{"id":"https://openalex.org/W4281552875","doi":"https://doi.org/10.48550/arxiv.2205.11747","title":"BabyBear: Cheap inference triage for expensive language models","display_name":"BabyBear: Cheap inference triage for expensive language models","publication_year":2022,"publication_date":"2022-05-24","ids":{"openalex":"https://openalex.org/W4281552875","doi":"https://doi.org/10.48550/arxiv.2205.11747"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2205.11747","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2205.11747","pdf_url":"https://arxiv.org/pdf/2205.11747","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2205.11747","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":null,"display_name":"Khalili, Leila","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Khalili, Leila","raw_affiliation_strings":["Primer Technologies Inc. San Francisco , California"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Primer Technologies Inc. San Francisco , California","institution_ids":[]}]},{"author_position":"middle","author":{"id":null,"display_name":"You, Yao","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"You, Yao","raw_affiliation_strings":["Primer Technologies Inc. San Francisco , California"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Primer Technologies Inc. San Francisco , California","institution_ids":[]}]},{"author_position":"last","author":{"id":null,"display_name":"Bohannon, John","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Bohannon, John","raw_affiliation_strings":["Primer Technologies Inc. San Francisco , California"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Primer Technologies Inc. San Francisco , California","institution_ids":[]}]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":14,"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":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.9315999746322632,"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":0.9315999746322632,"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":0.03889999911189079,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.019300000742077827,"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/large-language-models","display_name":"large language models","score":0.6439999938011169},{"id":"https://openalex.org/keywords/early-exit","display_name":"early exit","score":0.5199999809265137},{"id":"https://openalex.org/keywords/document-classification","display_name":"document classification","score":0.6690000295639038},{"id":"https://openalex.org/keywords/named-entity-recognition","display_name":"named entity recognition","score":0.9990000128746033},{"id":"https://openalex.org/keywords/transformer","display_name":"transformer models","score":0.44999998807907104},{"id":"https://openalex.org/keywords/energy-efficiency","display_name":"energy efficiency","score":0.4300000071525574}],"concepts":[{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.8435973525047302},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8367822170257568},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.6824420690536499},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.6794109344482422},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.5823871493339539},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.5690043568611145},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.5492179989814758},{"id":"https://openalex.org/C2777120189","wikidata":"https://www.wikidata.org/wiki/Q780067","display_name":"Triage","level":2,"score":0.5485303401947021},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.497743159532547},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.46571946144104004},{"id":"https://openalex.org/C44291984","wikidata":"https://www.wikidata.org/wiki/Q1074173","display_name":"Question answering","level":2,"score":0.41174018383026123},{"id":"https://openalex.org/C194828623","wikidata":"https://www.wikidata.org/wiki/Q2861470","display_name":"Emergency medicine","level":1,"score":0.0},{"id":"https://openalex.org/C165801399","wikidata":"https://www.wikidata.org/wiki/Q25428","display_name":"Voltage","level":2,"score":0.0},{"id":"https://openalex.org/C71924100","wikidata":"https://www.wikidata.org/wiki/Q11190","display_name":"Medicine","level":0,"score":0.0},{"id":"https://openalex.org/C62520636","wikidata":"https://www.wikidata.org/wiki/Q944","display_name":"Quantum mechanics","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},{"id":"https://openalex.org/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"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":2,"locations":[{"id":"pmh:oai:arXiv.org:2205.11747","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2205.11747","pdf_url":"https://arxiv.org/pdf/2205.11747","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":"doi:10.48550/arxiv.2205.11747","doi":"https://doi.org/10.48550/arxiv.2205.11747","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2205.11747","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2205.11747","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2205.11747","pdf_url":"https://arxiv.org/pdf/2205.11747","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"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":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4281552875.pdf","grobid_xml":"https://content.openalex.org/works/W4281552875.grobid-xml"},"referenced_works_count":12,"referenced_works":["https://openalex.org/W2552110825","https://openalex.org/W2767421475","https://openalex.org/W2794557536","https://openalex.org/W2896457183","https://openalex.org/W2952732525","https://openalex.org/W2965373594","https://openalex.org/W2978017171","https://openalex.org/W3157700140","https://openalex.org/W3201357245","https://openalex.org/W3211899209","https://openalex.org/W4224075666","https://openalex.org/W4300668032"],"related_works":["https://openalex.org/W4224075666","https://openalex.org/W4281381191","https://openalex.org/W6957687027","https://openalex.org/W4399354026","https://openalex.org/W3115348206","https://openalex.org/W3016791490","https://openalex.org/W4411403396","https://openalex.org/W3022719076","https://openalex.org/W4416016882","https://openalex.org/W7166589956"],"abstract_inverted_index":{"Transformer":[0],"language":[1,35],"models":[2,8,32,99],"provide":[3],"superior":[4],"accuracy":[5],"over":[6],"previous":[7],"but":[9],"they":[10],"are":[11],"computationally":[12],"and":[13,76],"environmentally":[14],"expensive.":[15],"Borrowing":[16],"the":[17,51,56,90,114,137,150],"concept":[18],"of":[19,89,117,136],"model":[20,54],"cascading":[21,31],"from":[22],"computer":[23],"vision,":[24],"we":[25,133],"introduce":[26],"BabyBear,":[27],"a":[28,59,86,105],"framework":[29],"for":[30,33,82],"natural":[34],"processing":[36],"(NLP)":[37],"tasks":[38,85],"to":[39,73,112],"minimize":[40],"cost.":[41],"The":[42],"core":[43],"strategy":[44],"is":[45],"inference":[46,91],"triage,":[47],"exiting":[48],"early":[49],"when":[50],"least":[52],"expensive":[53],"in":[55],"cascade":[57],"achieves":[58],"sufficiently":[60],"high-confidence":[61],"prediction.":[62],"We":[63,79],"test":[64],"BabyBear":[65],"on":[66,149],"several":[67],"open":[68],"source":[69],"data":[70],"sets":[71],"related":[72],"document":[74],"classification":[75,119],"entity":[77,131],"recognition.":[78],"find":[80],"that":[81,100],"common":[83],"NLP":[84],"high":[87],"proportion":[88],"load":[92],"can":[93],"be":[94],"accomplished":[95],"with":[96],"cheap,":[97],"fast":[98],"have":[101],"learned":[102],"by":[103,121],"observing":[104],"deep":[106,138],"learning":[107,139],"model.":[108],"This":[109],"allows":[110],"us":[111],"reduce":[113],"compute":[115,140],"cost":[116],"large-scale":[118],"jobs":[120],"more":[122],"than":[123,147],"50%":[124],"while":[125,141],"retaining":[126],"overall":[127],"accuracy.":[128],"For":[129],"named":[130],"recognition,":[132],"save":[134],"33%":[135],"maintaining":[142],"an":[143],"F1":[144],"score":[145],"higher":[146],"95%":[148],"CoNLL":[151],"benchmark.":[152]},"counts_by_year":[{"year":2026,"cited_by_count":2},{"year":2025,"cited_by_count":5},{"year":2024,"cited_by_count":6},{"year":2023,"cited_by_count":1}],"updated_date":"2026-11-10T11:39:43.021295","created_date":"2022-05-27T00:00:00"}