{"id":"https://openalex.org/W4206792915","doi":"https://doi.org/10.1109/tpds.2021.3138862","title":"NeoFlow: A Flexible Framework for Enabling Efficient Compilation for High Performance DNN Training","display_name":"NeoFlow: A Flexible Framework for Enabling Efficient Compilation for High Performance DNN Training","publication_year":2021,"publication_date":"2021-01-01","ids":{"openalex":"https://openalex.org/W4206792915","doi":"https://doi.org/10.1109/tpds.2021.3138862"},"language":"en","primary_location":{"id":"doi:10.1109/tpds.2021.3138862","is_oa":true,"landing_page_url":"https://doi.org/10.1109/tpds.2021.3138862","pdf_url":"https://ieeexplore.ieee.org/ielx7/71/4359390/09664259.pdf","source":{"id":"https://openalex.org/S97130795","display_name":"IEEE Transactions on Parallel and Distributed Systems","issn_l":"1045-9219","issn":["1045-9219","1558-2183","2161-9883"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Transactions on Parallel and Distributed Systems","raw_type":"journal-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"hybrid","oa_url":"https://ieeexplore.ieee.org/ielx7/71/4359390/09664259.pdf","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5072186367","display_name":"Size Zheng","orcid":"https://orcid.org/0000-0002-9471-1780"},"institutions":[{"id":"https://openalex.org/I20231570","display_name":"Peking University","ror":"https://ror.org/02v51f717","country_code":"CN","type":"education","lineage":["https://openalex.org/I20231570"]}],"countries":["CN"],"is_corresponding":true,"raw_author_name":"Size Zheng","raw_affiliation_strings":["School of EECS, Peking University, 12465 Beijing, Beijing, China, (e-mail: zhengsz@pku.edu.cn)","Department of Computer Sci-ence, School of EECS, Peking University, Beijing, China"],"affiliations":[{"raw_affiliation_string":"School of EECS, Peking University, 12465 Beijing, Beijing, China, (e-mail: zhengsz@pku.edu.cn)","institution_ids":["https://openalex.org/I20231570"]},{"raw_affiliation_string":"Department of Computer Sci-ence, School of EECS, Peking University, Beijing, China","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5004459947","display_name":"Renze Chen","orcid":"https://orcid.org/0000-0001-5938-7965"},"institutions":[{"id":"https://openalex.org/I20231570","display_name":"Peking University","ror":"https://ror.org/02v51f717","country_code":"CN","type":"education","lineage":["https://openalex.org/I20231570"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Renze Chen","raw_affiliation_strings":["School of EECS, Peking University, 12465 Beijing, Beijing, China, (e-mail: crz@pku.edu.cn)","Department of Computer Sci-ence, School of EECS, Peking University, Beijing, China"],"affiliations":[{"raw_affiliation_string":"School of EECS, Peking University, 12465 Beijing, Beijing, China, (e-mail: crz@pku.edu.cn)","institution_ids":["https://openalex.org/I20231570"]},{"raw_affiliation_string":"Department of Computer Sci-ence, School of EECS, Peking University, Beijing, China","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102997450","display_name":"Yicheng Jin","orcid":"https://orcid.org/0000-0001-6966-1219"},"institutions":[{"id":"https://openalex.org/I20231570","display_name":"Peking University","ror":"https://ror.org/02v51f717","country_code":"CN","type":"education","lineage":["https://openalex.org/I20231570"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yicheng Jin","raw_affiliation_strings":["School of EECS, Peking University, 12465 Beijing, Beijing, China, (e-mail: yicheng.jin@pku.edu.cn)","Department of Computer Sci-ence, School of EECS, Peking University, Beijing, China"],"affiliations":[{"raw_affiliation_string":"School of EECS, Peking University, 12465 Beijing, Beijing, China, (e-mail: yicheng.jin@pku.edu.cn)","institution_ids":["https://openalex.org/I20231570"]},{"raw_affiliation_string":"Department of Computer Sci-ence, School of EECS, Peking University, Beijing, China","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5017132290","display_name":"Anjiang Wei","orcid":"https://orcid.org/0000-0003-1654-6027"},"institutions":[{"id":"https://openalex.org/I20231570","display_name":"Peking University","ror":"https://ror.org/02v51f717","country_code":"CN","type":"education","lineage":["https://openalex.org/I20231570"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Anjiang Wei","raw_affiliation_strings":["School of EECS, Peking University, 12465 Beijing, Beijing, China, (e-mail: weianjiang@pku.edu.cn)","Department of Computer Sci-ence, School of EECS, Peking University, Beijing, China"],"affiliations":[{"raw_affiliation_string":"School of EECS, Peking University, 12465 Beijing, Beijing, China, (e-mail: weianjiang@pku.edu.cn)","institution_ids":["https://openalex.org/I20231570"]},{"raw_affiliation_string":"Department of Computer Sci-ence, School of EECS, Peking University, Beijing, China","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100660218","display_name":"Bingyang Wu","orcid":"https://orcid.org/0000-0001-7221-8007"},"institutions":[{"id":"https://openalex.org/I20231570","display_name":"Peking University","ror":"https://ror.org/02v51f717","country_code":"CN","type":"education","lineage":["https://openalex.org/I20231570"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Bingyang Wu","raw_affiliation_strings":["School of EECS, Peking University, 12465 Beijing, Beijing, China, (e-mail: bingyangwu@pku.edu.cn)","Department of Computer Sci-ence, School of EECS, Peking University, Beijing, China"],"affiliations":[{"raw_affiliation_string":"School of EECS, Peking University, 12465 Beijing, Beijing, China, (e-mail: bingyangwu@pku.edu.cn)","institution_ids":["https://openalex.org/I20231570"]},{"raw_affiliation_string":"Department of Computer Sci-ence, School of EECS, Peking University, Beijing, China","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100663885","display_name":"Xiuhong Li","orcid":"https://orcid.org/0000-0002-5327-0907"},"institutions":[{"id":"https://openalex.org/I4210092218","display_name":"Sensimetrics Corporation","ror":"https://ror.org/00h7y2d26","country_code":"US","type":"company","lineage":["https://openalex.org/I4210092218"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Xiuhong Li","raw_affiliation_strings":["SenseTime Research & Shanghai AI Lab, SenseTime Research, Hong Kong, Hong Kong, China, (e-mail: lixiuhong@sensetime.com)","Department of Computer Sci-ence, School of EECS, Peking University, Beijing, China"],"affiliations":[{"raw_affiliation_string":"SenseTime Research & Shanghai AI Lab, SenseTime Research, Hong Kong, Hong Kong, China, (e-mail: lixiuhong@sensetime.com)","institution_ids":["https://openalex.org/I4210092218"]},{"raw_affiliation_string":"Department of Computer Sci-ence, School of EECS, Peking University, Beijing, China","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5049910854","display_name":"Shengen Yan","orcid":"https://orcid.org/0009-0005-3858-7972"},"institutions":[{"id":"https://openalex.org/I4210092218","display_name":"Sensimetrics Corporation","ror":"https://ror.org/00h7y2d26","country_code":"US","type":"company","lineage":["https://openalex.org/I4210092218"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Shengen Yan","raw_affiliation_strings":["SenseTime Research, SenseTime Research, Hong Kong, Hong Kong, China, (e-mail: yanshengen@sensetime.com)","Department of Computer Sci-ence, School of EECS, Peking University, Beijing, China"],"affiliations":[{"raw_affiliation_string":"SenseTime Research, SenseTime Research, Hong Kong, Hong Kong, China, (e-mail: yanshengen@sensetime.com)","institution_ids":["https://openalex.org/I4210092218"]},{"raw_affiliation_string":"Department of Computer Sci-ence, School of EECS, Peking University, Beijing, China","institution_ids":[]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5100604860","display_name":"Yun Liang","orcid":"https://orcid.org/0000-0002-9076-7998"},"institutions":[{"id":"https://openalex.org/I20231570","display_name":"Peking University","ror":"https://ror.org/02v51f717","country_code":"CN","type":"education","lineage":["https://openalex.org/I20231570"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yun Liang","raw_affiliation_strings":["School of EECS, Peking University, 12465 Beijing, Beijing, China, (e-mail: ericlyun@pku.edu.cn)","Department of Computer Sci-ence, School of EECS, Peking University, Beijing, China"],"affiliations":[{"raw_affiliation_string":"School of EECS, Peking University, 12465 Beijing, Beijing, China, (e-mail: ericlyun@pku.edu.cn)","institution_ids":["https://openalex.org/I20231570"]},{"raw_affiliation_string":"Department of Computer Sci-ence, School of EECS, Peking University, Beijing, China","institution_ids":[]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":8,"corresponding_author_ids":["https://openalex.org/A5072186367"],"corresponding_institution_ids":["https://openalex.org/I20231570"],"apc_list":null,"apc_paid":null,"fwci":0.8646,"has_fulltext":true,"cited_by_count":10,"citation_normalized_percentile":{"value":0.76359477,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":91,"max":97},"biblio":{"volume":null,"issue":null,"first_page":"1","last_page":"1"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.9998999834060669,"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":0.9998999834060669,"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/T12535","display_name":"Machine Learning and Data Classification","score":0.9921000003814697,"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/T11948","display_name":"Machine Learning in Materials Science","score":0.9828000068664551,"subfield":{"id":"https://openalex.org/subfields/2505","display_name":"Materials Chemistry"},"field":{"id":"https://openalex.org/fields/25","display_name":"Materials 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":0.8783042430877686},{"id":"https://openalex.org/keywords/implementation","display_name":"Implementation","score":0.835965633392334},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.5700088143348694},{"id":"https://openalex.org/keywords/computer-architecture","display_name":"Computer architecture","score":0.4432294964790344},{"id":"https://openalex.org/keywords/graph","display_name":"Graph","score":0.4408951997756958},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.40518394112586975},{"id":"https://openalex.org/keywords/computer-engineering","display_name":"Computer engineering","score":0.3435688316822052},{"id":"https://openalex.org/keywords/distributed-computing","display_name":"Distributed computing","score":0.3290855288505554},{"id":"https://openalex.org/keywords/machine-learning","display_name":"Machine learning","score":0.32818156480789185},{"id":"https://openalex.org/keywords/theoretical-computer-science","display_name":"Theoretical computer science","score":0.2774190902709961},{"id":"https://openalex.org/keywords/programming-language","display_name":"Programming language","score":0.2459089159965515}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8783042430877686},{"id":"https://openalex.org/C26713055","wikidata":"https://www.wikidata.org/wiki/Q245962","display_name":"Implementation","level":2,"score":0.835965633392334},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.5700088143348694},{"id":"https://openalex.org/C118524514","wikidata":"https://www.wikidata.org/wiki/Q173212","display_name":"Computer architecture","level":1,"score":0.4432294964790344},{"id":"https://openalex.org/C132525143","wikidata":"https://www.wikidata.org/wiki/Q141488","display_name":"Graph","level":2,"score":0.4408951997756958},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.40518394112586975},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.3435688316822052},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.3290855288505554},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.32818156480789185},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.2774190902709961},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.2459089159965515}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/tpds.2021.3138862","is_oa":true,"landing_page_url":"https://doi.org/10.1109/tpds.2021.3138862","pdf_url":"https://ieeexplore.ieee.org/ielx7/71/4359390/09664259.pdf","source":{"id":"https://openalex.org/S97130795","display_name":"IEEE Transactions on Parallel and Distributed Systems","issn_l":"1045-9219","issn":["1045-9219","1558-2183","2161-9883"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Transactions on Parallel and Distributed Systems","raw_type":"journal-article"}],"best_oa_location":{"id":"doi:10.1109/tpds.2021.3138862","is_oa":true,"landing_page_url":"https://doi.org/10.1109/tpds.2021.3138862","pdf_url":"https://ieeexplore.ieee.org/ielx7/71/4359390/09664259.pdf","source":{"id":"https://openalex.org/S97130795","display_name":"IEEE Transactions on Parallel and Distributed Systems","issn_l":"1045-9219","issn":["1045-9219","1558-2183","2161-9883"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Transactions on Parallel and Distributed Systems","raw_type":"journal-article"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/9","display_name":"Industry, innovation and infrastructure","score":0.47999998927116394}],"awards":[{"id":"https://openalex.org/G8806450724","display_name":null,"funder_award_id":"U21B2017","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"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4206792915.pdf","grobid_xml":"https://content.openalex.org/works/W4206792915.grobid-xml"},"referenced_works_count":73,"referenced_works":["https://openalex.org/W2966661","https://openalex.org/W1686810756","https://openalex.org/W2055312318","https://openalex.org/W2118776487","https://openalex.org/W2186615578","https://openalex.org/W2194775991","https://openalex.org/W2293634267","https://openalex.org/W2295598076","https://openalex.org/W2412412865","https://openalex.org/W2476548250","https://openalex.org/W2531409750","https://openalex.org/W2549139847","https://openalex.org/W2590246587","https://openalex.org/W2784561332","https://openalex.org/W2785994986","https://openalex.org/W2786320458","https://openalex.org/W2792643794","https://openalex.org/W2796649226","https://openalex.org/W2804032941","https://openalex.org/W2804500013","https://openalex.org/W2896457183","https://openalex.org/W2899971035","https://openalex.org/W2940126124","https://openalex.org/W2953384591","https://openalex.org/W2961619211","https://openalex.org/W2963125010","https://openalex.org/W2963125769","https://openalex.org/W2963573053","https://openalex.org/W2963703618","https://openalex.org/W2963840672","https://openalex.org/W2966258469","https://openalex.org/W2981758446","https://openalex.org/W3001542287","https://openalex.org/W3012249773","https://openalex.org/W3037749908","https://openalex.org/W3096395190","https://openalex.org/W3157104645","https://openalex.org/W3181900527","https://openalex.org/W4246166885","https://openalex.org/W4292779060","https://openalex.org/W4294408596","https://openalex.org/W4295312788","https://openalex.org/W4297775537","https://openalex.org/W4300749827","https://openalex.org/W4302296459","https://openalex.org/W6637151318","https://openalex.org/W6637373629","https://openalex.org/W6677548962","https://openalex.org/W6686509673","https://openalex.org/W6696085341","https://openalex.org/W6713134421","https://openalex.org/W6719700797","https://openalex.org/W6728184133","https://openalex.org/W6737664043","https://openalex.org/W6743446608","https://openalex.org/W6743596144","https://openalex.org/W6745744969","https://openalex.org/W6745914641","https://openalex.org/W6748053814","https://openalex.org/W6748088138","https://openalex.org/W6748324931","https://openalex.org/W6749029207","https://openalex.org/W6750448596","https://openalex.org/W6751349269","https://openalex.org/W6752057402","https://openalex.org/W6755207826","https://openalex.org/W6756195904","https://openalex.org/W6761772364","https://openalex.org/W6766978945","https://openalex.org/W6772040440","https://openalex.org/W6778883912","https://openalex.org/W6779728309","https://openalex.org/W6779965347"],"related_works":["https://openalex.org/W2120447654","https://openalex.org/W2977179488","https://openalex.org/W2144453115","https://openalex.org/W2128223750","https://openalex.org/W4238532390","https://openalex.org/W2188872161","https://openalex.org/W2002978035","https://openalex.org/W2961779879","https://openalex.org/W797688974","https://openalex.org/W2209382646"],"abstract_inverted_index":{"Deep":[0],"neural":[1,69,107,193],"networks":[2,70,108,269],"(DNNs)":[3],"are":[4,282],"increasingly":[5],"deployed":[6],"in":[7,28,113],"various":[8],"image":[9],"recognition":[10],"and":[11,21,31,45,54,157,166,201,219,236,255,280,285],"natural":[12],"language":[13],"processing":[14],"applications.":[15],"The":[16,240],"continuous":[17],"demand":[18],"for":[19,63,96,105,131,135,205,259,267],"accuracy":[20,235],"high":[22,163,167],"performance":[23,237,251],"has":[24],"led":[25],"to":[26,59,77,101,143,151,154,175,215,262,277],"innovations":[27],"DNN":[29,39,97,137],"design":[30],"a":[32,49,93,128,222],"proliferation":[33],"of":[34,52,116,238,275],"new":[35,72,82,110,149,181],"operators.":[36,65,182],"However,":[37],"existing":[38,85],"training":[40,98,220],"frameworks":[41,99],"such":[42],"as":[43,148,221],"PyTorch":[44],"TensorFlow":[46],"only":[47],"support":[48,176],"limited":[50],"range":[51],"operators":[53,83,86,111,150],"rely":[55],"on":[56],"hand-optimized":[57],"libraries":[58,204],"provide":[60,88,102],"efficient":[61,117,133,187,212],"implementations":[62,90,104,159],"these":[64],"To":[66],"evaluate":[67],"novel":[68,268],"with":[71,84,180],"operators,":[73],"the":[74,80,106,114,141,192,217,229,233,253,271],"programmers":[75,142],"have":[76],"either":[78],"replace":[79],"holistic":[81],"or":[87,248],"low-level":[89,158],"manually.":[91],"Therefore,":[92],"critical":[94],"requirement":[95],"is":[100,127],"high-performance":[103,136,203],"containing":[109],"automatically":[112],"absence":[115],"library":[118],"support.":[119],"In":[120,228],"this":[121],"paper,":[122],"we":[123,231],"introduce":[124],"NeoFlow,":[125],"which":[126],"flexible":[129],"framework":[130],"enabling":[132],"compilation":[134,188,218],"training.":[138],"NeoFlow":[139,170,184,209,244,276],"allows":[140],"directly":[144],"write":[145],"customized":[146,177],"expressions":[147],"be":[152],"mapped":[153],"graph":[155,195,257],"representation":[156],"automatically,":[160],"providing":[161],"both":[162],"programming":[164],"productivity":[165],"performance.":[168],"First,":[169],"provides":[171],"expression-based":[172],"automatic":[173],"differentiation":[174],"model":[178],"definitions":[179],"Then,":[183],"proposes":[185],"an":[186,211],"system":[189,214],"that":[190,243],"partitions":[191],"network":[194],"into":[196],"subgraphs,":[197],"explores":[198],"optimized":[199],"schedules,":[200],"generates":[202],"subgraphs":[206],"automatically.":[207],"Finally,":[208],"develops":[210],"runtime":[213],"combine":[216],"whole":[223,256],"by":[224],"overlapping":[225],"their":[226],"execution.":[227],"experiments,":[230],"examine":[232],"numerical":[234],"NeoFlow.":[239],"results":[241],"show":[242],"can":[245],"achieve":[246],"similar":[247],"even":[249],"better":[250],"at":[252],"operator":[254],"level":[258],"DNNs":[260],"compared":[261],"deep":[263],"learning":[264],"frameworks.":[265],"Especially,":[266],"training,":[270],"geometric":[272],"mean":[273],"speedups":[274],"PyTorch,":[278],"TensorFlow,":[279],"CuDNN":[281],"3.16X,":[283],"2.43X,":[284],"1.92X,":[286],"respectively.":[287]},"counts_by_year":[{"year":2025,"cited_by_count":1},{"year":2024,"cited_by_count":3},{"year":2023,"cited_by_count":3},{"year":2022,"cited_by_count":3}],"updated_date":"2025-11-06T03:46:38.306776","created_date":"2025-10-10T00:00:00"}