{"id":"https://openalex.org/W4388874804","doi":"https://doi.org/10.1145/3552326.3587438","title":"Tabi: An Efficient Multi-Level Inference System for Large Language Models","display_name":"Tabi: An Efficient Multi-Level Inference System for Large Language Models","publication_year":2023,"publication_date":"2023-05-08","ids":{"openalex":"https://openalex.org/W4388874804","doi":"https://doi.org/10.1145/3552326.3587438"},"language":"en","primary_location":{"id":"doi:10.1145/3552326.3587438","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3552326.3587438","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the Eighteenth European Conference on Computer Systems","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://repository.hkust.edu.hk/ir/bitstream/1783.1-125990/1/125990-1.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5100607418","display_name":"Yiding Wang","orcid":"https://orcid.org/0000-0002-6596-2337"},"institutions":[{"id":"https://openalex.org/I200769079","display_name":"Hong Kong University of Science and Technology","ror":"https://ror.org/00q4vv597","country_code":"HK","type":"education","lineage":["https://openalex.org/I200769079"]}],"countries":["HK"],"is_corresponding":true,"raw_author_name":"Yiding Wang","raw_affiliation_strings":["Hong Kong University of Science and Technology, Clear Water Bay, Hong Kong"],"affiliations":[{"raw_affiliation_string":"Hong Kong University of Science and Technology, Clear Water Bay, Hong Kong","institution_ids":["https://openalex.org/I200769079"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100438001","display_name":"Kai Chen","orcid":"https://orcid.org/0000-0003-2587-6028"},"institutions":[{"id":"https://openalex.org/I200769079","display_name":"Hong Kong University of Science and Technology","ror":"https://ror.org/00q4vv597","country_code":"HK","type":"education","lineage":["https://openalex.org/I200769079"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Kai Chen","raw_affiliation_strings":["Hong Kong University of Science and Technology, Clear Water Bay, Hong Kong"],"affiliations":[{"raw_affiliation_string":"Hong Kong University of Science and Technology, Clear Water Bay, Hong Kong","institution_ids":["https://openalex.org/I200769079"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5067465324","display_name":"Haisheng Tan","orcid":"https://orcid.org/0000-0002-3133-1430"},"institutions":[{"id":"https://openalex.org/I126520041","display_name":"University of Science and Technology of China","ror":"https://ror.org/04c4dkn09","country_code":"CN","type":"education","lineage":["https://openalex.org/I126520041","https://openalex.org/I19820366"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Haisheng Tan","raw_affiliation_strings":["University of Science and Technology of China, Hefei, China"],"affiliations":[{"raw_affiliation_string":"University of Science and Technology of China, Hefei, China","institution_ids":["https://openalex.org/I126520041"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5100765132","display_name":"Kun Guo","orcid":"https://orcid.org/0000-0002-6270-2468"},"institutions":[{"id":"https://openalex.org/I80947539","display_name":"Fuzhou University","ror":"https://ror.org/011xvna82","country_code":"CN","type":"education","lineage":["https://openalex.org/I80947539"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Kun Guo","raw_affiliation_strings":["Fuzhou University, Fuzhou, China"],"affiliations":[{"raw_affiliation_string":"Fuzhou University, Fuzhou, China","institution_ids":["https://openalex.org/I80947539"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":4,"corresponding_author_ids":["https://openalex.org/A5100607418"],"corresponding_institution_ids":["https://openalex.org/I200769079"],"apc_list":null,"apc_paid":null,"fwci":9.4874,"has_fulltext":false,"cited_by_count":55,"citation_normalized_percentile":{"value":0.98567794,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":98,"max":100},"biblio":{"volume":null,"issue":null,"first_page":"233","last_page":"248"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.9998999834060669,"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":0.9998999834060669,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.9998000264167786,"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/T10201","display_name":"Speech Recognition and Synthesis","score":0.9876999855041504,"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":0.8013357520103455},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.7291878461837769},{"id":"https://openalex.org/keywords/discriminative-model","display_name":"Discriminative model","score":0.5384656190872192},{"id":"https://openalex.org/keywords/pruning","display_name":"Pruning","score":0.5216389298439026},{"id":"https://openalex.org/keywords/generative-model","display_name":"Generative model","score":0.4848405122756958},{"id":"https://openalex.org/keywords/offset","display_name":"Offset (computer science)","score":0.4839799404144287},{"id":"https://openalex.org/keywords/latency","display_name":"Latency (audio)","score":0.46985188126564026},{"id":"https://openalex.org/keywords/generative-grammar","display_name":"Generative grammar","score":0.42258888483047485},{"id":"https://openalex.org/keywords/machine-learning","display_name":"Machine learning","score":0.39352524280548096},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.35907724499702454},{"id":"https://openalex.org/keywords/telecommunications","display_name":"Telecommunications","score":0.08909836411476135}],"concepts":[{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.8013357520103455},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7291878461837769},{"id":"https://openalex.org/C97931131","wikidata":"https://www.wikidata.org/wiki/Q5282087","display_name":"Discriminative model","level":2,"score":0.5384656190872192},{"id":"https://openalex.org/C108010975","wikidata":"https://www.wikidata.org/wiki/Q500094","display_name":"Pruning","level":2,"score":0.5216389298439026},{"id":"https://openalex.org/C167966045","wikidata":"https://www.wikidata.org/wiki/Q5532625","display_name":"Generative model","level":3,"score":0.4848405122756958},{"id":"https://openalex.org/C175291020","wikidata":"https://www.wikidata.org/wiki/Q1156822","display_name":"Offset (computer science)","level":2,"score":0.4839799404144287},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.46985188126564026},{"id":"https://openalex.org/C39890363","wikidata":"https://www.wikidata.org/wiki/Q36108","display_name":"Generative grammar","level":2,"score":0.42258888483047485},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.39352524280548096},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.35907724499702454},{"id":"https://openalex.org/C76155785","wikidata":"https://www.wikidata.org/wiki/Q418","display_name":"Telecommunications","level":1,"score":0.08909836411476135},{"id":"https://openalex.org/C86803240","wikidata":"https://www.wikidata.org/wiki/Q420","display_name":"Biology","level":0,"score":0.0},{"id":"https://openalex.org/C6557445","wikidata":"https://www.wikidata.org/wiki/Q173113","display_name":"Agronomy","level":1,"score":0.0},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.0}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.1145/3552326.3587438","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3552326.3587438","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the Eighteenth European Conference on Computer Systems","raw_type":"proceedings-article"},{"id":"pmh:oai:repository.hkust.edu.hk:1783.1-125990","is_oa":true,"landing_page_url":"http://repository.hkust.edu.hk/ir/Record/1783.1-125990","pdf_url":"https://repository.hkust.edu.hk/ir/bitstream/1783.1-125990/1/125990-1.pdf","source":{"id":"https://openalex.org/S4306401796","display_name":"Rare & Special e-Zone (The Hong Kong University of Science and Technology)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I200769079","host_organization_name":"Hong Kong University of Science and Technology","host_organization_lineage":["https://openalex.org/I200769079"],"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":"Conference paper"}],"best_oa_location":{"id":"pmh:oai:repository.hkust.edu.hk:1783.1-125990","is_oa":true,"landing_page_url":"http://repository.hkust.edu.hk/ir/Record/1783.1-125990","pdf_url":"https://repository.hkust.edu.hk/ir/bitstream/1783.1-125990/1/125990-1.pdf","source":{"id":"https://openalex.org/S4306401796","display_name":"Rare & Special e-Zone (The Hong Kong University of Science and Technology)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I200769079","host_organization_name":"Hong Kong University of Science and Technology","host_organization_lineage":["https://openalex.org/I200769079"],"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":"Conference paper"},"sustainable_development_goals":[{"display_name":"Reduced inequalities","id":"https://metadata.un.org/sdg/10","score":0.699999988079071}],"awards":[{"id":"https://openalex.org/G1121271761","display_name":null,"funder_award_id":"Program","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G1477544716","display_name":null,"funder_award_id":"Guangdong","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G1683999813","display_name":null,"funder_award_id":"62062005","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G2243393327","display_name":null,"funder_award_id":"RGC TRS T41-603/20-R","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G3317480652","display_name":null,"funder_award_id":"Science","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G3606743899","display_name":null,"funder_award_id":"132009","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G391238517","display_name":null,"funder_award_id":", and","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G5994120800","display_name":null,"funder_award_id":"Natural","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G7033924410","display_name":null,"funder_award_id":"62132009","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G7049551338","display_name":null,"funder_award_id":"2022J01","funder_id":"https://openalex.org/F4320321878","funder_display_name":"Natural Science Foundation of Fujian Province"},{"id":"https://openalex.org/G7617473759","display_name":null,"funder_award_id":"2022J01118","funder_id":"https://openalex.org/F4320321878","funder_display_name":"Natural Science Foundation of Fujian Province"},{"id":"https://openalex.org/G7726157001","display_name":null,"funder_award_id":"Grant No.","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G8083849180","display_name":null,"funder_award_id":"2022J0111","funder_id":"https://openalex.org/F4320321878","funder_display_name":"Natural Science Foundation of Fujian Province"},{"id":"https://openalex.org/G8208620042","display_name":null,"funder_award_id":"2132009","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"}],"funders":[{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"},{"id":"https://openalex.org/F4320321878","display_name":"Natural Science Foundation of Fujian Province","ror":null}],"has_content":{"grobid_xml":false,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4388874804.pdf"},"referenced_works_count":63,"referenced_works":["https://openalex.org/W95608104","https://openalex.org/W1850527962","https://openalex.org/W2128073546","https://openalex.org/W2565600385","https://openalex.org/W2606492274","https://openalex.org/W2612690371","https://openalex.org/W2734941459","https://openalex.org/W2739542029","https://openalex.org/W2750779823","https://openalex.org/W2810469995","https://openalex.org/W2896958211","https://openalex.org/W2923014074","https://openalex.org/W2946794439","https://openalex.org/W2951333013","https://openalex.org/W2952357537","https://openalex.org/W2956461999","https://openalex.org/W2962677625","https://openalex.org/W2963516811","https://openalex.org/W2970454332","https://openalex.org/W2971118045","https://openalex.org/W2972324944","https://openalex.org/W2982157693","https://openalex.org/W2986193249","https://openalex.org/W3007007518","https://openalex.org/W3021636956","https://openalex.org/W3033527233","https://openalex.org/W3033737024","https://openalex.org/W3034742519","https://openalex.org/W3035038672","https://openalex.org/W3046754651","https://openalex.org/W3094502228","https://openalex.org/W3098605233","https://openalex.org/W3101163004","https://openalex.org/W3102725307","https://openalex.org/W3104939451","https://openalex.org/W3118485687","https://openalex.org/W3130689885","https://openalex.org/W3130716829","https://openalex.org/W3133702157","https://openalex.org/W3153004963","https://openalex.org/W3159727696","https://openalex.org/W3164703200","https://openalex.org/W3165698711","https://openalex.org/W3168124404","https://openalex.org/W3172942063","https://openalex.org/W3181029726","https://openalex.org/W3196976833","https://openalex.org/W3214897310","https://openalex.org/W4212883601","https://openalex.org/W4220738176","https://openalex.org/W4225729912","https://openalex.org/W4282577879","https://openalex.org/W4287777801","https://openalex.org/W4288347855","https://openalex.org/W4290991649","https://openalex.org/W6639048329","https://openalex.org/W6739651123","https://openalex.org/W6739901393","https://openalex.org/W6762945437","https://openalex.org/W6771626834","https://openalex.org/W6777017071","https://openalex.org/W6781533629","https://openalex.org/W6798686915"],"related_works":["https://openalex.org/W4396941953","https://openalex.org/W2093104230","https://openalex.org/W2987280934","https://openalex.org/W4390874210","https://openalex.org/W4384918963","https://openalex.org/W4365211920","https://openalex.org/W2128027845","https://openalex.org/W3014948380","https://openalex.org/W4386184937","https://openalex.org/W4394728283"],"abstract_inverted_index":{"Today's":[0],"trend":[1],"of":[2,13,32,52,109],"building":[3],"ever":[4],"larger":[5],"language":[6,15],"models":[7,74,86,111],"(LLMs),":[8],"while":[9,166],"pushing":[10],"the":[11,21,29,42,97,106,133,164],"performance":[12],"natural":[14],"processing,":[16],"adds":[17],"significant":[18],"latency":[19,157],"to":[20,28,35,101,104,117,131],"inference":[22,62,67],"stage.":[23],"We":[24,139],"observe":[25],"that":[26,69,152],"due":[27],"diminishing":[30],"returns":[31],"adding":[33],"parameters":[34],"LLMs,":[36],"a":[37,46,50,65,92],"smaller":[38],"model":[39],"could":[40],"make":[41],"same":[43],"prediction":[44],"as":[45],"costly":[47],"LLM":[48],"for":[49,78,84],"majority":[51],"queries.":[53],"Based":[54],"on":[55],"this":[56],"observation,":[57],"we":[58],"design":[59],"Tabi,":[60],"an":[61],"system":[63,134],"with":[64,144],"multi-level":[66],"engine":[68],"serves":[70],"queries":[71],"using":[72],"small":[73,110],"and":[75,127,136,141,147],"optional":[76],"LLMs":[77],"demanding":[79],"applications.":[80],"Tabi":[81,95,143,153],"is":[82],"optimized":[83],"discriminative":[85],"(i.e.,":[87],"not":[88],"generative":[89],"LLMs)":[90],"in":[91],"serving":[93],"framework.":[94],"uses":[96,123],"calibrated":[98],"confidence":[99],"score":[100],"decide":[102],"whether":[103],"return":[105],"accurate":[107],"results":[108],"extremely":[112],"fast":[113],"or":[114],"re-route":[115],"them":[116],"LLMs.":[118],"For":[119],"re-routed":[120],"queries,":[121],"it":[122],"attention-based":[124],"word":[125],"pruning":[126],"weighted":[128],"ensemble":[129],"techniques":[130],"offset":[132],"overhead":[135],"accuracy":[137,170],"loss.":[138],"implement":[140],"evaluate":[142],"multiple":[145],"tasks":[146],"models.":[148],"Our":[149],"result":[150],"shows":[151],"achieves":[154],"21%-40%":[155],"average":[156],"reduction":[158],"(with":[159],"comparable":[160],"tail":[161],"latency)":[162],"over":[163],"state-of-the-art":[165],"meeting":[167],"LLM-grade":[168],"high":[169],"targets.":[171]},"counts_by_year":[{"year":2026,"cited_by_count":4},{"year":2025,"cited_by_count":23},{"year":2024,"cited_by_count":22},{"year":2023,"cited_by_count":6}],"updated_date":"2026-04-13T07:58:08.660418","created_date":"2025-10-10T00:00:00"}