"set":[83],"learnable":[85],"tokens":[86,119],"that":[87,111],"serve":[88],"fixed-capacity":[91],"explicit":[92],"pooling":[93],"mechanism.":[94],"For":[95],"training,":[96],"propose":[98],"Generative":[99],"Information":[100],"Condensation:":[101],"next-token":[103],"prediction":[104],"objective":[105],"coupled":[106],"Condensation":[109],"Mask":[110],"severs":[112],"direct":[114],"attention":[115],"path":[116],"from":[117],"target":[118],"to":[120],"query":[121],"tokens.":[122],"All":[123],"predictive":[124],"signals":[125],"are":[126,151],"thereby":[127],"forced":[128],"through":[129],"BToks,":[131],"converting":[132],"generative":[134],"loss":[135],"dense,":[137],"supervision":[139],"semantic":[141],"compression.":[142],"At":[143],"inference":[144],"time,":[145],"only":[146],"input":[148],"and":[149],"BToks":[150],"processed":[152],"in":[153],"single":[155],"forward":[156],"pass":[157],"negligible":[159],"overhead":[160],"over":[161,191],"conventional":[162],"last-token":[163],"pooling.":[164],"On":[165],"MMEB-V2":[166],"(78":[167],"datasets,":[168],"3":[169],"modalities,":[170],"9":[171],"meta-tasks),":[172],"our":[173],"approach":[174],"achieves":[175],"state-of-the-art":[176],"among":[177],"2B-scale":[178],"under":[180],"comparable":[181],"data":[182],"conditions,":[183],"attaining":[184],"an":[185],"Overall":[186],"score":[187],"59.0":[189],"(+3.6":[190],"VLM2Vec-V2)":[192],"substantial":[194],"gains":[195],"semantically":[197],"demanding":[198],"tasks":[199],"+12.6":[201],"Video-QA).":[203]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-04-15T00:00:00"}