{"id":"https://openalex.org/W7160642440","doi":"https://doi.org/10.48550/arxiv.2605.06285","title":"LatentRAG: Latent Reasoning and Retrieval for Efficient Agentic RAG","display_name":"LatentRAG: Latent Reasoning and Retrieval for Efficient Agentic RAG","publication_year":2026,"publication_date":"2026-05-07","ids":{"openalex":"https://openalex.org/W7160642440","doi":"https://doi.org/10.48550/arxiv.2605.06285"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.06285","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.06285","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":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2605.06285","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5101750180","display_name":"Yijia Zheng","orcid":"https://orcid.org/0000-0002-6585-8273"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zheng, Yijia","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5070684680","display_name":"Marcel Worring","orcid":"https://orcid.org/0000-0003-4097-4136"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Worring, Marcel","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/T10028","display_name":"Topic Modeling","score":0.8805000185966492,"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"}},"topics":[{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.8805000185966492,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.07259999960660934,"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/T10664","display_name":"Sentiment Analysis and Opinion Mining","score":0.00430000014603138,"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":0.670799970626831},{"id":"https://openalex.org/keywords/natural-language","display_name":"Natural language","score":0.46860000491142273},{"id":"https://openalex.org/keywords/question-answering","display_name":"Question answering","score":0.4458000063896179},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.40119999647140503},{"id":"https://openalex.org/keywords/latent-semantic-analysis","display_name":"Latent semantic analysis","score":0.3880000114440918},{"id":"https://openalex.org/keywords/probabilistic-latent-semantic-analysis","display_name":"Probabilistic latent semantic analysis","score":0.3741999864578247},{"id":"https://openalex.org/keywords/transparency","display_name":"Transparency (behavior)","score":0.34689998626708984},{"id":"https://openalex.org/keywords/process","display_name":"Process (computing)","score":0.3409000039100647}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7414000034332275},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.670799970626831},{"id":"https://openalex.org/C195324797","wikidata":"https://www.wikidata.org/wiki/Q33742","display_name":"Natural language","level":2,"score":0.46860000491142273},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.46050000190734863},{"id":"https://openalex.org/C44291984","wikidata":"https://www.wikidata.org/wiki/Q1074173","display_name":"Question answering","level":2,"score":0.4458000063896179},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.4296000003814697},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.40119999647140503},{"id":"https://openalex.org/C170133592","wikidata":"https://www.wikidata.org/wiki/Q1806883","display_name":"Latent semantic analysis","level":2,"score":0.3880000114440918},{"id":"https://openalex.org/C112933361","wikidata":"https://www.wikidata.org/wiki/Q2845258","display_name":"Probabilistic latent semantic analysis","level":2,"score":0.3741999864578247},{"id":"https://openalex.org/C2780233690","wikidata":"https://www.wikidata.org/wiki/Q535347","display_name":"Transparency (behavior)","level":2,"score":0.34689998626708984},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.3409000039100647},{"id":"https://openalex.org/C51167844","wikidata":"https://www.wikidata.org/wiki/Q4422623","display_name":"Latent variable","level":2,"score":0.3224000036716461},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.31779998540878296},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.3005000054836273},{"id":"https://openalex.org/C184337299","wikidata":"https://www.wikidata.org/wiki/Q1437428","display_name":"Semantics (computer science)","level":2,"score":0.2896000146865845},{"id":"https://openalex.org/C2780586882","wikidata":"https://www.wikidata.org/wiki/Q7520643","display_name":"Simple (philosophy)","level":2,"score":0.287200003862381},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.2831999957561493},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.26980000734329224},{"id":"https://openalex.org/C2778572836","wikidata":"https://www.wikidata.org/wiki/Q380933","display_name":"Space (punctuation)","level":2,"score":0.2696000039577484},{"id":"https://openalex.org/C137878579","wikidata":"https://www.wikidata.org/wiki/Q9636076","display_name":"Joint attention","level":3,"score":0.2653999924659729},{"id":"https://openalex.org/C2776608160","wikidata":"https://www.wikidata.org/wiki/Q4785462","display_name":"Natural (archaeology)","level":2,"score":0.2651999890804291},{"id":"https://openalex.org/C160920958","wikidata":"https://www.wikidata.org/wiki/Q7662746","display_name":"Synthetic data","level":2,"score":0.25440001487731934},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.25380000472068787}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.06285","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.06285","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":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2605.06285","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.06285","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":"Preprint"},"sustainable_development_goals":[{"display_name":"Quality Education","id":"https://metadata.un.org/sdg/4","score":0.6702796816825867}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Single-step":[0],"retrieval-augmented":[1],"generation":[2,70],"(RAG)":[3],"provides":[4],"an":[5],"efficient":[6],"way":[7],"to":[8,53,67,96,188],"incorporate":[9,162],"external":[10],"information":[11],"for":[12,116],"simple":[13],"question":[14],"answering":[15],"tasks":[16],"but":[17],"struggles":[18],"with":[19,31,56,133,205],"complex":[20],"questions.":[21],"Agentic":[22],"RAG":[23,191],"extends":[24],"this":[25,78],"paradigm":[26],"by":[27,197],"replacing":[28],"single-step":[29,207],"retrieval":[30,58,91,135,142],"a":[32,44,83,126,163],"multi-step":[33],"process,":[34],"in":[35,125,137],"which":[36],"the":[37,57,68,122,138,202],"large":[38],"language":[39,94,107],"model":[40],"(LLM)":[41],"acts":[42],"as":[43],"search":[45],"agent":[46],"that":[47,86,104,168,183],"generates":[48],"intermediate":[49],"thoughts":[50,73,108,117],"and":[51,74,90,118,147,155],"subqueries":[52,110,119],"iteratively":[54],"interact":[55],"system.":[59],"This":[60],"iterative":[61],"process":[62],"incurs":[63],"substantial":[64],"latency":[65,196,203],"due":[66],"autoregressive":[69],"of":[71],"lengthy":[72],"subqueries.":[75],"To":[76,152],"address":[77],"limitation,":[79],"we":[80,161],"propose":[81],"LatentRAG,":[82],"novel":[84],"framework":[85],"shifts":[87],"both":[88],"reasoning":[89],"from":[92,121],"discrete":[93],"space":[95],"continuous":[97],"latent":[98,114,139,144,159,165,170],"space.":[99],"Unlike":[100],"existing":[101],"explicit":[102,189],"methods":[103,192],"generate":[105],"natural":[106,174],"or":[109],"token-by-token,":[111],"LatentRAG":[112,184],"produces":[113],"tokens":[115,146,171],"directly":[120],"hidden":[123],"states":[124],"single":[127],"forward":[128],"pass.":[129],"We":[130],"align":[131],"LLMs":[132],"dense":[134],"models":[136],"space,":[140],"enabling":[141],"over":[143],"subquery":[145],"supporting":[148],"end-to-end":[149],"joint":[150],"optimization.":[151],"improve":[153],"transparency":[154],"encourage":[156],"semantically":[157],"meaningful":[158],"representations,":[160],"parallel":[164],"decoding":[166],"mechanism":[167],"translates":[169],"back":[172],"into":[173],"language.":[175],"Extensive":[176],"experiments":[177],"on":[178],"seven":[179],"benchmark":[180],"datasets":[181],"show":[182],"achieves":[185],"performance":[186],"comparable":[187],"agentic":[190],"while":[193],"reducing":[194],"inference":[195],"approximately":[198],"90%,":[199],"substantially":[200],"narrowing":[201],"gap":[204],"traditional":[206],"RAG.":[208]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-05-09T00:00:00"}