{"id":"https://openalex.org/W4386566441","doi":"https://doi.org/10.18653/v1/2023.findings-eacl.43","title":"Task-specific Compression for Multi-task Language Models using Attribution-based Pruning","display_name":"Task-specific Compression for Multi-task Language Models using Attribution-based Pruning","publication_year":2023,"publication_date":"2023-01-01","ids":{"openalex":"https://openalex.org/W4386566441","doi":"https://doi.org/10.18653/v1/2023.findings-eacl.43"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2023.findings-eacl.43","doi":"https://doi.org/10.18653/v1/2023.findings-eacl.43","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2023.findings-eacl.43","pdf_url":"https://aclanthology.org/2023.findings-eacl.43.pdf","source":{"id":"https://openalex.org/S7407087051","display_name":"Findings of the Association for Computational Linguistics: EACL","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: EACL 2023","raw_type":"proceedings-article"},"volume":null,"type":"conference-paper","indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"hybrid","oa_url":"https://aclanthology.org/2023.findings-eacl.43.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5036753528","display_name":"Nakyeong Yang","orcid":"https://orcid.org/0000-0002-2196-5149","observed_orcids":["https://orcid.org/0000-0002-2196-5149"]},"institutions":[{"id":"https://openalex.org/I139264467","display_name":"Seoul National University","ror":"https://ror.org/04h9pn542","country_code":"KR","type":"education","lineage":["https://openalex.org/I139264467"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Nakyeong Yang","raw_affiliation_strings":["Seoul National University,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Seoul National University,","institution_ids":["https://openalex.org/I139264467"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5088353147","display_name":"Yunah Jang","orcid":"https://orcid.org/0000-0002-2805-7530","observed_orcids":["https://orcid.org/0000-0002-2805-7530"]},"institutions":[{"id":"https://openalex.org/I139264467","display_name":"Seoul National University","ror":"https://ror.org/04h9pn542","country_code":"KR","type":"education","lineage":["https://openalex.org/I139264467"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Yunah Jang","raw_affiliation_strings":["Seoul National University,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Seoul National University,","institution_ids":["https://openalex.org/I139264467"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5063029769","display_name":"Hwanhee Lee","orcid":"https://orcid.org/0000-0002-9367-9811","observed_orcids":["https://orcid.org/0000-0002-9367-9811"]},"institutions":[{"id":"https://openalex.org/I67900169","display_name":"Chung-Ang University","ror":"https://ror.org/01r024a98","country_code":"KR","type":"education","lineage":["https://openalex.org/I67900169"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Hwanhee Lee","raw_affiliation_strings":["Chung-Ang University,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Chung-Ang University,","institution_ids":["https://openalex.org/I67900169"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5104083391","display_name":"Seohyeong Jeong","orcid":"https://orcid.org/0000-0002-3292-0691","observed_orcids":["https://orcid.org/0000-0002-3292-0691"]},"institutions":[{"id":"https://openalex.org/I197312522","display_name":"Hyundai Motor Group (South Korea)","ror":"https://ror.org/05kxbz959","country_code":"KR","type":"company","lineage":["https://openalex.org/I197312522"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Seohyeong Jeong","raw_affiliation_strings":["Hyundai Motor Group and 42dot Inc"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Hyundai Motor Group and 42dot Inc","institution_ids":["https://openalex.org/I197312522"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5077832834","display_name":"Kyomin Jung","orcid":"https://orcid.org/0000-0003-2547-7051","observed_orcids":["https://orcid.org/0000-0003-2547-7051"]},"institutions":[{"id":"https://openalex.org/I139264467","display_name":"Seoul National University","ror":"https://ror.org/04h9pn542","country_code":"KR","type":"education","lineage":["https://openalex.org/I139264467"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Kyomin Jung","raw_affiliation_strings":["Seoul National University,"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Seoul National University,","institution_ids":["https://openalex.org/I139264467"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":3,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.9596,"has_fulltext":true,"cited_by_count":9,"citation_normalized_percentile":{"value":0.84612813,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":94,"max":98},"biblio":{"volume":null,"issue":null,"first_page":"594","last_page":"604"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.8428000211715698,"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":0.8428000211715698,"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/T10028","display_name":"Topic Modeling","score":0.1006999984383583,"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":0.013399999588727951,"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/training-free-compression","display_name":"training-free compression","score":0.6610000133514404},{"id":"https://openalex.org/keywords/neuron-pruning","display_name":"neuron pruning","score":0.7739999890327454},{"id":"https://openalex.org/keywords/parameter-efficiency","display_name":"parameter efficiency","score":0.4729999899864197},{"id":"https://openalex.org/keywords/resource-limited-settings","display_name":"resource-limited settings","score":0.843999981880188},{"id":"https://openalex.org/keywords/unseen-domain-generalization","display_name":"unseen domain generalization","score":0.6869999766349792}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8685317039489746},{"id":"https://openalex.org/C108010975","wikidata":"https://www.wikidata.org/wiki/Q500094","display_name":"Pruning","level":2,"score":0.8089052438735962},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.7818666696548462},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.7305420637130737},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.6210461258888245},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.5416399240493774},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.4837653636932373},{"id":"https://openalex.org/C195324797","wikidata":"https://www.wikidata.org/wiki/Q33742","display_name":"Natural language","level":2,"score":0.4749565124511719},{"id":"https://openalex.org/C175154964","wikidata":"https://www.wikidata.org/wiki/Q380077","display_name":"Task analysis","level":3,"score":0.422243595123291},{"id":"https://openalex.org/C28006648","wikidata":"https://www.wikidata.org/wiki/Q6934509","display_name":"Multi-task learning","level":3,"score":0.4106122553348541},{"id":"https://openalex.org/C162324750","wikidata":"https://www.wikidata.org/wiki/Q8134","display_name":"Economics","level":0,"score":0.0},{"id":"https://openalex.org/C187736073","wikidata":"https://www.wikidata.org/wiki/Q2920921","display_name":"Management","level":1,"score":0.0},{"id":"https://openalex.org/C6557445","wikidata":"https://www.wikidata.org/wiki/Q173113","display_name":"Agronomy","level":1,"score":0.0},{"id":"https://openalex.org/C86803240","wikidata":"https://www.wikidata.org/wiki/Q420","display_name":"Biology","level":0,"score":0.0}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.18653/v1/2023.findings-eacl.43","doi":"https://doi.org/10.18653/v1/2023.findings-eacl.43","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2023.findings-eacl.43","pdf_url":"https://aclanthology.org/2023.findings-eacl.43.pdf","source":{"id":"https://openalex.org/S7407087051","display_name":"Findings of the Association for Computational Linguistics: EACL","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: EACL 2023","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2205.04157","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2205.04157","pdf_url":"https://arxiv.org/pdf/2205.04157","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.2205.04157","doi":"https://doi.org/10.48550/arxiv.2205.04157","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2205.04157","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"article-journal"}],"best_oa_location":{"id":"doi:10.18653/v1/2023.findings-eacl.43","doi":"https://doi.org/10.18653/v1/2023.findings-eacl.43","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2023.findings-eacl.43","pdf_url":"https://aclanthology.org/2023.findings-eacl.43.pdf","source":{"id":"https://openalex.org/S7407087051","display_name":"Findings of the Association for Computational Linguistics: EACL","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: EACL 2023","raw_type":"proceedings-article"},"sustainable_development_goals":[],"study_designs":[],"awards":[{"id":"https://openalex.org/G1152552616","display_name":null,"funder_award_id":"2021\u20130-01343","funder_id":"https://openalex.org/F4320335489","funder_display_name":"Institute for Information and Communications Technology Promotion"},{"id":"https://openalex.org/G1478288009","display_name":null,"funder_award_id":"2021-0-02068","funder_id":"https://openalex.org/F4320328359","funder_display_name":"Ministry of Science and ICT, South Korea"},{"id":"https://openalex.org/G2206812781","display_name":null,"funder_award_id":"2021-0-02068","funder_id":"https://openalex.org/F4320321292","funder_display_name":"Seoul National University"},{"id":"https://openalex.org/G3450174791","display_name":null,"funder_award_id":"2022-0-00184","funder_id":"https://openalex.org/F4320335489","funder_display_name":"Institute for Information and Communications Technology Promotion"},{"id":"https://openalex.org/G4643994530","display_name":null,"funder_award_id":"2021-0-02068","funder_id":"https://openalex.org/F4320335489","funder_display_name":"Institute for Information and Communications Technology Promotion"},{"id":"https://openalex.org/G4700831490","display_name":null,"funder_award_id":"2022-","funder_id":"https://openalex.org/F4320335489","funder_display_name":"Institute for Information and Communications Technology Promotion"},{"id":"https://openalex.org/G5065061179","display_name":null,"funder_award_id":"2021-0-01343","funder_id":"https://openalex.org/F4320328359","funder_display_name":"Ministry of Science and ICT, South Korea"},{"id":"https://openalex.org/G5635159257","display_name":null,"funder_award_id":"2021-0-01343","funder_id":"https://openalex.org/F4320321292","funder_display_name":"Seoul National University"},{"id":"https://openalex.org/G6758007335","display_name":null,"funder_award_id":"2022-0-00184","funder_id":"https://openalex.org/F4320328359","funder_display_name":"Ministry of Science and ICT, South Korea"}],"funders":[{"id":"https://openalex.org/F4320321292","display_name":"Seoul National University","ror":"https://ror.org/04h9pn542"},{"id":"https://openalex.org/F4320328359","display_name":"Ministry of Science and ICT, South Korea","ror":"https://ror.org/01wpjm123"},{"id":"https://openalex.org/F4320335489","display_name":"Institute for Information and Communications Technology Promotion","ror":"https://ror.org/01g0hqq23"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4386566441.pdf","grobid_xml":"https://content.openalex.org/works/W4386566441.grobid-xml"},"referenced_works_count":36,"referenced_works":["https://openalex.org/W1682403713","https://openalex.org/W1821462560","https://openalex.org/W2119144962","https://openalex.org/W2123045220","https://openalex.org/W2150165932","https://openalex.org/W2346578521","https://openalex.org/W2495425901","https://openalex.org/W2513419314","https://openalex.org/W2515385951","https://openalex.org/W2560647685","https://openalex.org/W2626778328","https://openalex.org/W2803552920","https://openalex.org/W2896457183","https://openalex.org/W2903996579","https://openalex.org/W2923014074","https://openalex.org/W2943552823","https://openalex.org/W2965373594","https://openalex.org/W2969515962","https://openalex.org/W2975381464","https://openalex.org/W2978017171","https://openalex.org/W2981852735","https://openalex.org/W2982399380","https://openalex.org/W2998183051","https://openalex.org/W3015468748","https://openalex.org/W3015609966","https://openalex.org/W3030163527","https://openalex.org/W3033529678","https://openalex.org/W3034457371","https://openalex.org/W3034742519","https://openalex.org/W3105966348","https://openalex.org/W3115511229","https://openalex.org/W3116594510","https://openalex.org/W3174173633","https://openalex.org/W3175505246","https://openalex.org/W3180037928","https://openalex.org/W3208397797"],"related_works":["https://openalex.org/W4226483909","https://openalex.org/W4200631806","https://openalex.org/W4221164185","https://openalex.org/W4406880340","https://openalex.org/W7101941617","https://openalex.org/W2963972328","https://openalex.org/W7103754027","https://openalex.org/W7159671713","https://openalex.org/W4312228839","https://openalex.org/W7131870106"],"abstract_inverted_index":{"Multi-task":[0],"language":[1,9,18,45,106],"models":[2,19,46],"show":[3,114],"outstanding":[4],"performance":[5,132],"for":[6,32,43,62],"various":[7],"natural":[8],"understanding":[10],"tasks":[11],"with":[12],"only":[13,31,73],"a":[14,33,38,48,64],"single":[15],"model.However,":[16],"these":[17],"utilize":[20],"an":[21,53,135],"unnecessarily":[22],"large":[23],"number":[24],"of":[25,105],"model":[26],"parameters,":[27],"even":[28,133],"when":[29],"used":[30],"specific":[34,65],"task.This":[35],"paper":[36],"proposes":[37],"novel":[39],"training-free":[40],"compression":[41,89],"method":[42,55,79,90,119,130],"multi-task":[44],"using":[47],"pruning":[49,118,123],"method.Specifically,":[50],"we":[51,76,126],"use":[52],"attribution":[54],"to":[56,80],"determine":[57],"which":[58],"neurons":[59,70],"are":[60],"essential":[61],"performing":[63],"task.We":[66],"task-specifically":[67],"prune":[68],"unimportant":[69],"and":[71,85,98],"leave":[72],"task-specific":[74],"parameters.Furthermore,":[75],"extend":[77],"our":[78,88,116,129],"be":[81],"applicable":[82],"in":[83,134],"low-resource":[84],"unsupervised":[86],"settings.Since":[87],"is":[91],"training-free,":[92],"it":[93],"uses":[94],"few":[95],"computing":[96],"resources":[97],"does":[99],"not":[100],"destroy":[101],"the":[102,110],"pre-trained":[103],"knowledge":[104],"models.Experimental":[107],"results":[108],"on":[109],"six":[111],"widelyused":[112],"datasets":[113],"that":[115,128],"proposed":[117],"significantly":[120],"outperforms":[121],"baseline":[122],"methods.In":[124],"addition,":[125],"demonstrate":[127],"preserves":[131],"unseen":[136],"domain":[137],"setting.":[138]},"counts_by_year":[{"year":2026,"cited_by_count":2},{"year":2025,"cited_by_count":5},{"year":2024,"cited_by_count":2}],"updated_date":"2026-10-30T12:39:25.321706","created_date":"2023-09-10T00:00:00"}