{"id":"https://openalex.org/W4288065691","doi":"https://doi.org/10.48550/arxiv.2112.08718","title":"Prompt Tuning GPT-2 language model for parameter-efficient domain adaptation of ASR systems","display_name":"Prompt Tuning GPT-2 language model for parameter-efficient domain adaptation of ASR systems","publication_year":2021,"publication_date":"2021-12-16","ids":{"openalex":"https://openalex.org/W4288065691","doi":"https://doi.org/10.48550/arxiv.2112.08718"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2112.08718","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2112.08718","pdf_url":"https://arxiv.org/pdf/2112.08718","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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2112.08718","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5027419911","display_name":"Saket Dingliwal","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Dingliwal, Saket","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5030041600","display_name":"Ashish Shenoy","orcid":"https://orcid.org/0000-0003-1401-262X","observed_orcids":["https://orcid.org/0000-0003-1401-262X"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shenoy, Ashish","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5050552820","display_name":"Sravan Bodapati","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Bodapati, Sravan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5014054572","display_name":"Ankur Gandhe","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Gandhe, Ankur","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5050071432","display_name":"Ravi Teja Gadde","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Gadde, Ravi Teja","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5050325468","display_name":"Katrin Kirchhoff","orcid":"https://orcid.org/0000-0002-6645-6030","observed_orcids":["https://orcid.org/0000-0002-6645-6030"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kirchhoff, Katrin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"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":4,"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/T10201","display_name":"Speech Recognition and Synthesis","score":0.9926999807357788,"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/T10201","display_name":"Speech Recognition and Synthesis","score":0.9926999807357788,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.006800000090152025,"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/T10028","display_name":"Topic Modeling","score":0.00019999999494757503,"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/prompt-tuning","display_name":"prompt tuning","score":0.9340000152587891},{"id":"https://openalex.org/keywords/gpt-2","display_name":"GPT-2","score":0.8960000276565552},{"id":"https://openalex.org/keywords/domain-adaptation","display_name":"domain adaptation","score":0.8100000023841858},{"id":"https://openalex.org/keywords/speech-recognition","display_name":"speech recognition","score":0.9570000171661377},{"id":"https://openalex.org/keywords/domain-embedding","display_name":"domain embedding","score":0.8199999928474426},{"id":"https://openalex.org/keywords/parameter-efficient-fine-tuning","display_name":"parameter-efficient fine-tuning","score":0.7770000100135803}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8227996826171875},{"id":"https://openalex.org/C2776434776","wikidata":"https://www.wikidata.org/wiki/Q19246213","display_name":"Domain adaptation","level":3,"score":0.8042882680892944},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.6914181709289551},{"id":"https://openalex.org/C105339364","wikidata":"https://www.wikidata.org/wiki/Q2297740","display_name":"Software deployment","level":2,"score":0.6361324191093445},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.6349056959152222},{"id":"https://openalex.org/C36503486","wikidata":"https://www.wikidata.org/wiki/Q11235244","display_name":"Domain (mathematical analysis)","level":2,"score":0.6153600811958313},{"id":"https://openalex.org/C2779960059","wikidata":"https://www.wikidata.org/wiki/Q7113681","display_name":"Overhead (engineering)","level":2,"score":0.5651347041130066},{"id":"https://openalex.org/C41608201","wikidata":"https://www.wikidata.org/wiki/Q980509","display_name":"Embedding","level":2,"score":0.5568700432777405},{"id":"https://openalex.org/C139807058","wikidata":"https://www.wikidata.org/wiki/Q352374","display_name":"Adaptation (eye)","level":2,"score":0.4813152551651001},{"id":"https://openalex.org/C157524613","wikidata":"https://www.wikidata.org/wiki/Q2828883","display_name":"Fine-tuning","level":2,"score":0.48116788268089294},{"id":"https://openalex.org/C123657996","wikidata":"https://www.wikidata.org/wiki/Q12271","display_name":"Architecture","level":2,"score":0.4163036644458771},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.3843376934528351},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.369276225566864},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.09920772910118103},{"id":"https://openalex.org/C115903868","wikidata":"https://www.wikidata.org/wiki/Q80993","display_name":"Software engineering","level":1,"score":0.09380605816841125},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.06624817848205566},{"id":"https://openalex.org/C119599485","wikidata":"https://www.wikidata.org/wiki/Q43035","display_name":"Electrical engineering","level":1,"score":0.059599995613098145},{"id":"https://openalex.org/C134306372","wikidata":"https://www.wikidata.org/wiki/Q7754","display_name":"Mathematical analysis","level":1,"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/C62520636","wikidata":"https://www.wikidata.org/wiki/Q944","display_name":"Quantum mechanics","level":1,"score":0.0},{"id":"https://openalex.org/C120665830","wikidata":"https://www.wikidata.org/wiki/Q14620","display_name":"Optics","level":1,"score":0.0},{"id":"https://openalex.org/C142362112","wikidata":"https://www.wikidata.org/wiki/Q735","display_name":"Art","level":0,"score":0.0},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"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/C95623464","wikidata":"https://www.wikidata.org/wiki/Q1096149","display_name":"Classifier (UML)","level":2,"score":0.0},{"id":"https://openalex.org/C153349607","wikidata":"https://www.wikidata.org/wiki/Q36649","display_name":"Visual arts","level":1,"score":0.0}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2112.08718","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2112.08718","pdf_url":"https://arxiv.org/pdf/2112.08718","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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2112.08718","doi":"https://doi.org/10.48550/arxiv.2112.08718","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2112.08718","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:2112.08718","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2112.08718","pdf_url":"https://arxiv.org/pdf/2112.08718","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":null,"license_id":null,"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":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4225963373","https://openalex.org/W3205893931","https://openalex.org/W3208017279","https://openalex.org/W4402402306","https://openalex.org/W3049591882","https://openalex.org/W4406461904","https://openalex.org/W7117706738","https://openalex.org/W4391021457","https://openalex.org/W3131668513","https://openalex.org/W4214631905"],"abstract_inverted_index":{"Automatic":[0],"Speech":[1],"Recognition":[2],"(ASR)":[3],"systems":[4,148],"have":[5],"found":[6],"their":[7],"use":[8],"in":[9,13,146],"numerous":[10],"industrial":[11],"applications":[12],"very":[14],"diverse":[15],"domains":[16,24],"creating":[17],"a":[18,37,42,51,57,74],"need":[19],"to":[20,22,49,56,120,123,149,153],"adapt":[21],"new":[23,75,154],"with":[25,77],"small":[26,43],"memory":[27],"and":[28],"deployment":[29],"overhead.":[30],"In":[31],"this":[32,61],"work,":[33],"we":[34],"introduce":[35],"domain-prompts,":[36],"methodology":[38],"that":[39],"involves":[40],"training":[41],"number":[44],"of":[45,99,102,143],"domain":[46,76,116],"embedding":[47],"parameters":[48,101],"prime":[50],"Transformer-based":[52],"Language":[53],"Model":[54],"(LM)":[55],"particular":[58],"domain.":[59],"Using":[60],"domain-adapted":[62],"LM":[63,105],"for":[64,73,140],"rescoring":[65],"ASR":[66,147],"hypotheses":[67],"can":[68],"achieve":[69],"7-13%":[70],"WER":[71],"reduction":[72],"just":[78,97],"1000":[79],"unlabeled":[80],"textual":[81],"domain-specific":[82],"sentences.":[83],"This":[84],"improvement":[85],"is":[86,111,136],"comparable":[87],"or":[88],"even":[89,95],"better":[90],"than":[91,127],"fully":[92],"fine-tuned":[93],"models":[94],"though":[96],"0.02%":[98],"the":[100,103,114,121,124,129],"base":[104,130],"are":[106,118],"updated.":[107],"Additionally,":[108],"our":[109,134],"method":[110,135],"deployment-friendly":[112],"as":[113],"learnt":[115],"embeddings":[117],"prefixed":[119],"input":[122],"model":[125,131],"rather":[126],"changing":[128],"architecture.":[132],"Therefore,":[133],"an":[137],"ideal":[138],"choice":[139],"on-the-fly":[141],"adaptation":[142],"LMs":[144],"used":[145],"progressively":[150],"scale":[151],"it":[152],"domains.":[155]},"counts_by_year":[{"year":2025,"cited_by_count":1},{"year":2024,"cited_by_count":2},{"year":2023,"cited_by_count":1}],"updated_date":"2026-11-04T10:54:05.112534","created_date":"2022-07-28T00:00:00"}