{"id":"https://openalex.org/W7162072155","doi":"https://doi.org/10.1007/s10462-026-11538-1","title":"On-device large language models: a survey of model compression and system optimization","display_name":"On-device large language models: a survey of model compression and system optimization","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7162072155","doi":"https://doi.org/10.1007/s10462-026-11538-1"},"language":"en","primary_location":{"id":"pmh:oai:repository.hkust.edu.hk:1783.1-173943","is_oa":false,"landing_page_url":"http://repository.hkust.edu.hk/ir/Record/1783.1-173943","pdf_url":null,"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":"Article"},"type":"article","indexed_in":[],"open_access":{"is_oa":false,"oa_status":"closed","oa_url":null,"any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5149078277","display_name":"Wanyi Chen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen, Wanyi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5149109789","display_name":"Junhao Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Junhao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5149014044","display_name":"Yiwei Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Yiwei","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5148984705","display_name":"Yufan Shi","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shi, Yufan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5149100508","display_name":"Tianyi Jiang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jiang, Tianyi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5110123643","display_name":"Shudong Zhou","orcid":"https://orcid.org/0009-0001-3887-2432"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhou, Shengxian","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5149014310","display_name":"Chenxu Wu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wu, Chenxu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5077911588","display_name":"Andi Zhang","orcid":"https://orcid.org/0009-0007-4855-5442"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Andi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5149096562","display_name":"Chenyue Zhou","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhou, Chenyue","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5033050248","display_name":"Minxuan Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Minxuan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5149079089","display_name":"Xinyu Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Xinyu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5149077515","display_name":"Xiaoshuai Hao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hao, Xiaoshuai","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5029617620","display_name":"Yinan Wu","orcid":"https://orcid.org/0009-0000-9149-3389"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wu, Yinan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5149096028","display_name":"Yichen Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Yichen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5149061916","display_name":"Yuwei Hu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hu, Yuwei","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5000560558","display_name":"Zhao Cao","orcid":"https://orcid.org/0000-0001-7648-0184"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Cao, Zhao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5148977864","display_name":"Yang Lu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lu, Yang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5149029348","display_name":"Mengke Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Mengke","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5020661081","display_name":"Yanbiao Ma","orcid":"https://orcid.org/0000-0002-8472-1475"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ma, Yanbiao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5148958442","display_name":"Zhiwu Lu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lu, Zhiwu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5046605531","display_name":"Jungong Han","orcid":"https://orcid.org/0000-0003-4361-956X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Han, Jungong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5149005382","display_name":"Yike Guo","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Guo, Yike","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":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.53199566,"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/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.39100000262260437,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"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/T10054","display_name":"Parallel Computing and Optimization Techniques","score":0.39100000262260437,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"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/T14347","display_name":"Big Data and Digital Economy","score":0.10279999673366547,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.06030000001192093,"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/memory-footprint","display_name":"Memory footprint","score":0.5425000190734863},{"id":"https://openalex.org/keywords/pruning","display_name":"Pruning","score":0.49549999833106995},{"id":"https://openalex.org/keywords/cache","display_name":"Cache","score":0.4837000072002411},{"id":"https://openalex.org/keywords/unification","display_name":"Unification","score":0.4593999981880188},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.45210000872612},{"id":"https://openalex.org/keywords/compiler","display_name":"Compiler","score":0.42809998989105225},{"id":"https://openalex.org/keywords/kernel","display_name":"Kernel (algebra)","score":0.3817000091075897},{"id":"https://openalex.org/keywords/quantization","display_name":"Quantization (signal processing)","score":0.35280001163482666},{"id":"https://openalex.org/keywords/rank","display_name":"Rank (graph theory)","score":0.3440999984741211}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8269000053405762},{"id":"https://openalex.org/C74912251","wikidata":"https://www.wikidata.org/wiki/Q6815727","display_name":"Memory footprint","level":2,"score":0.5425000190734863},{"id":"https://openalex.org/C108010975","wikidata":"https://www.wikidata.org/wiki/Q500094","display_name":"Pruning","level":2,"score":0.49549999833106995},{"id":"https://openalex.org/C115537543","wikidata":"https://www.wikidata.org/wiki/Q165596","display_name":"Cache","level":2,"score":0.4837000072002411},{"id":"https://openalex.org/C96146094","wikidata":"https://www.wikidata.org/wiki/Q609057","display_name":"Unification","level":2,"score":0.4593999981880188},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.45210000872612},{"id":"https://openalex.org/C169590947","wikidata":"https://www.wikidata.org/wiki/Q47506","display_name":"Compiler","level":2,"score":0.42809998989105225},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.42160001397132874},{"id":"https://openalex.org/C74193536","wikidata":"https://www.wikidata.org/wiki/Q574844","display_name":"Kernel (algebra)","level":2,"score":0.3817000091075897},{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.35280001163482666},{"id":"https://openalex.org/C164226766","wikidata":"https://www.wikidata.org/wiki/Q7293202","display_name":"Rank (graph theory)","level":2,"score":0.3440999984741211},{"id":"https://openalex.org/C43521106","wikidata":"https://www.wikidata.org/wiki/Q2165493","display_name":"Pipeline (software)","level":2,"score":0.34150001406669617},{"id":"https://openalex.org/C189783530","wikidata":"https://www.wikidata.org/wiki/Q352090","display_name":"CPU cache","level":3,"score":0.3328000009059906},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.32280001044273376},{"id":"https://openalex.org/C162319229","wikidata":"https://www.wikidata.org/wiki/Q175263","display_name":"Data structure","level":2,"score":0.32249999046325684},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.32190001010894775},{"id":"https://openalex.org/C100776233","wikidata":"https://www.wikidata.org/wiki/Q2532492","display_name":"Bridge (graph theory)","level":2,"score":0.31299999356269836},{"id":"https://openalex.org/C8797682","wikidata":"https://www.wikidata.org/wiki/Q2115","display_name":"XML","level":2,"score":0.3127000033855438},{"id":"https://openalex.org/C12186640","wikidata":"https://www.wikidata.org/wiki/Q6815743","display_name":"Memory model","level":3,"score":0.31209999322891235},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.30979999899864197},{"id":"https://openalex.org/C2780512708","wikidata":"https://www.wikidata.org/wiki/Q850661","display_name":"XQuery","level":4,"score":0.3075000047683716},{"id":"https://openalex.org/C46743427","wikidata":"https://www.wikidata.org/wiki/Q1341685","display_name":"Inference engine","level":3,"score":0.30090001225471497},{"id":"https://openalex.org/C2777212361","wikidata":"https://www.wikidata.org/wiki/Q5127848","display_name":"Class (philosophy)","level":2,"score":0.29409998655319214},{"id":"https://openalex.org/C67186912","wikidata":"https://www.wikidata.org/wiki/Q367664","display_name":"Data modeling","level":2,"score":0.28949999809265137},{"id":"https://openalex.org/C118643609","wikidata":"https://www.wikidata.org/wiki/Q189210","display_name":"Web application","level":2,"score":0.28690001368522644},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.28690001368522644},{"id":"https://openalex.org/C147497476","wikidata":"https://www.wikidata.org/wiki/Q54872","display_name":"RDF","level":3,"score":0.28690001368522644},{"id":"https://openalex.org/C206729178","wikidata":"https://www.wikidata.org/wiki/Q2271896","display_name":"Scheduling (production processes)","level":2,"score":0.2687000036239624},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.2646999955177307},{"id":"https://openalex.org/C45493050","wikidata":"https://www.wikidata.org/wiki/Q7884934","display_name":"Unified Model","level":2,"score":0.2635999917984009},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.26330000162124634},{"id":"https://openalex.org/C5655090","wikidata":"https://www.wikidata.org/wiki/Q192588","display_name":"Relational database","level":2,"score":0.2574999928474426},{"id":"https://openalex.org/C162307627","wikidata":"https://www.wikidata.org/wiki/Q204833","display_name":"Enhanced Data Rates for GSM Evolution","level":2,"score":0.25690001249313354}],"mesh":[],"locations_count":1,"locations":[{"id":"pmh:oai:repository.hkust.edu.hk:1783.1-173943","is_oa":false,"landing_page_url":"http://repository.hkust.edu.hk/ir/Record/1783.1-173943","pdf_url":null,"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":"Article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Large":[0],"language":[1],"models":[2,105],"are":[3],"increasingly":[4],"deployed":[5],"on":[6,103,208],"device":[7,209],"and":[8,17,51,59,75,80,98,100,122,135,148,155,167,175,177,179,186,206,210],"at":[9],"the":[10,26,35,64,84,137,217],"edge,":[11],"where":[12,55],"memory":[13,121],"capacity,":[14],"bandwidth,":[15],"latency,":[16],"privacy":[18],"requirements":[19],"dominate":[20],"system":[21,65],"behavior.":[22],"This":[23],"survey":[24],"systematizes":[25],"end":[27],"side":[28],"stack":[29],"from":[30,106,200],"algorithms":[31],"to":[32,71,108,112,124,203],"systems.":[33],"On":[34,63],"model":[36],"side,":[37,66],"we":[38,67,151],"present":[39],"a":[40,90,141,158,197,214],"clear":[41],"taxonomy":[42],"of":[43,83],"quantization,":[44],"pruning,":[45],"knowledge":[46],"distillation,":[47,176],"low":[48,132,160,187],"rank":[49,133,188],"adaptation,":[50],"hybrid":[52],"pipelines,":[53],"explaining":[54],"representative":[56,104],"methods":[57],"belong":[58],"how":[60],"they":[61],"compose.":[62],"link":[68],"these":[69],"techniques":[70],"inference":[72,191],"frameworks,":[73],"compiler":[74],"runtime":[76],"optimizations,":[77],"kernel":[78,168],"fusion,":[79,169],"explicit":[81],"management":[82],"KV":[85,138],"cache.":[86],"We":[87],"further":[88],"propose":[89],"unified":[91,159],"ALEM":[92],"protocol,":[93],"namely":[94],"Accuracy,":[95],"Latency,":[96],"Energy,":[97],"Memory,":[99],"instantiate":[101],"it":[102],"1":[107],"4":[109],"billion":[110],"parameters":[111,189],"reveal":[113],"practical":[114,198],"trade":[115],"offs:":[116],"apply":[117],"quantization":[118],"first":[119,125,142],"for":[120,220],"time":[123],"token,":[126],"pair":[127],"structured":[128,173],"pruning":[129,174],"with":[130],"mergeable":[131],"compensation,":[134],"treat":[136],"cache":[139],"as":[140],"class":[143],"subsystem":[144],"through":[145],"paging,":[146],"compression,":[147],"eviction.":[149],"Finally,":[150],"outline":[152],"open":[153],"problems":[154],"directions,":[156],"including":[157],"bit":[161],"pipeline":[162],"that":[163,182],"couples":[164],"transform,":[165],"calibration,":[166],"joint":[170],"search":[171],"over":[172],"train":[178],"serve":[180],"unification":[181],"collapses":[183],"sparse,":[184],"quantized,":[185],"into":[190],"ready":[192],"weights.":[193],"The":[194],"goal":[195],"is":[196],"bridge":[199],"algorithmic":[201],"compression":[202],"resource":[204],"aligned":[205],"reliable":[207],"edge":[211],"deployment.":[212],"https://github.com/LumosJiang/Awesome-On-Device-LLMs:":[213],"repository":[215],"hosting":[216],"complete":[218],"references":[219],"Sects.":[221],"3\u20134.":[222]},"counts_by_year":[],"updated_date":"2026-09-02T18:50:13.731487","created_date":"2026-09-02T00:00:00"}
