{"id":"https://openalex.org/W7204909591","doi":"https://doi.org/10.48550/arxiv.2608.28968","title":"Efficient GPU Retrieval for Semantic Search","display_name":"Efficient GPU Retrieval for Semantic Search","publication_year":2026,"publication_date":"2026-08-29","ids":{"openalex":"https://openalex.org/W7204909591","doi":"https://doi.org/10.48550/arxiv.2608.28968"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2608.28968","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.28968","pdf_url":"https://arxiv.org/pdf/2608.28968","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2608.28968","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5021439763","display_name":"Dhritiman Das","orcid":"https://orcid.org/0000-0001-6627-0618","observed_orcids":["https://orcid.org/0000-0001-6627-0618"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Das, Dhritiman","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5148906059","display_name":"Chujie Zheng","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zheng, Chujie","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5010096373","display_name":"Ronak Kaoshik","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kaoshik, Ronak","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5148930864","display_name":"Pratik Dixit","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Dixit, Pratik","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5159888508","display_name":"Vishal Shah","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shah, Vishal","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5148848887","display_name":"Yanbo Li","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Yanbo","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5148835707","display_name":"Jiahao Xu","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xu, Jiahao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5148916765","display_name":"Manika Agarwal","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Agarwal, Manika","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5000407229","display_name":"Chinmay Naik","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Naik, Chinmay","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5148928879","display_name":"Lingyu Zhang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Lingyu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5067046150","display_name":"Chetan Bhole","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Bhole, Chetan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5148875651","display_name":"Chirag Bhanuprasad Mehta","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Mehta, Chirag Bhanuprasad","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5038942290","display_name":"Meng Zheng","orcid":"https://orcid.org/0000-0002-6677-2017","observed_orcids":["https://orcid.org/0000-0002-6677-2017"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zheng, Meng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5148925233","display_name":"Puneet Singh Ahluwalia","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ahluwalia, Puneet Singh","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5148859753","display_name":"Shirisha Singh","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Singh, Shirisha","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5103254650","display_name":"Ping Jin","orcid":"https://orcid.org/0000-0003-2498-2654","observed_orcids":["https://orcid.org/0000-0003-2498-2654"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jin, Ping","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5047871918","display_name":"Manas Apte","orcid":"https://orcid.org/0009-0001-1775-7152","observed_orcids":["https://orcid.org/0009-0001-1775-7152"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Apte, Manas","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125538468","display_name":"Gokulraj Mohanasundaram","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Mohanasundaram, Gokulraj","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5107077744","display_name":"Tugrul Bingol","orcid":"https://orcid.org/0009-0008-3082-0356","observed_orcids":["https://orcid.org/0009-0008-3082-0356"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Bingol, Tugrul","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125564425","display_name":"Raghavan Muthuregunathan","orcid":"https://orcid.org/0009-0003-9652-1539","observed_orcids":["https://orcid.org/0009-0003-9652-1539"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Muthuregunathan, Raghavan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5079826085","display_name":"Fedor Borisyuk","orcid":"https://orcid.org/0009-0005-8171-7656","observed_orcids":["https://orcid.org/0009-0005-8171-7656"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Borisyuk, Fedor","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/T10286","display_name":"Information Retrieval and Search Behavior","score":0.9843999743461609,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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/T10286","display_name":"Information Retrieval and Search Behavior","score":0.9843999743461609,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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":0.01119999960064888,"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/T10203","display_name":"Recommender Systems and Techniques","score":0.001500000013038516,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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/semantic-search","display_name":"semantic search","score":0.8849999904632568},{"id":"https://openalex.org/keywords/multi-vector-retrieval","display_name":"multi-vector retrieval","score":0.6690000295639038}],"concepts":[{"id":"https://openalex.org/C158154518","wikidata":"https://www.wikidata.org/wiki/Q7310970","display_name":"Relevance (law)","level":2,"score":0.7493000030517578},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7189000248908997},{"id":"https://openalex.org/C43122875","wikidata":"https://www.wikidata.org/wiki/Q5428522","display_name":"Facet (psychology)","level":4,"score":0.6902999877929688},{"id":"https://openalex.org/C100660578","wikidata":"https://www.wikidata.org/wiki/Q18733","display_name":"Recall","level":2,"score":0.609000027179718},{"id":"https://openalex.org/C2776359362","wikidata":"https://www.wikidata.org/wiki/Q2145286","display_name":"Representation (politics)","level":3,"score":0.5809999704360962},{"id":"https://openalex.org/C2780762811","wikidata":"https://www.wikidata.org/wiki/Q1784941","display_name":"Cosine similarity","level":3,"score":0.5600000023841858},{"id":"https://openalex.org/C103278499","wikidata":"https://www.wikidata.org/wiki/Q254465","display_name":"Similarity (geometry)","level":3,"score":0.4754999876022339},{"id":"https://openalex.org/C23123220","wikidata":"https://www.wikidata.org/wiki/Q816826","display_name":"Information retrieval","level":1,"score":0.4514000117778778},{"id":"https://openalex.org/C81669768","wikidata":"https://www.wikidata.org/wiki/Q2359161","display_name":"Precision and recall","level":2,"score":0.40619999170303345},{"id":"https://openalex.org/C1667742","wikidata":"https://www.wikidata.org/wiki/Q10927554","display_name":"Image retrieval","level":3,"score":0.37549999356269836},{"id":"https://openalex.org/C157764524","wikidata":"https://www.wikidata.org/wiki/Q1383412","display_name":"Throughput","level":3,"score":0.36059999465942383},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3431999981403351},{"id":"https://openalex.org/C143271835","wikidata":"https://www.wikidata.org/wiki/Q254515","display_name":"Similitude","level":2,"score":0.30320000648498535},{"id":"https://openalex.org/C2779227376","wikidata":"https://www.wikidata.org/wiki/Q6505497","display_name":"Layer (electronics)","level":2,"score":0.3009999990463257},{"id":"https://openalex.org/C115961682","wikidata":"https://www.wikidata.org/wiki/Q860623","display_name":"Image (mathematics)","level":2,"score":0.3003000020980835},{"id":"https://openalex.org/C132525143","wikidata":"https://www.wikidata.org/wiki/Q141488","display_name":"Graph","level":2,"score":0.2976999878883362},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.29670000076293945},{"id":"https://openalex.org/C130318100","wikidata":"https://www.wikidata.org/wiki/Q2268914","display_name":"Semantic similarity","level":2,"score":0.2838999927043915},{"id":"https://openalex.org/C2987098735","wikidata":"https://www.wikidata.org/wiki/Q3808900","display_name":"Recall rate","level":2,"score":0.27810001373291016},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.274399995803833},{"id":"https://openalex.org/C90805587","wikidata":"https://www.wikidata.org/wiki/Q10944557","display_name":"Word (group theory)","level":2,"score":0.2702000141143799},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.26829999685287476},{"id":"https://openalex.org/C14036430","wikidata":"https://www.wikidata.org/wiki/Q3736076","display_name":"Function (biology)","level":2,"score":0.2648000121116638},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.26109999418258667}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2608.28968","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.28968","pdf_url":"https://arxiv.org/pdf/2608.28968","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2608.28968","doi":"https://doi.org/10.48550/arxiv.2608.28968","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2608.28968","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2608.28968","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.28968","pdf_url":"https://arxiv.org/pdf/2608.28968","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W2963605131","https://openalex.org/W4403577926","https://openalex.org/W7128513156","https://openalex.org/W4414035082","https://openalex.org/W7130706945","https://openalex.org/W7148445520","https://openalex.org/W4379933301","https://openalex.org/W7151489934","https://openalex.org/W7122627845","https://openalex.org/W2323996792"],"abstract_inverted_index":{"Semantic":[0],"Search":[1],"on":[2,71,76],"LinkedIn":[3],"must":[4,41],"retrieve":[5],"relevant":[6],"profiles":[7],"from":[8,230,237],"a":[9,45,55,68,88,157,172,218,242],"corpus":[10],"of":[11,13,81,207],"hundreds":[12],"millions":[14],"in":[15,25,29],"response":[16],"to":[17,232,239],"natural-language":[18],"queries":[19],"such":[20],"as":[21],"\"a":[22],"fintech":[23],"founder":[24],"Berlin":[26],"who":[27],"worked":[28],"payments.\"":[30],"The":[31],"deployed":[32],"relevance":[33,155],"policy":[34],"is":[35,115],"bottleneck-oriented:":[36],"every":[37],"active":[38],"non-negotiable":[39],"facet":[40,60,73,165],"be":[42],"satisfied,":[43],"and":[44,122,145,189,234],"pre-existing":[46],"LLM":[47],"Graded":[48],"Relevance":[49],"(GR)":[50],"judge":[51,228],"operationalizes":[52],"this":[53,112,151,169],"through":[54],"fixed":[56],"min/median":[57,104],"aggregation":[58],"over":[59,156,211],"grades.":[61],"Cosine":[62],"similarity":[63],"instead":[64],"averages":[65],"evidence,":[66],"letting":[67],"strong":[69],"match":[70],"one":[72],"mask":[74],"failure":[75],"another,":[77],"capping":[78],"the":[79,82,102,181,225,248],"recall":[80,209],"first-stage":[83],"(L0)":[84],"retriever.":[85],"We":[86,167],"present":[87],"policy-aligned":[89],"retrieval":[90],"framework:":[91],"embeddings":[92],"are":[93],"partitioned":[94],"into":[95],"eight":[96],"category-supervised":[97],"segments":[98],"whose":[99],"scores":[100,180],"follow":[101],"same":[103],"rule":[105],"at":[106,210],"serving":[107],"time;":[108],"for":[109],"multi-vector":[110],"retrieval,":[111],"segment":[113],"score":[114],"computed":[116],"independently":[117,246],"per":[118,214],"tagged":[119],"document":[120],"slot":[121],"maximized":[123],"across":[124,142,164],"slots.":[125],"A":[126],"lightweight":[127],"single-slot":[128],"Stage-1":[129,190],"scorer":[130],"generates":[131],"high-recall":[132],"candidates,":[133],"while":[134],"scale-invariant":[135],"relative-norm":[136],"gating":[137],"keeps":[138],"category":[139],"activation":[140],"consistent":[141],"training,":[143],"evaluation,":[144],"serving.":[146],"On":[147],"21K":[148],"held-out":[149],"queries,":[150],"representation":[152],"improves":[153],"offline":[154],"matched-capacity":[158],"baseline,":[159],"with":[160,171,241],"gains":[161],"broadly":[162],"distributed":[163],"combinations.":[166],"serve":[168],"framework":[170],"two-stage":[173],"GPU":[174],"architecture:":[175],"an":[176,196,201],"FP8":[177],"coarse":[178],"ranker":[179],"full":[182],"corpus,":[183],"increasing":[184],"per-shard":[185],"capacity":[186],"by":[187,193],"71%":[188],"matmul":[191],"throughput":[192],"36%,":[194],"then":[195],"FP16":[197],"stage":[198],"exactly":[199],"re-ranks":[200],"oversampled":[202],"candidate":[203],"set,":[204],"recovering":[205],"99.6-99.8%":[206],"full-FP16":[208],"500":[212],"QPS":[213],"shard":[215],"replica.":[216],"In":[217],"member-randomized":[219],"A/B":[220],"test,":[221],"exploratory-query":[222],"Precision@10":[223,249],"under":[224],"unchanged":[226],"GR":[227],"rises":[229],"63.7%":[231],"79.0%":[233],"navigational":[235],"Precision@1":[236],"65.5%":[238],"74.7%,":[240],"blinded":[243],"human":[244],"evaluation":[245],"confirming":[247],"gain.":[250]},"counts_by_year":[],"updated_date":"2026-10-28T10:33:24.094632","created_date":"2026-09-02T00:00:00"}