{"id":"https://openalex.org/W7010087891","doi":null,"title":"Full Stack Approach for Efficient Deep Learning Inference","display_name":"Full Stack Approach for Efficient Deep Learning Inference","publication_year":2024,"publication_date":"2024-01-01","ids":{"openalex":"https://openalex.org/W7010087891"},"language":"en","primary_location":{"id":"pmh:oai:escholarship.org:ark:/13030/qt4wf834q8","doi":null,"is_oa":true,"landing_page_url":"https://escholarship.org/uc/item/4wf834q8","pdf_url":"https://escholarship.org/content/qt4wf834q8/qt4wf834q8.pdf","source":{"id":"https://openalex.org/S4306400115","display_name":"eScholarship (California Digital Library)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I2801248553","host_organization_name":"California Digital Library","host_organization_lineage":["https://openalex.org/I2801248553"],"host_organization_lineage_names":[],"type":"repository"},"license":"public-domain","license_id":"https://openalex.org/licenses/public-domain","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"etd"},"volume":null,"type":"dissertation","is_traditional":true,"indexed_in":[],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://escholarship.org/content/qt4wf834q8/qt4wf834q8.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5002255210","display_name":"Sehoon Kim","orcid":"https://orcid.org/0000-0002-9339-5480","observed_orcids":["https://orcid.org/0000-0002-9339-5480"]},"institutions":[{"id":"https://openalex.org/I95457486","display_name":"University of California, Berkeley","ror":"https://ror.org/01an7q238","country_code":"US","type":"education","lineage":["https://openalex.org/I95457486"]}],"countries":["US"],"is_corresponding":true,"raw_author_name":"Kim, Sehoon","raw_affiliation_strings":["Doctor of Philosophy in Computer Science University of California , Berkeley"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Doctor of Philosophy in Computer Science University of California , Berkeley","institution_ids":["https://openalex.org/I95457486"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":["https://openalex.org/A5002255210"],"corresponding_institution_ids":["https://openalex.org/I95457486"],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"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":true,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.9682999849319458,"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"}},"topics":[{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.9682999849319458,"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/T10028","display_name":"Topic Modeling","score":0.013500000350177288,"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":0.003800000064074993,"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/large-language-model-inference","display_name":"large language model inference","score":0.7179999947547913},{"id":"https://openalex.org/keywords/inference-efficiency","display_name":"inference efficiency","score":0.6819999814033508},{"id":"https://openalex.org/keywords/speculative-decoding","display_name":"speculative decoding","score":0.953000009059906},{"id":"https://openalex.org/keywords/speech-recognition","display_name":"speech recognition","score":0.6850000023841858},{"id":"https://openalex.org/keywords/memory-reduction","display_name":"memory reduction","score":0.48500001430511475},{"id":"https://openalex.org/keywords/latency-reduction","display_name":"latency reduction","score":0.781000018119812}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8375999927520752},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.8066999912261963},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5468999743461609},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.5175999999046326},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.5095000267028809},{"id":"https://openalex.org/C68339613","wikidata":"https://www.wikidata.org/wiki/Q1549489","display_name":"Speedup","level":2,"score":0.45239999890327454},{"id":"https://openalex.org/C105339364","wikidata":"https://www.wikidata.org/wiki/Q2297740","display_name":"Software deployment","level":2,"score":0.447299987077713},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.44519999623298645},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.3833000063896179},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.3700999915599823},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.352400004863739},{"id":"https://openalex.org/C186108316","wikidata":"https://www.wikidata.org/wiki/Q352530","display_name":"Adaptive neuro fuzzy inference system","level":4,"score":0.3400000035762787},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.3215000033378601},{"id":"https://openalex.org/C2777472644","wikidata":"https://www.wikidata.org/wiki/Q16968992","display_name":"Approximate inference","level":3,"score":0.31769999861717224},{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.3111000061035156},{"id":"https://openalex.org/C157170001","wikidata":"https://www.wikidata.org/wiki/Q4781507","display_name":"Applications of artificial intelligence","level":2,"score":0.30000001192092896},{"id":"https://openalex.org/C12186640","wikidata":"https://www.wikidata.org/wiki/Q6815743","display_name":"Memory model","level":3,"score":0.29440000653266907},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.28679999709129333},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.2809000015258789},{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.2727000117301941},{"id":"https://openalex.org/C2781357197","wikidata":"https://www.wikidata.org/wiki/Q5757597","display_name":"High memory","level":2,"score":0.2694999873638153},{"id":"https://openalex.org/C159877910","wikidata":"https://www.wikidata.org/wiki/Q2202883","display_name":"Autoregressive model","level":2,"score":0.25529998540878296},{"id":"https://openalex.org/C195672273","wikidata":"https://www.wikidata.org/wiki/Q6888132","display_name":"Model-based design","level":2,"score":0.2538999915122986}],"mesh":[],"locations_count":1,"locations":[{"id":"pmh:oai:escholarship.org:ark:/13030/qt4wf834q8","doi":null,"is_oa":true,"landing_page_url":"https://escholarship.org/uc/item/4wf834q8","pdf_url":"https://escholarship.org/content/qt4wf834q8/qt4wf834q8.pdf","source":{"id":"https://openalex.org/S4306400115","display_name":"eScholarship (California Digital Library)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I2801248553","host_organization_name":"California Digital Library","host_organization_lineage":["https://openalex.org/I2801248553"],"host_organization_lineage_names":[],"type":"repository"},"license":"public-domain","license_id":"https://openalex.org/licenses/public-domain","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"etd"}],"best_oa_location":{"id":"pmh:oai:escholarship.org:ark:/13030/qt4wf834q8","doi":null,"is_oa":true,"landing_page_url":"https://escholarship.org/uc/item/4wf834q8","pdf_url":"https://escholarship.org/content/qt4wf834q8/qt4wf834q8.pdf","source":{"id":"https://openalex.org/S4306400115","display_name":"eScholarship (California Digital Library)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I2801248553","host_organization_name":"California Digital Library","host_organization_lineage":["https://openalex.org/I2801248553"],"host_organization_lineage_names":[],"type":"repository"},"license":"public-domain","license_id":"https://openalex.org/licenses/public-domain","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"etd"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7010087891.pdf","grobid_xml":"https://content.openalex.org/works/W7010087891.grobid-xml"},"referenced_works_count":184,"referenced_works":["https://openalex.org/W1549358575","https://openalex.org/W1690739335","https://openalex.org/W1828163288","https://openalex.org/W1836465849","https://openalex.org/W2131172946","https://openalex.org/W2319920447","https://openalex.org/W2405920868","https://openalex.org/W2469490737","https://openalex.org/W2515385951","https://openalex.org/W2525332836","https://openalex.org/W2531327146","https://openalex.org/W2607892599","https://openalex.org/W2612445135","https://openalex.org/W2707890836","https://openalex.org/W2739351760","https://openalex.org/W2767206889","https://openalex.org/W2769312802","https://openalex.org/W2784372305","https://openalex.org/W2786771851","https://openalex.org/W2787752464","https://openalex.org/W2789142637","https://openalex.org/W2799054028","https://openalex.org/W2809624076","https://openalex.org/W2866343820","https://openalex.org/W2885185669","https://openalex.org/W2896457183","https://openalex.org/W2899663614","https://openalex.org/W2903260438","https://openalex.org/W2915589364","https://openalex.org/W2916954108","https://openalex.org/W2924902521","https://openalex.org/W2936774411","https://openalex.org/W2937808806","https://openalex.org/W2944255943","https://openalex.org/W2945767825","https://openalex.org/W2946794439","https://openalex.org/W2946948417","https://openalex.org/W2947946877","https://openalex.org/W2948981900","https://openalex.org/W2952862139","https://openalex.org/W2952942550","https://openalex.org/W2962915948","https://openalex.org/W2963403664","https://openalex.org/W2963434219","https://openalex.org/W2963748441","https://openalex.org/W2965373594","https://openalex.org/W2969515962","https://openalex.org/W2973215447","https://openalex.org/W2973727699","https://openalex.org/W2975059944","https://openalex.org/W2975381464","https://openalex.org/W2975429091","https://openalex.org/W2976965654","https://openalex.org/W2978017171","https://openalex.org/W2979314664","https://openalex.org/W2981757109","https://openalex.org/W2981852735","https://openalex.org/W2986772082","https://openalex.org/W2997710335","https://openalex.org/W3005700362","https://openalex.org/W3006683367","https://openalex.org/W3006801027","https://openalex.org/W3006963874","https://openalex.org/W3017022649","https://openalex.org/W3024171804","https://openalex.org/W3030163527","https://openalex.org/W3033529678","https://openalex.org/W3034457371","https://openalex.org/W3034560159","https://openalex.org/W3035007693","https://openalex.org/W3035251378","https://openalex.org/W3036463250","https://openalex.org/W3037932933","https://openalex.org/W3040573126","https://openalex.org/W3086977883","https://openalex.org/W3093502935","https://openalex.org/W3093517588","https://openalex.org/W3093579165","https://openalex.org/W3094502228","https://openalex.org/W3095173472","https://openalex.org/W3095838132","https://openalex.org/W3097558625","https://openalex.org/W3097777922","https://openalex.org/W3098576111","https://openalex.org/W3101045333","https://openalex.org/W3101248447","https://openalex.org/W3105966348","https://openalex.org/W3108424616","https://openalex.org/W3115511229","https://openalex.org/W3118568258","https://openalex.org/W3122752795","https://openalex.org/W3125507956","https://openalex.org/W3141188098","https://openalex.org/W3141854550","https://openalex.org/W3147414526","https://openalex.org/W3152956381","https://openalex.org/W3166109569","https://openalex.org/W3173374050","https://openalex.org/W3177813494","https://openalex.org/W3185686823","https://openalex.org/W3195140297","https://openalex.org/W3200543719","https://openalex.org/W3202028501","https://openalex.org/W3209984917","https://openalex.org/W4221145109","https://openalex.org/W4221161695","https://openalex.org/W4221165032","https://openalex.org/W4221167110","https://openalex.org/W4224308101","https://openalex.org/W4225887996","https://openalex.org/W4226399820","https://openalex.org/W4229005866","https://openalex.org/W4281690148","https://openalex.org/W4287254789","https://openalex.org/W4293025109","https://openalex.org/W4304195432","https://openalex.org/W4307934016","https://openalex.org/W4308760184","https://openalex.org/W4308760226","https://openalex.org/W4309591663","https://openalex.org/W4309953112","https://openalex.org/W4310561894","https://openalex.org/W4312769131","https://openalex.org/W4313484599","https://openalex.org/W4319166707","https://openalex.org/W4322718191","https://openalex.org/W4322718253","https://openalex.org/W4327810158","https://openalex.org/W4362598949","https://openalex.org/W4378942217","https://openalex.org/W4379260375","https://openalex.org/W4379548477","https://openalex.org/W4380552023","https://openalex.org/W4384918448","https://openalex.org/W4385436532","https://openalex.org/W4385474529","https://openalex.org/W4385570088","https://openalex.org/W4385570209","https://openalex.org/W4385571189","https://openalex.org/W4385572634","https://openalex.org/W4385573982","https://openalex.org/W4385681800","https://openalex.org/W4386231419","https://openalex.org/W4386528753","https://openalex.org/W4386841261","https://openalex.org/W4387164935","https://openalex.org/W4387560733","https://openalex.org/W4387636003","https://openalex.org/W4388626886","https://openalex.org/W4389326242","https://openalex.org/W4389422760","https://openalex.org/W4389500120","https://openalex.org/W4389520103","https://openalex.org/W4389520747","https://openalex.org/W4389520750","https://openalex.org/W4389524393","https://openalex.org/W4390833798","https://openalex.org/W4390872747","https://openalex.org/W4391833450","https://openalex.org/W4392679398","https://openalex.org/W4393160302","https://openalex.org/W4394670654","https://openalex.org/W4394698525","https://openalex.org/W4399317989","https://openalex.org/W4411630358","https://openalex.org/W4412042427","https://openalex.org/W4415795271","https://openalex.org/W4415798413","https://openalex.org/W4415799217","https://openalex.org/W6966825443","https://openalex.org/W7133186778","https://openalex.org/W7133202489","https://openalex.org/W7133204437","https://openalex.org/W7133207436","https://openalex.org/W7133208400","https://openalex.org/W7133209886","https://openalex.org/W7133214447","https://openalex.org/W7133219248","https://openalex.org/W7133222210","https://openalex.org/W7133225605","https://openalex.org/W7133227044","https://openalex.org/W7133231030","https://openalex.org/W7133231369","https://openalex.org/W7133236450"],"related_works":["https://openalex.org/W7079662407","https://openalex.org/W7213516412","https://openalex.org/W4414315357","https://openalex.org/W7204963055","https://openalex.org/W7166855115","https://openalex.org/W4416257191","https://openalex.org/W4410887251","https://openalex.org/W4403678670","https://openalex.org/W7133235396","https://openalex.org/W7138928919"],"abstract_inverted_index":{"Recent":[0],"advancements":[1],"in":[2,10,29,213],"AI":[3,74,243,261],"technologies":[4],"have":[5,25],"led":[6],"to":[7,92,109,162,249,252],"unprecedented":[8],"growth":[9],"model":[11,77,81,86,173,244],"sizes,":[12],"particularly":[13],"with":[14],"the":[15,73,118,147,199,254],"advent":[16],"of":[17,72,117,259],"large":[18,170],"language":[19],"models":[20,24],"(LLMs).":[21],"While":[22],"these":[23,59,235],"shown":[26],"great":[27],"capabilities":[28],"various":[30],"domains,":[31],"their":[32],"exponential":[33],"scaling":[34],"has":[35],"introduced":[36],"significant":[37],"inference-time":[38,94],"overheads,":[39],"such":[40],"as":[41],"increased":[42],"memory":[43,97,133],"requirements,":[44],"latency,":[45],"and":[46,53,83,96,114,169,220,257],"computational":[47],"costs,":[48],"thereby":[49],"making":[50],"efficient":[51,178,255],"deployment":[52,116,256],"serving":[54,258],"challenging.":[55],"This":[56],"thesis":[57],"addresses":[58],"challenges":[60],"through":[61,164],"a":[62,110,151,165,184,205,238],"full-stack":[63,239],"approach":[64],"that":[65,155],"enhances":[66],"efficiency":[67,191],"across":[68],"four":[69],"key":[70],"components":[71],"inference":[75,79,143,159,190,245],"stack:":[76],"optimization,":[78,87],"methods,":[80,144],"architectures,":[82,174],"applications.":[84],"For":[85,141],"we":[88,145,175,202],"introduce":[89,203],"quantization":[90],"techniques":[91],"optimize":[93],"compute":[95,101],"requirements.":[98],"I-BERT":[99],"optimizes":[100],"by":[102,160,192,225],"leveraging":[103],"integer-only":[104,122],"quantization,":[105,130],"which":[106,125,188,216],"achieves":[107],"up":[108,161],"3.5\u00d7":[111],"latency":[112,219],"speedup":[113],"enables":[115],"Transformer":[119],"architectures":[120],"on":[121],"hardware.":[123],"SqueezeLLM,":[124],"employs":[126],"extremely":[127],"low-bit":[128],"weight":[129],"effectively":[131],"reduces":[132,217],"requirements":[134],"without":[135],"sacrificing":[136],"accuracy":[137],"during":[138],"LLM":[139,158],"inference.":[140],"enhanced":[142],"present":[146],"Big":[148],"Little":[149],"Decoder,":[150],"speculative":[152],"decoding":[153],"framework":[154,206],"accelerates":[156],"autoregressive":[157],"2x":[163],"collaboration":[166],"between":[167],"small":[168],"models.":[171],"Regarding":[172],"propose":[176],"an":[177],"design":[179],"for":[180,207,241],"speech":[181],"recognition":[182],"using":[183],"Temporal":[185],"U-Net":[186],"structure,":[187],"improves":[189],"shortening":[193],"input":[194],"sequence":[195],"lengths.":[196],"Finally,":[197],"at":[198],"application":[200],"level,":[201],"LLMCompiler,":[204],"efficiently":[208],"orchestrating":[209],"multiple":[210],"function":[211],"calls":[212],"LLM-based":[214],"applications,":[215],"execution":[218],"costs":[221],"while":[222],"enhancing":[223],"robustness":[224],"decomposing":[226],"complex":[227],"user":[228],"inputs":[229],"into":[230],"smaller,":[231],"easier":[232],"tasks.":[233],"Collectively,":[234],"contributions":[236],"provide":[237],"strategy":[240],"optimizing":[242],"from":[246],"low-level":[247],"systems":[248],"high-level":[250],"applications":[251],"enable":[253],"state-of-the-art":[260],"solutions.":[262]},"counts_by_year":[],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2025-10-10T00:00:00"}