{"id":"https://openalex.org/W3203657632","doi":"https://doi.org/10.1109/ijcnn52387.2021.9534402","title":"RefBERT: Compressing BERT by Referencing to Pre-computed Representations","display_name":"RefBERT: Compressing BERT by Referencing to Pre-computed Representations","publication_year":2021,"publication_date":"2021-07-18","ids":{"openalex":"https://openalex.org/W3203657632","doi":"https://doi.org/10.1109/ijcnn52387.2021.9534402","mag":"3203657632"},"language":"en","primary_location":{"id":"doi:10.1109/ijcnn52387.2021.9534402","is_oa":false,"landing_page_url":"https://doi.org/10.1109/ijcnn52387.2021.9534402","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2021 International Joint Conference on Neural Networks (IJCNN)","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"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/A5100382950","display_name":"Xinyi Wang","orcid":"https://orcid.org/0000-0002-6280-1300"},"institutions":[{"id":"https://openalex.org/I154570441","display_name":"University of California, Santa Barbara","ror":"https://ror.org/02t274463","country_code":"US","type":"education","lineage":["https://openalex.org/I154570441"]}],"countries":["US"],"is_corresponding":true,"raw_author_name":"Xinyi Wang","raw_affiliation_strings":["University of California, Santa Barbara, USA"],"affiliations":[{"raw_affiliation_string":"University of California, Santa Barbara, USA","institution_ids":["https://openalex.org/I154570441"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5081662508","display_name":"Haiqin Yang","orcid":"https://orcid.org/0000-0001-5453-476X"},"institutions":[{"id":"https://openalex.org/I4401726822","display_name":"Ping An (China)","ror":"https://ror.org/004yv2z91","country_code":null,"type":"company","lineage":["https://openalex.org/I4401726822"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Haiqin Yang","raw_affiliation_strings":["Ping An Life Insurance Company of China, Ltd., Shenzhen, China"],"affiliations":[{"raw_affiliation_string":"Ping An Life Insurance Company of China, Ltd., Shenzhen, China","institution_ids":["https://openalex.org/I4401726822"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5064078067","display_name":"Liang Zhao","orcid":"https://orcid.org/0000-0001-6301-1311"},"institutions":[{"id":"https://openalex.org/I4401726822","display_name":"Ping An (China)","ror":"https://ror.org/004yv2z91","country_code":null,"type":"company","lineage":["https://openalex.org/I4401726822"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Liang Zhao","raw_affiliation_strings":["Ping An Life Insurance Company of China, Ltd., Shenzhen, China"],"affiliations":[{"raw_affiliation_string":"Ping An Life Insurance Company of China, Ltd., Shenzhen, China","institution_ids":["https://openalex.org/I4401726822"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100549617","display_name":"Yang Mo","orcid":null},"institutions":[{"id":"https://openalex.org/I4401726822","display_name":"Ping An (China)","ror":"https://ror.org/004yv2z91","country_code":null,"type":"company","lineage":["https://openalex.org/I4401726822"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yang Mo","raw_affiliation_strings":["Ping An Life Insurance Company of China, Ltd., Shenzhen, China"],"affiliations":[{"raw_affiliation_string":"Ping An Life Insurance Company of China, Ltd., Shenzhen, China","institution_ids":["https://openalex.org/I4401726822"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5108756348","display_name":"Jianping Shen","orcid":null},"institutions":[{"id":"https://openalex.org/I4401726822","display_name":"Ping An (China)","ror":"https://ror.org/004yv2z91","country_code":null,"type":"company","lineage":["https://openalex.org/I4401726822"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jianping Shen","raw_affiliation_strings":["Ping An Life Insurance Company of China, Ltd., Shenzhen, China"],"affiliations":[{"raw_affiliation_string":"Ping An Life Insurance Company of China, Ltd., Shenzhen, China","institution_ids":["https://openalex.org/I4401726822"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":5,"corresponding_author_ids":["https://openalex.org/A5100382950"],"corresponding_institution_ids":["https://openalex.org/I154570441"],"apc_list":null,"apc_paid":null,"fwci":0.4079,"has_fulltext":false,"cited_by_count":4,"citation_normalized_percentile":{"value":0.6913843,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":95,"max":99},"biblio":{"volume":null,"issue":null,"first_page":"1","last_page":"8"},"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":0.998199999332428,"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":0.991599977016449,"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/leverage","display_name":"Leverage (statistics)","score":0.7691686153411865},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.7447596788406372},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.62000572681427},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.5288977026939392},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.5114845037460327},{"id":"https://openalex.org/keywords/benchmark","display_name":"Benchmark (surveying)","score":0.49486035108566284},{"id":"https://openalex.org/keywords/natural-language-processing","display_name":"Natural language processing","score":0.4808800518512726},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.4602462649345398},{"id":"https://openalex.org/keywords/overhead","display_name":"Overhead (engineering)","score":0.4339107275009155},{"id":"https://openalex.org/keywords/theoretical-computer-science","display_name":"Theoretical computer science","score":0.3817283511161804},{"id":"https://openalex.org/keywords/machine-learning","display_name":"Machine learning","score":0.38067665696144104},{"id":"https://openalex.org/keywords/programming-language","display_name":"Programming language","score":0.28259509801864624}],"concepts":[{"id":"https://openalex.org/C153083717","wikidata":"https://www.wikidata.org/wiki/Q6535263","display_name":"Leverage (statistics)","level":2,"score":0.7691686153411865},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7447596788406372},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.62000572681427},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5288977026939392},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.5114845037460327},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.49486035108566284},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.4808800518512726},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.4602462649345398},{"id":"https://openalex.org/C2779960059","wikidata":"https://www.wikidata.org/wiki/Q7113681","display_name":"Overhead (engineering)","level":2,"score":0.4339107275009155},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.3817283511161804},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.38067665696144104},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.28259509801864624},{"id":"https://openalex.org/C13280743","wikidata":"https://www.wikidata.org/wiki/Q131089","display_name":"Geodesy","level":1,"score":0.0},{"id":"https://openalex.org/C205649164","wikidata":"https://www.wikidata.org/wiki/Q1071","display_name":"Geography","level":0,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/ijcnn52387.2021.9534402","is_oa":false,"landing_page_url":"https://doi.org/10.1109/ijcnn52387.2021.9534402","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2021 International Joint Conference on Neural Networks (IJCNN)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/4","display_name":"Quality Education","score":0.8799999952316284}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":72,"referenced_works":["https://openalex.org/W131533222","https://openalex.org/W135467536","https://openalex.org/W1821462560","https://openalex.org/W2153579005","https://openalex.org/W2250539671","https://openalex.org/W2396767181","https://openalex.org/W2565286634","https://openalex.org/W2593634001","https://openalex.org/W2739351760","https://openalex.org/W2741217076","https://openalex.org/W2896457183","https://openalex.org/W2907947679","https://openalex.org/W2923014074","https://openalex.org/W2924902521","https://openalex.org/W2939362715","https://openalex.org/W2962739339","https://openalex.org/W2963310665","https://openalex.org/W2963403868","https://openalex.org/W2963453233","https://openalex.org/W2963846996","https://openalex.org/W2964230360","https://openalex.org/W2964299589","https://openalex.org/W2965373594","https://openalex.org/W2969601108","https://openalex.org/W2970233825","https://openalex.org/W2970454332","https://openalex.org/W2970597249","https://openalex.org/W2974875810","https://openalex.org/W2978017171","https://openalex.org/W2978670439","https://openalex.org/W2996035354","https://openalex.org/W2996428491","https://openalex.org/W2996834012","https://openalex.org/W3037576147","https://openalex.org/W3083485178","https://openalex.org/W3092642435","https://openalex.org/W3093215104","https://openalex.org/W3104033643","https://openalex.org/W3105966348","https://openalex.org/W3106118231","https://openalex.org/W3118999024","https://openalex.org/W3123936891","https://openalex.org/W3156386096","https://openalex.org/W3176971293","https://openalex.org/W3185784178","https://openalex.org/W3201197320","https://openalex.org/W4287824654","https://openalex.org/W4288112596","https://openalex.org/W4294170691","https://openalex.org/W4297749952","https://openalex.org/W4385245566","https://openalex.org/W6605323724","https://openalex.org/W6638523607","https://openalex.org/W6682691769","https://openalex.org/W6739901393","https://openalex.org/W6750615492","https://openalex.org/W6755207826","https://openalex.org/W6757750581","https://openalex.org/W6760732026","https://openalex.org/W6762073003","https://openalex.org/W6763701032","https://openalex.org/W6766673545","https://openalex.org/W6766795243","https://openalex.org/W6766961549","https://openalex.org/W6768021236","https://openalex.org/W6768086466","https://openalex.org/W6768851824","https://openalex.org/W6771917389","https://openalex.org/W6779928756","https://openalex.org/W6784001418","https://openalex.org/W6794709657","https://openalex.org/W6797710381"],"related_works":["https://openalex.org/W2378211422","https://openalex.org/W2745001401","https://openalex.org/W4321353415","https://openalex.org/W2130974462","https://openalex.org/W972276598","https://openalex.org/W4246352526","https://openalex.org/W2028665553","https://openalex.org/W4230315250","https://openalex.org/W2086519370","https://openalex.org/W4379251913"],"abstract_inverted_index":{"Recently":[0],"developed":[1],"large":[2],"pre-trained":[3,20],"language":[4,16,21],"models,":[5],"e.g.,":[6],"BERT,":[7],"have":[8],"achieved":[9],"remarkable":[10],"performance":[11,60,193],"in":[12,37,61,89,92,162],"many":[13],"downstream":[14,62],"natural":[15],"processing":[17],"applications.":[18,39,63],"These":[19],"models":[22,50],"often":[23],"contain":[24],"hundreds":[25],"of":[26,28,48,66,139,191,194],"millions":[27],"parameters":[29],"and":[30,35,54,116,136,173,186,209],"suffer":[31],"from":[32,103],"high":[33],"computation":[34,46],"latency":[36],"real-world":[38],"It":[40],"is":[41,204],"desirable":[42],"to":[43,71,76,98],"reduce":[44],"the":[45,49,58,73,85,100,104,108,113,133,137,143,149,154,159,164,170,180,192,199],"overhead":[47],"for":[51],"fast":[52],"training":[53],"inference":[55,215],"while":[56],"keeping":[57],"model":[59,75],"Several":[64],"lines":[65],"work":[67],"utilize":[68],"knowledge":[69,87,101],"distillation":[70],"compress":[72],"teacher":[74],"a":[77,120],"smaller":[78,121,208],"student":[79,122,165],"model.":[80,123,166],"However,":[81],"they":[82],"usually":[83],"discard":[84],"teacher's":[86,151],"when":[88],"inference.":[90],"Differently,":[91],"this":[93],"paper,":[94],"we":[95,128,168],"propose":[96],"RemERT":[97,177,203],"leverage":[99],"learned":[102],"teacher,":[105],"i.e.,":[106],"facilitating":[107],"pre-computed":[109,150],"BERT":[110,118],"representation":[111],"on":[112,132,153,198,214],"reference":[114,140,155],"sample":[115],"compressing":[117],"into":[119],"To":[124],"guarantee":[125],"our":[126,176],"proposal,":[127],"provide":[129],"theoretical":[130,144],"justification":[131],"loss":[134],"function":[135],"usage":[138],"samples.":[141],"Significantly,":[142],"result":[145],"shows":[146],"that":[147,175],"including":[148],"representations":[152],"samples":[156],"indeed":[157],"increases":[158],"mutual":[160],"information":[161],"learning":[163],"Finally,":[167],"conduct":[169],"empirical":[171],"evaluation":[172],"show":[174],"can":[178],"beat":[179],"vanilla":[181],"TinyBERT":[182],"over":[183],"8.1":[184],"%":[185],"achieves":[187],"more":[188],"than":[189,216],"94%":[190],"<tex":[195,205,210,217],"xmlns:mml=\"http://www.w3.org/1998/Math/MathML\"":[196,206,211,218],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">$\\mathbf{BERT}_{\\mathbf{BASE}}$</tex>":[197,219],"GLUE":[200],"benchmark.":[201],"Meanwhile,":[202],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">$\\mathbf{7.4x}$</tex>":[207],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">$\\mathbf{9.5x}$</tex>":[212],"faster":[213],".":[220]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2021,"cited_by_count":3}],"updated_date":"2026-01-13T01:12:25.745995","created_date":"2025-10-10T00:00:00"}