{"id":"https://openalex.org/W4417271153","doi":"https://doi.org/10.18653/v1/2026.findings-acl.1355","title":"Soft Head Selection for Injecting ICL-Derived Task Embeddings","display_name":"Soft Head Selection for Injecting ICL-Derived Task Embeddings","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W4417271153","doi":"https://doi.org/10.18653/v1/2026.findings-acl.1355"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2026.findings-acl.1355","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1355","pdf_url":"https://aclanthology.org/2026.findings-acl.1355.pdf","source":null,"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: ACL 2026","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2026.findings-acl.1355.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5019065183","display_name":"Jungwon Park","orcid":"https://orcid.org/0000-0002-4930-5628"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jungwon Park","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Jimyeong Kim","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jimyeong Kim","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Changin Choi","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Changin Choi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5103058698","display_name":"Wonjong Rhee","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wonjong Rhee","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":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.03505335,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"27161","last_page":"27214"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.3885999917984009,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.3885999917984009,"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/T10429","display_name":"EEG and Brain-Computer Interfaces","score":0.13609999418258667,"subfield":{"id":"https://openalex.org/subfields/2805","display_name":"Cognitive Neuroscience"},"field":{"id":"https://openalex.org/fields/28","display_name":"Neuroscience"},"domain":{"id":"https://openalex.org/domains/1","display_name":"Life Sciences"}},{"id":"https://openalex.org/T13702","display_name":"Machine Learning in Healthcare","score":0.08020000159740448,"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/task","display_name":"Task (project management)","score":0.8165000081062317},{"id":"https://openalex.org/keywords/transfer-of-learning","display_name":"Transfer of learning","score":0.5157999992370605},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.4936999976634979},{"id":"https://openalex.org/keywords/head","display_name":"Head (geology)","score":0.47369998693466187},{"id":"https://openalex.org/keywords/space","display_name":"Space (punctuation)","score":0.4165000021457672},{"id":"https://openalex.org/keywords/multi-task-learning","display_name":"Multi-task learning","score":0.4097999930381775}],"concepts":[{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.8165000081062317},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7669000029563904},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5861999988555908},{"id":"https://openalex.org/C150899416","wikidata":"https://www.wikidata.org/wiki/Q1820378","display_name":"Transfer of learning","level":2,"score":0.5157999992370605},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.4936999976634979},{"id":"https://openalex.org/C2780312720","wikidata":"https://www.wikidata.org/wiki/Q5689100","display_name":"Head (geology)","level":2,"score":0.47369998693466187},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.451200008392334},{"id":"https://openalex.org/C2778572836","wikidata":"https://www.wikidata.org/wiki/Q380933","display_name":"Space (punctuation)","level":2,"score":0.4165000021457672},{"id":"https://openalex.org/C28006648","wikidata":"https://www.wikidata.org/wiki/Q6934509","display_name":"Multi-task learning","level":3,"score":0.4097999930381775},{"id":"https://openalex.org/C175154964","wikidata":"https://www.wikidata.org/wiki/Q380077","display_name":"Task analysis","level":3,"score":0.3978999853134155},{"id":"https://openalex.org/C138777275","wikidata":"https://www.wikidata.org/wiki/Q6884054","display_name":"Mixing (physics)","level":2,"score":0.3910999894142151},{"id":"https://openalex.org/C2776175482","wikidata":"https://www.wikidata.org/wiki/Q1195816","display_name":"Transfer (computing)","level":2,"score":0.34769999980926514},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.32409998774528503},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.2736999988555908}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.18653/v1/2026.findings-acl.1355","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1355","pdf_url":"https://aclanthology.org/2026.findings-acl.1355.pdf","source":null,"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: ACL 2026","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2507.20906","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2507.20906","pdf_url":"https://arxiv.org/pdf/2507.20906","source":{"id":"https://openalex.org/S4393918464","display_name":"ArXiv.org","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"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.2507.20906","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2507.20906","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,"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":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.findings-acl.1355","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1355","pdf_url":"https://aclanthology.org/2026.findings-acl.1355.pdf","source":null,"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: ACL 2026","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G1317682478","display_name":null,"funder_award_id":"NO. RS-2021-II211343","funder_id":"https://openalex.org/F4320328359","funder_display_name":"Ministry of Science and ICT, South Korea"},{"id":"https://openalex.org/G269467657","display_name":null,"funder_award_id":"NO.RS-2021-II211343","funder_id":"https://openalex.org/F4320321292","funder_display_name":"Seoul National University"},{"id":"https://openalex.org/G2784508656","display_name":null,"funder_award_id":"NO.RS-2021-II211343","funder_id":"https://openalex.org/F4320335489","funder_display_name":"Institute for Information and Communications Technology Promotion"},{"id":"https://openalex.org/G2947324332","display_name":null,"funder_award_id":"RS-2021II211343","funder_id":"https://openalex.org/F4320335489","funder_display_name":"Institute for Information and Communications Technology Promotion"},{"id":"https://openalex.org/G4475359334","display_name":null,"funder_award_id":"RS2021-II211343","funder_id":"https://openalex.org/F4320321292","funder_display_name":"Seoul National University"},{"id":"https://openalex.org/G6120464043","display_name":null,"funder_award_id":"RS2021-II211343","funder_id":"https://openalex.org/F4320328359","funder_display_name":"Ministry of Science and ICT, South Korea"},{"id":"https://openalex.org/G7580349198","display_name":null,"funder_award_id":"RS-2023-00235293","funder_id":"https://openalex.org/F4320321292","funder_display_name":"Seoul National University"},{"id":"https://openalex.org/G8441597283","display_name":null,"funder_award_id":"RS-2023-","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":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4417271153.pdf","grobid_xml":"https://content.openalex.org/works/W4417271153.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Large":[0],"language":[1,98],"models":[2],"(LLMs)":[3],"are":[4],"commonly":[5],"adapted":[6],"to":[7,84,125],"downstream":[8],"tasks":[9],"using":[10,36,112],"parameterefficient":[11],"fine-tuning":[12],"(PEFT)":[13],"or":[14,57],"in-context":[15],"learning":[16],"(ICL).Recently,":[17],"ICL-driven":[18],"embeddingbased":[19],"adaptation":[20,28,106],"has":[21,49],"been":[22,64],"proposed":[23],"as":[24],"a":[25,76],"distinct":[26],"task":[27,87,149],"paradigm.It":[29],"derives":[30],"taskspecific":[31],"embeddings":[32],"from":[33,123],"intermediate":[34],"activations":[35],"few-shot":[37,109],"prompts":[38],"and":[39,59,96,108,134,136],"injects":[40],"them":[41],"during":[42],"inference.Despite":[43],"its":[44,60],"conceptual":[45],"appeal,":[46],"this":[47],"approach":[48],"not":[50],"demonstrated":[51],"consistent":[52],"performance":[53],"gains":[54],"over":[55],"PEFT":[56],"ICL,":[58,110],"empirical":[61],"advantages":[62],"have":[63],"limited":[65],"in":[66,151],"practice.We":[67],"propose":[68],"Soft":[69],"head-selection":[70],"for":[71],"ICLderived":[72],"Task":[73],"Embeddings":[74],"(SITE),":[75],"gradientbased":[77],"method":[78],"that":[79],"identifies":[80],"task-relevant":[81],"attention":[82,152],"heads":[83],"enable":[85],"effective":[86],"embedding":[88],"injection.Across":[89],"various":[90],"types":[91],"of":[92,131],"openended":[93],"generation,":[94],"reasoning,":[95],"natural":[97],"understanding":[99],"tasks,":[100],"SITE":[101],"significantly":[102],"outperforms":[103],"prior":[104],"embedding-based":[105],"methods":[107],"while":[111],"substantially":[113],"fewer":[114],"trainable":[115],"parameters":[116,127],"than":[117],"PEFT.Experiments":[118],"on":[119],"12":[120],"LLMs":[121],"ranging":[122],"4B":[124],"70B":[126],"demonstrate":[128],"the":[129],"generality":[130],"our":[132],"approach,":[133],"intra-task":[135],"inter-task":[137],"activation":[138],"patching":[139],"analyses":[140],"further":[141],"provide":[142],"new":[143],"mechanistic":[144],"insights":[145],"by":[146],"revealing":[147],"strong":[148],"dependence":[150],"head":[153],"functionality.":[154]},"counts_by_year":[],"updated_date":"2026-08-02T14:50:37.381335","created_date":"2025-10-10T00:00:00"}