{"id":"https://openalex.org/W4403487110","doi":"https://doi.org/10.3233/faia240927","title":"SUBLLM: A Novel Efficient Architecture with Token Sequence Subsampling for LLM","display_name":"SUBLLM: A Novel Efficient Architecture with Token Sequence Subsampling for LLM","publication_year":2024,"publication_date":"2024-10-16","ids":{"openalex":"https://openalex.org/W4403487110","doi":"https://doi.org/10.3233/faia240927"},"language":"en","primary_location":{"id":"doi:10.3233/faia240927","is_oa":true,"landing_page_url":"https://doi.org/10.3233/faia240927","pdf_url":null,"source":{"id":"https://openalex.org/S4210201731","display_name":"Frontiers in artificial intelligence and applications","issn_l":"0922-6389","issn":["0922-6389","1879-8314"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"journal"},"license":"cc-by-nc","license_id":"https://openalex.org/licenses/cc-by-nc","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Frontiers in Artificial Intelligence and Applications","raw_type":"book-chapter"},"type":"book-chapter","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"hybrid","oa_url":"https://doi.org/10.3233/faia240927","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5013004114","display_name":"Quandong Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Quandong Wang","raw_affiliation_strings":["Xiaomi AI Lab, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Xiaomi AI Lab, China","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5103263121","display_name":"Yuxuan Yuan","orcid":"https://orcid.org/0009-0003-2850-4366"},"institutions":[{"id":"https://openalex.org/I191208505","display_name":"Xiamen University","ror":"https://ror.org/00mcjh785","country_code":"CN","type":"education","lineage":["https://openalex.org/I191208505"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yuxuan Yuan","raw_affiliation_strings":["Department of Artificial Intelligence, School of Informatics, Xiamen University, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Artificial Intelligence, School of Informatics, Xiamen University, China","institution_ids":["https://openalex.org/I191208505"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101744929","display_name":"Xiaoyu Yang","orcid":"https://orcid.org/0000-0002-7418-3239"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xiaoyu Yang","raw_affiliation_strings":["Xiaomi AI Lab, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Xiaomi AI Lab, China","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5111227759","display_name":"Ruike Zhang","orcid":null},"institutions":[{"id":"https://openalex.org/I19820366","display_name":"Chinese Academy of Sciences","ror":"https://ror.org/034t30j35","country_code":"CN","type":"government","lineage":["https://openalex.org/I19820366"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Ruike Zhang","raw_affiliation_strings":["Institute of Automation, Chinese Academy of Sciences"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Institute of Automation, Chinese Academy of Sciences","institution_ids":["https://openalex.org/I19820366"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100675363","display_name":"Kang Zhao","orcid":"https://orcid.org/0000-0001-7927-4079"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kang Zhao","raw_affiliation_strings":["Xiaomi AI Lab, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Xiaomi AI Lab, China","institution_ids":[]}]},{"author_position":"middle","author":{"id":null,"display_name":"Wei Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wei Liu","raw_affiliation_strings":["Xiaomi AI Lab, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Xiaomi AI Lab, China","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5054843960","display_name":"Jian Luan","orcid":"https://orcid.org/0000-0002-2383-226X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jian Luan","raw_affiliation_strings":["Xiaomi AI Lab, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Xiaomi AI Lab, China","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5084286453","display_name":"Daniel Povey","orcid":"https://orcid.org/0000-0002-0611-3634"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Daniel Povey","raw_affiliation_strings":["Xiaomi AI Lab, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Xiaomi AI Lab, China","institution_ids":[]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5101341567","display_name":"Bin Wang","orcid":"https://orcid.org/0009-0008-2038-1927"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Bin Wang","raw_affiliation_strings":["Xiaomi AI Lab, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Xiaomi AI Lab, China","institution_ids":[]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":9,"corresponding_author_ids":["https://openalex.org/A5013004114"],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.43170932,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"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/T12479","display_name":"Web Application Security Vulnerabilities","score":0.9013000130653381,"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/T12479","display_name":"Web Application Security Vulnerabilities","score":0.9013000130653381,"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/security-token","display_name":"Security token","score":0.6931906938552856},{"id":"https://openalex.org/keywords/architecture","display_name":"Architecture","score":0.5921387672424316},{"id":"https://openalex.org/keywords/sequence","display_name":"Sequence (biology)","score":0.5907109379768372},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.588354766368866},{"id":"https://openalex.org/keywords/computer-architecture","display_name":"Computer architecture","score":0.3550547957420349},{"id":"https://openalex.org/keywords/computer-network","display_name":"Computer network","score":0.20002350211143494},{"id":"https://openalex.org/keywords/biology","display_name":"Biology","score":0.11267063021659851},{"id":"https://openalex.org/keywords/art","display_name":"Art","score":0.08494901657104492},{"id":"https://openalex.org/keywords/visual-arts","display_name":"Visual arts","score":0.03185385465621948}],"concepts":[{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.6931906938552856},{"id":"https://openalex.org/C123657996","wikidata":"https://www.wikidata.org/wiki/Q12271","display_name":"Architecture","level":2,"score":0.5921387672424316},{"id":"https://openalex.org/C2778112365","wikidata":"https://www.wikidata.org/wiki/Q3511065","display_name":"Sequence (biology)","level":2,"score":0.5907109379768372},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.588354766368866},{"id":"https://openalex.org/C118524514","wikidata":"https://www.wikidata.org/wiki/Q173212","display_name":"Computer architecture","level":1,"score":0.3550547957420349},{"id":"https://openalex.org/C31258907","wikidata":"https://www.wikidata.org/wiki/Q1301371","display_name":"Computer network","level":1,"score":0.20002350211143494},{"id":"https://openalex.org/C86803240","wikidata":"https://www.wikidata.org/wiki/Q420","display_name":"Biology","level":0,"score":0.11267063021659851},{"id":"https://openalex.org/C142362112","wikidata":"https://www.wikidata.org/wiki/Q735","display_name":"Art","level":0,"score":0.08494901657104492},{"id":"https://openalex.org/C153349607","wikidata":"https://www.wikidata.org/wiki/Q36649","display_name":"Visual arts","level":1,"score":0.03185385465621948},{"id":"https://openalex.org/C54355233","wikidata":"https://www.wikidata.org/wiki/Q7162","display_name":"Genetics","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.3233/faia240927","is_oa":true,"landing_page_url":"https://doi.org/10.3233/faia240927","pdf_url":null,"source":{"id":"https://openalex.org/S4210201731","display_name":"Frontiers in artificial intelligence and applications","issn_l":"0922-6389","issn":["0922-6389","1879-8314"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"journal"},"license":"cc-by-nc","license_id":"https://openalex.org/licenses/cc-by-nc","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Frontiers in Artificial Intelligence and Applications","raw_type":"book-chapter"}],"best_oa_location":{"id":"doi:10.3233/faia240927","is_oa":true,"landing_page_url":"https://doi.org/10.3233/faia240927","pdf_url":null,"source":{"id":"https://openalex.org/S4210201731","display_name":"Frontiers in artificial intelligence and applications","issn_l":"0922-6389","issn":["0922-6389","1879-8314"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"journal"},"license":"cc-by-nc","license_id":"https://openalex.org/licenses/cc-by-nc","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Frontiers in Artificial Intelligence and Applications","raw_type":"book-chapter"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W4388335561","https://openalex.org/W2970530566","https://openalex.org/W4288261899","https://openalex.org/W4307309205","https://openalex.org/W2967478618","https://openalex.org/W4385009901","https://openalex.org/W2038503502"],"abstract_inverted_index":{"While":[0],"Large":[1,32],"Language":[2,33],"Models":[3],"(LLMs)":[4],"have":[5],"achieved":[6],"remarkable":[7],"success":[8],"in":[9,84],"various":[10],"fields,":[11],"the":[12,40,58,61,65,69,78,144],"efficiency":[13],"of":[14],"training":[15,86,131],"and":[16,48,68,87,107,123,132,140],"inference":[17,88,133],"remains":[18],"a":[19],"major":[20],"challenge.":[21],"To":[22],"address":[23],"this":[24],"issue,":[25],"we":[26],"propose":[27],"SUBLLM,":[28],"short":[29],"for":[30,56],"Subsampling-Upsampling-Bypass":[31],"Model,":[34],"an":[35],"innovative":[36],"architecture":[37],"that":[38],"extends":[39],"core":[41],"decoder-only":[42],"framework":[43],"by":[44,105,110,119,126,138],"incorporating":[45],"subsampling,":[46],"upsampling,":[47],"bypass":[49,70],"modules.":[50],"The":[51,130],"subsampling":[52],"modules":[53,63,71],"are":[54],"responsible":[55],"shortening":[57],"sequence,":[59],"while":[60,95],"upsampling":[62],"restore":[64],"sequence":[66],"length,":[67],"enhance":[72],"convergence.":[73],"In":[74,114],"comparison":[75],"to":[76,121,149],"LLaMA,":[77],"proposed":[79],"SUBLLM":[80,102],"exhibits":[81],"significant":[82],"enhancements":[83],"both":[85],"speeds":[89,104,118,134],"as":[90,92],"well":[91],"memory":[93,109,125],"usage,":[94],"maintaining":[96],"competitive":[97],"few-shot":[98],"performance.":[99],"During":[100],"training,":[101],"increases":[103],"26%":[106],"cuts":[108],"10GB":[111],"per":[112,128],"GPU.":[113,129],"inference,":[115],"it":[116],"boosts":[117],"up":[120],"37%":[122],"reduces":[124],"1GB":[127],"can":[135],"be":[136],"enhanced":[137],"34%":[139],"52%":[141],"respectively":[142],"when":[143],"context":[145],"window":[146],"is":[147,153],"expanded":[148],"8192.":[150],"Our":[151],"code":[152],"available":[154],"at":[155],"https://github.com/XiaoMi/subllm.":[156]},"counts_by_year":[],"updated_date":"2026-05-21T06:26:12.895304","created_date":"2025-10-10T00:00:00"}