{"id":"https://openalex.org/W4394007410","doi":"https://doi.org/10.1145/3642970.3655832","title":"Towards Pareto Optimal Throughput in Small Language Model Serving","display_name":"Towards Pareto Optimal Throughput in Small Language Model Serving","publication_year":2024,"publication_date":"2024-04-19","ids":{"openalex":"https://openalex.org/W4394007410","doi":"https://doi.org/10.1145/3642970.3655832"},"language":"en","primary_location":{"id":"doi:10.1145/3642970.3655832","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3642970.3655832","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 4th Workshop on Machine Learning and Systems","raw_type":"proceedings-article"},"type":"preprint","indexed_in":["arxiv","crossref"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2404.03353","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5026630988","display_name":"Pol G. Recasens","orcid":"https://orcid.org/0009-0008-5535-0012"},"institutions":[{"id":"https://openalex.org/I2799803557","display_name":"Barcelona Supercomputing Center","ror":"https://ror.org/05sd8tv96","country_code":"ES","type":"facility","lineage":["https://openalex.org/I2799803557","https://openalex.org/I9617848"]},{"id":"https://openalex.org/I9617848","display_name":"Universitat Polit\u00e8cnica de Catalunya","ror":"https://ror.org/03mb6wj31","country_code":"ES","type":"education","lineage":["https://openalex.org/I9617848"]}],"countries":["ES"],"is_corresponding":true,"raw_author_name":"Pol G. Recasens","raw_affiliation_strings":["Barcelona Supercomputing Center"],"raw_orcid":"https://orcid.org/0009-0008-5535-0012","affiliations":[{"raw_affiliation_string":"Barcelona Supercomputing Center","institution_ids":["https://openalex.org/I9617848","https://openalex.org/I2799803557"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5004237513","display_name":"Yue Zhu","orcid":"https://orcid.org/0009-0005-2207-1904"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yue Zhu","raw_affiliation_strings":["IBM Research"],"raw_orcid":"https://orcid.org/0009-0005-2207-1904","affiliations":[{"raw_affiliation_string":"IBM Research","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100337643","display_name":"Chen Wang","orcid":"https://orcid.org/0000-0003-0204-2362"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen Wang","raw_affiliation_strings":["IBM Research"],"raw_orcid":"https://orcid.org/0000-0003-0204-2362","affiliations":[{"raw_affiliation_string":"IBM Research","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5023638218","display_name":"Eun Kyung Lee","orcid":"https://orcid.org/0000-0002-2335-4798"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Eun Kyung Lee","raw_affiliation_strings":["IBM Research"],"raw_orcid":"https://orcid.org/0000-0002-2335-4798","affiliations":[{"raw_affiliation_string":"IBM Research","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5086114919","display_name":"Olivier Tardieu","orcid":"https://orcid.org/0000-0002-8377-6757"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Olivier Tardieu","raw_affiliation_strings":["IBM Research"],"raw_orcid":"https://orcid.org/0000-0002-8377-6757","affiliations":[{"raw_affiliation_string":"IBM Research","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5027088517","display_name":"Alaa Youssef","orcid":"https://orcid.org/0000-0001-5346-7331"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Alaa Youssef","raw_affiliation_strings":["IBM Research"],"raw_orcid":"https://orcid.org/0000-0001-5346-7331","affiliations":[{"raw_affiliation_string":"IBM Research","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5039375454","display_name":"Jordi Torres","orcid":"https://orcid.org/0000-0003-1963-7418"},"institutions":[{"id":"https://openalex.org/I2799803557","display_name":"Barcelona Supercomputing Center","ror":"https://ror.org/05sd8tv96","country_code":"ES","type":"facility","lineage":["https://openalex.org/I2799803557","https://openalex.org/I9617848"]},{"id":"https://openalex.org/I9617848","display_name":"Universitat Polit\u00e8cnica de Catalunya","ror":"https://ror.org/03mb6wj31","country_code":"ES","type":"education","lineage":["https://openalex.org/I9617848"]}],"countries":["ES"],"is_corresponding":false,"raw_author_name":"Jordi Torres","raw_affiliation_strings":["Barcelona Supercomputing Center, Universitat Polit\u00e8cnica de Catalunya"],"raw_orcid":"https://orcid.org/0000-0003-1963-7418","affiliations":[{"raw_affiliation_string":"Barcelona Supercomputing Center, Universitat Polit\u00e8cnica de Catalunya","institution_ids":["https://openalex.org/I2799803557","https://openalex.org/I9617848"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5074281010","display_name":"Josep Ll. Berral","orcid":"https://orcid.org/0000-0003-3037-3580"},"institutions":[{"id":"https://openalex.org/I2799803557","display_name":"Barcelona Supercomputing Center","ror":"https://ror.org/05sd8tv96","country_code":"ES","type":"facility","lineage":["https://openalex.org/I2799803557","https://openalex.org/I9617848"]},{"id":"https://openalex.org/I9617848","display_name":"Universitat Polit\u00e8cnica de Catalunya","ror":"https://ror.org/03mb6wj31","country_code":"ES","type":"education","lineage":["https://openalex.org/I9617848"]}],"countries":["ES"],"is_corresponding":false,"raw_author_name":"Josep Ll. Berral","raw_affiliation_strings":["Universitat Polit\u00e8cnica de Catalunya, Barcelona Supercomputing Center"],"raw_orcid":"https://orcid.org/0000-0003-3037-3580","affiliations":[{"raw_affiliation_string":"Universitat Polit\u00e8cnica de Catalunya, Barcelona Supercomputing Center","institution_ids":["https://openalex.org/I2799803557","https://openalex.org/I9617848"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":8,"corresponding_author_ids":["https://openalex.org/A5026630988"],"corresponding_institution_ids":["https://openalex.org/I2799803557","https://openalex.org/I9617848"],"apc_list":null,"apc_paid":null,"fwci":3.106,"has_fulltext":true,"cited_by_count":10,"citation_normalized_percentile":{"value":0.9223541,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":90,"max":99},"biblio":{"volume":null,"issue":null,"first_page":"144","last_page":"152"},"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":0.9965000152587891,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.9869999885559082,"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/computer-science","display_name":"Computer science","score":0.7691764831542969},{"id":"https://openalex.org/keywords/throughput","display_name":"Throughput","score":0.6853702068328857},{"id":"https://openalex.org/keywords/pareto-principle","display_name":"Pareto principle","score":0.6138518452644348},{"id":"https://openalex.org/keywords/set","display_name":"Set (abstract data type)","score":0.5877289175987244},{"id":"https://openalex.org/keywords/benchmark","display_name":"Benchmark (surveying)","score":0.5740569233894348},{"id":"https://openalex.org/keywords/resource","display_name":"Resource (disambiguation)","score":0.5592646598815918},{"id":"https://openalex.org/keywords/enhanced-data-rates-for-gsm-evolution","display_name":"Enhanced Data Rates for GSM Evolution","score":0.4870753586292267},{"id":"https://openalex.org/keywords/footprint","display_name":"Footprint","score":0.4799135625362396},{"id":"https://openalex.org/keywords/resource-allocation","display_name":"Resource allocation","score":0.4760541617870331},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.45098716020584106},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.41579902172088623},{"id":"https://openalex.org/keywords/distributed-computing","display_name":"Distributed computing","score":0.35118764638900757},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.25146761536598206},{"id":"https://openalex.org/keywords/computer-network","display_name":"Computer network","score":0.20169419050216675},{"id":"https://openalex.org/keywords/engineering","display_name":"Engineering","score":0.1098356544971466},{"id":"https://openalex.org/keywords/programming-language","display_name":"Programming language","score":0.10331571102142334},{"id":"https://openalex.org/keywords/telecommunications","display_name":"Telecommunications","score":0.10318261384963989},{"id":"https://openalex.org/keywords/operations-management","display_name":"Operations management","score":0.08312124013900757}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7691764831542969},{"id":"https://openalex.org/C157764524","wikidata":"https://www.wikidata.org/wiki/Q1383412","display_name":"Throughput","level":3,"score":0.6853702068328857},{"id":"https://openalex.org/C137635306","wikidata":"https://www.wikidata.org/wiki/Q182667","display_name":"Pareto principle","level":2,"score":0.6138518452644348},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.5877289175987244},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.5740569233894348},{"id":"https://openalex.org/C206345919","wikidata":"https://www.wikidata.org/wiki/Q20380951","display_name":"Resource (disambiguation)","level":2,"score":0.5592646598815918},{"id":"https://openalex.org/C162307627","wikidata":"https://www.wikidata.org/wiki/Q204833","display_name":"Enhanced Data Rates for GSM Evolution","level":2,"score":0.4870753586292267},{"id":"https://openalex.org/C132943942","wikidata":"https://www.wikidata.org/wiki/Q2562511","display_name":"Footprint","level":2,"score":0.4799135625362396},{"id":"https://openalex.org/C29202148","wikidata":"https://www.wikidata.org/wiki/Q287260","display_name":"Resource allocation","level":2,"score":0.4760541617870331},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.45098716020584106},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.41579902172088623},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.35118764638900757},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.25146761536598206},{"id":"https://openalex.org/C31258907","wikidata":"https://www.wikidata.org/wiki/Q1301371","display_name":"Computer network","level":1,"score":0.20169419050216675},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.1098356544971466},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.10331571102142334},{"id":"https://openalex.org/C76155785","wikidata":"https://www.wikidata.org/wiki/Q418","display_name":"Telecommunications","level":1,"score":0.10318261384963989},{"id":"https://openalex.org/C21547014","wikidata":"https://www.wikidata.org/wiki/Q1423657","display_name":"Operations management","level":1,"score":0.08312124013900757},{"id":"https://openalex.org/C86803240","wikidata":"https://www.wikidata.org/wiki/Q420","display_name":"Biology","level":0,"score":0.0},{"id":"https://openalex.org/C205649164","wikidata":"https://www.wikidata.org/wiki/Q1071","display_name":"Geography","level":0,"score":0.0},{"id":"https://openalex.org/C555944384","wikidata":"https://www.wikidata.org/wiki/Q249","display_name":"Wireless","level":2,"score":0.0},{"id":"https://openalex.org/C151730666","wikidata":"https://www.wikidata.org/wiki/Q7205","display_name":"Paleontology","level":1,"score":0.0},{"id":"https://openalex.org/C13280743","wikidata":"https://www.wikidata.org/wiki/Q131089","display_name":"Geodesy","level":1,"score":0.0}],"mesh":[],"locations_count":4,"locations":[{"id":"doi:10.1145/3642970.3655832","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3642970.3655832","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 4th Workshop on Machine Learning and Systems","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2404.03353","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2404.03353","pdf_url":"https://arxiv.org/pdf/2404.03353","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"pmh:oai:upcommons.upc.edu:2117/409824","is_oa":true,"landing_page_url":"https://hdl.handle.net/2117/409824","pdf_url":"https://upcommons.upc.edu/bitstreams/5a874b1d-487b-4bcc-9fbf-7ab30f515c37/download","source":{"id":"https://openalex.org/S4377196262","display_name":"UPCommons institutional repository (Universitat Polit\u00e8cnica de Catalunya)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I9617848","host_organization_name":"Universitat Polit\u00e8cnica de Catalunya","host_organization_lineage":["https://openalex.org/I9617848"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by-nc-nd","license_id":"https://openalex.org/licenses/cc-by-nc-nd","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"info:eu-repo/semantics/conferenceObject"},{"id":"pmh:oai:upcommons.upc.edu:2117/428295","is_oa":true,"landing_page_url":"https://hdl.handle.net/2117/428295","pdf_url":"https://upcommons.upc.edu/bitstreams/31bf3538-b1df-44e8-b034-74700fcfb16b/download","source":{"id":"https://openalex.org/S4377196262","display_name":"UPCommons institutional repository (Universitat Polit\u00e8cnica de Catalunya)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I9617848","host_organization_name":"Universitat Polit\u00e8cnica de Catalunya","host_organization_lineage":["https://openalex.org/I9617848"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by-nc-nd","license_id":"https://openalex.org/licenses/cc-by-nc-nd","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Conference report"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2404.03353","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2404.03353","pdf_url":"https://arxiv.org/pdf/2404.03353","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G1321411063","display_name":null,"funder_award_id":"2021-SGR-00478","funder_id":"https://openalex.org/F4320334830","funder_display_name":"Ag\u00e8ncia de Gesti\u00f3 d'Ajuts Universitaris i de Recerca"},{"id":"https://openalex.org/G2262748287","display_name":null,"funder_award_id":"501100011033","funder_id":"https://openalex.org/F4320335598","funder_display_name":"Agencia Estatal de Investigaci\u00f3n"},{"id":"https://openalex.org/G2309174360","display_name":null,"funder_award_id":"10.13039","funder_id":"https://openalex.org/F4320322930","funder_display_name":"Ministerio de Ciencia e Innovaci\u00f3n"},{"id":"https://openalex.org/G2778551575","display_name":null,"funder_award_id":"2021-SGR-00478","funder_id":"https://openalex.org/F4320321505","funder_display_name":"Generalitat de Catalunya"},{"id":"https://openalex.org/G3005324705","display_name":null,"funder_award_id":"501100011033","funder_id":"https://openalex.org/F4320321505","funder_display_name":"Generalitat de Catalunya"},{"id":"https://openalex.org/G3480869486","display_name":null,"funder_award_id":"13039","funder_id":"https://openalex.org/F4320335598","funder_display_name":"Agencia Estatal de Investigaci\u00f3n"},{"id":"https://openalex.org/G3815932828","display_name":null,"funder_award_id":"10.13039/ 501100011033","funder_id":"https://openalex.org/F4320335322","funder_display_name":"European Regional Development Fund"},{"id":"https://openalex.org/G5080475149","display_name":null,"funder_award_id":"10.13039","funder_id":"https://openalex.org/F4320335598","funder_display_name":"Agencia Estatal de Investigaci\u00f3n"},{"id":"https://openalex.org/G5421974903","display_name":null,"funder_award_id":"501100011033","funder_id":"https://openalex.org/F4320322930","funder_display_name":"Ministerio de Ciencia e Innovaci\u00f3n"},{"id":"https://openalex.org/G5967599077","display_name":null,"funder_award_id":"501100011033","funder_id":"https://openalex.org/F4320335322","funder_display_name":"European Regional Development Fund"},{"id":"https://openalex.org/G6272053785","display_name":null,"funder_award_id":"MCIN/AEI/ 10.13039/ 501100011033","funder_id":"https://openalex.org/F4320322930","funder_display_name":"Ministerio de Ciencia e Innovaci\u00f3n"},{"id":"https://openalex.org/G7040512654","display_name":null,"funder_award_id":"10.13039/ 501100011033","funder_id":"https://openalex.org/F4320335598","funder_display_name":"Agencia Estatal de Investigaci\u00f3n"}],"funders":[{"id":"https://openalex.org/F4320321505","display_name":"Generalitat de Catalunya","ror":"https://ror.org/01bg62x04"},{"id":"https://openalex.org/F4320322930","display_name":"Ministerio de Ciencia e Innovaci\u00f3n","ror":"https://ror.org/034900433"},{"id":"https://openalex.org/F4320334830","display_name":"Ag\u00e8ncia de Gesti\u00f3 d'Ajuts Universitaris i de Recerca","ror":"https://ror.org/01n4pqe45"},{"id":"https://openalex.org/F4320335322","display_name":"European Regional Development Fund","ror":"https://ror.org/00k4n6c32"},{"id":"https://openalex.org/F4320335598","display_name":"Agencia Estatal de Investigaci\u00f3n","ror":null}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4394007410.pdf","grobid_xml":"https://content.openalex.org/works/W4394007410.grobid-xml"},"referenced_works_count":7,"referenced_works":["https://openalex.org/W2979826702","https://openalex.org/W3159727696","https://openalex.org/W4224308101","https://openalex.org/W4321636575","https://openalex.org/W4387321091","https://openalex.org/W4390189088","https://openalex.org/W6810081322"],"related_works":["https://openalex.org/W2368534554","https://openalex.org/W2148469257","https://openalex.org/W2999244305","https://openalex.org/W2330537534","https://openalex.org/W2136583354","https://openalex.org/W2111238207","https://openalex.org/W2168827721","https://openalex.org/W2031194604","https://openalex.org/W4249517016","https://openalex.org/W2094506573"],"abstract_inverted_index":{"Large":[0],"language":[1,12],"models":[2,43],"(LLMs)":[3],"have":[4],"revolutionized":[5],"the":[6,23,76,85,89],"state-of-the-art":[7],"of":[8,25,54,80,92,104],"many":[9],"different":[10],"natural":[11],"processing":[13],"tasks.":[14],"Although":[15],"serving":[16,116],"LLMs":[17],"is":[18],"computationally":[19],"and":[20,63],"memory":[21,78],"demanding,":[22],"rise":[24],"Small":[26],"Language":[27],"Models":[28],"(SLMs)":[29],"offers":[30],"new":[31,70],"opportunities":[32],"for":[33,83,115],"resource-constrained":[34],"users,":[35],"who":[36],"now":[37],"are":[38],"able":[39],"to":[40,57],"serve":[41],"small":[42,77],"with":[44],"cutting-edge":[45],"performance.":[46],"In":[47,96],"this":[48,97],"paper,":[49],"we":[50,99],"present":[51,100],"a":[52,69,93],"set":[53,103],"experiments":[55],"designed":[56],"benchmark":[58],"SLM":[59],"inference":[60],"at":[61],"performance":[62],"energy":[64],"levels.":[65],"Our":[66],"analysis":[67],"provides":[68],"perspective":[71],"in":[72],"serving,":[73],"highlighting":[74],"that":[75],"footprint":[79],"SLMs":[81],"allows":[82],"reaching":[84],"Pareto-optimal":[86],"throughput":[87],"within":[88],"resource":[90,113],"capacity":[91],"single":[94],"accelerator.":[95],"regard,":[98],"an":[101],"initial":[102],"findings":[105],"demonstrating":[106],"how":[107],"model":[108],"replication":[109],"can":[110],"effectively":[111],"improve":[112],"utilization":[114],"SLMs.":[117]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2025,"cited_by_count":8},{"year":2024,"cited_by_count":1}],"updated_date":"2026-06-02T09:04:35.204637","created_date":"2025-10-10T00:00:00"}