{"id":"https://openalex.org/W4414809307","doi":"https://doi.org/10.48550/arxiv.2510.00810","title":"Family Matters: Language Transfer and Merging for Adapting Small LLMs to Faroese","display_name":"Family Matters: Language Transfer and Merging for Adapting Small LLMs to Faroese","publication_year":2025,"publication_date":"2025-10-01","ids":{"openalex":"https://openalex.org/W4414809307","doi":"https://doi.org/10.48550/arxiv.2510.00810"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2510.00810","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2510.00810","pdf_url":"https://arxiv.org/pdf/2510.00810","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-sa","license_id":"https://openalex.org/licenses/cc-by-sa","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/2510.00810","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":null,"display_name":"Kunz, Jenny","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I102134673","display_name":"Link\u00f6ping University","ror":"https://ror.org/05ynxx418","country_code":"SE","type":"education","lineage":["https://openalex.org/I102134673"]}],"countries":["SE"],"is_corresponding":false,"raw_author_name":"Kunz, Jenny","raw_affiliation_strings":["Link\u00f6ping University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Link\u00f6ping University","institution_ids":["https://openalex.org/I102134673"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5076535318","display_name":"Iben Nyholm Debess","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I103146751","display_name":"University of the Faroe Islands","ror":"https://ror.org/05mwmd090","country_code":"FO","type":"education","lineage":["https://openalex.org/I103146751"]}],"countries":["FO"],"is_corresponding":false,"raw_author_name":"Debess, Iben Nyholm","raw_affiliation_strings":["University of the Faroe Islands"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of the Faroe Islands","institution_ids":["https://openalex.org/I103146751"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5037337204","display_name":"Annika Simonsen","orcid":"https://orcid.org/0009-0007-2565-6152","observed_orcids":["https://orcid.org/0009-0007-2565-6152"]},"institutions":[{"id":"https://openalex.org/I165368041","display_name":"University of Iceland","ror":"https://ror.org/01db6h964","country_code":"IS","type":"education","lineage":["https://openalex.org/I165368041"]}],"countries":["IS"],"is_corresponding":false,"raw_author_name":"Simonsen, Annika","raw_affiliation_strings":["University of Iceland"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Iceland","institution_ids":["https://openalex.org/I165368041"]}]}],"institutions":[],"countries_distinct_count":3,"institutions_distinct_count":3,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":0,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.5268999934196472,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.5268999934196472,"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":0.4050000011920929,"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":0.05270000174641609,"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/faroese-language","display_name":"Faroese language","score":0.8349999785423279},{"id":"https://openalex.org/keywords/low-resource-languages","display_name":"low-resource languages","score":0.765999972820282},{"id":"https://openalex.org/keywords/small-language-models","display_name":"small language models","score":0.7480000257492065},{"id":"https://openalex.org/keywords/language-transfer","display_name":"language transfer","score":0.5910000205039978},{"id":"https://openalex.org/keywords/model-merging","display_name":"model merging","score":0.8500000238418579},{"id":"https://openalex.org/keywords/lora","display_name":"LoRa","score":0.8330000042915344},{"id":"https://openalex.org/keywords/parameter-efficient-fine-tuning","display_name":"parameter-efficient fine-tuning","score":0.8190000057220459},{"id":"https://openalex.org/keywords/full-fine-tuning","display_name":"full fine-tuning","score":0.8790000081062317},{"id":"https://openalex.org/keywords/nordic-languages","display_name":"Scandinavian languages","score":0.6669999957084656},{"id":"https://openalex.org/keywords/icelandic","display_name":"Icelandic language","score":0.8460000157356262},{"id":"https://openalex.org/keywords/danish-language","display_name":"Danish language","score":0.9810000061988831},{"id":"https://openalex.org/keywords/linguistic-acceptability","display_name":"linguistic acceptability","score":0.9470000267028809},{"id":"https://openalex.org/keywords/text-comprehension","display_name":"text comprehension","score":0.9200000166893005},{"id":"https://openalex.org/keywords/human-evaluation","display_name":"human evaluation","score":0.5220000147819519},{"id":"https://openalex.org/keywords/multilingual-pretraining","display_name":"multilingual pre-training","score":0.5070000290870667}],"concepts":[{"id":"https://openalex.org/C2776957530","wikidata":"https://www.wikidata.org/wiki/Q294","display_name":"Icelandic","level":2,"score":0.715399980545044},{"id":"https://openalex.org/C112313634","wikidata":"https://www.wikidata.org/wiki/Q7886648","display_name":"Complement (music)","level":5,"score":0.5942000150680542},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5591999888420105},{"id":"https://openalex.org/C41895202","wikidata":"https://www.wikidata.org/wiki/Q8162","display_name":"Linguistics","level":1,"score":0.5044999718666077},{"id":"https://openalex.org/C2780801425","wikidata":"https://www.wikidata.org/wiki/Q5164392","display_name":"Construct (python library)","level":2,"score":0.5016000270843506},{"id":"https://openalex.org/C164622146","wikidata":"https://www.wikidata.org/wiki/Q9035","display_name":"Danish","level":2,"score":0.46380001306533813},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.45509999990463257},{"id":"https://openalex.org/C205649164","wikidata":"https://www.wikidata.org/wiki/Q1071","display_name":"Geography","level":0,"score":0.445499986410141},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.4065000116825104},{"id":"https://openalex.org/C2993724205","wikidata":"https://www.wikidata.org/wiki/Q315","display_name":"Human language","level":2,"score":0.4036000072956085},{"id":"https://openalex.org/C2776175482","wikidata":"https://www.wikidata.org/wiki/Q1195816","display_name":"Transfer (computing)","level":2,"score":0.38989999890327454},{"id":"https://openalex.org/C511192102","wikidata":"https://www.wikidata.org/wiki/Q5156948","display_name":"Comprehension","level":2,"score":0.38589999079704285},{"id":"https://openalex.org/C95457728","wikidata":"https://www.wikidata.org/wiki/Q309","display_name":"History","level":0,"score":0.3140000104904175},{"id":"https://openalex.org/C171041071","wikidata":"https://www.wikidata.org/wiki/Q36870","display_name":"First language","level":2,"score":0.31279999017715454},{"id":"https://openalex.org/C42058472","wikidata":"https://www.wikidata.org/wiki/Q810214","display_name":"Base (topology)","level":2,"score":0.2948000133037567},{"id":"https://openalex.org/C98184364","wikidata":"https://www.wikidata.org/wiki/Q1780131","display_name":"Argument (complex analysis)","level":2,"score":0.2615000009536743},{"id":"https://openalex.org/C59985594","wikidata":"https://www.wikidata.org/wiki/Q1758140","display_name":"Contrastive linguistics","level":3,"score":0.26100000739097595},{"id":"https://openalex.org/C150899416","wikidata":"https://www.wikidata.org/wiki/Q1820378","display_name":"Transfer of learning","level":2,"score":0.25679999589920044}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2510.00810","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2510.00810","pdf_url":"https://arxiv.org/pdf/2510.00810","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-sa","license_id":"https://openalex.org/licenses/cc-by-sa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2510.00810","doi":"https://doi.org/10.48550/arxiv.2510.00810","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2510.00810","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":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2510.00810","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2510.00810","pdf_url":"https://arxiv.org/pdf/2510.00810","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-sa","license_id":"https://openalex.org/licenses/cc-by-sa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"sustainable_development_goals_aurora":[],"study_designs":[],"awards":[{"id":"https://openalex.org/G3800857974","display_name":"Democratize Trustworthy and Efficient Large Language Model Technology for Europe","funder_award_id":"101135671","funder_id":"https://openalex.org/F4320320300","funder_display_name":"European Commission"},{"id":"https://openalex.org/G7318745325","display_name":null,"funder_award_id":"2022-06725","funder_id":"https://openalex.org/F4320336380","funder_display_name":"National Supercomputer Centre, Link\u00f6pings Universitet"},{"id":"https://openalex.org/G8869864247","display_name":"NAISS","funder_award_id":"2022-06725","funder_id":"https://openalex.org/F4320322581","funder_display_name":"Vetenskapsr\u00e5det"}],"funders":[{"id":"https://openalex.org/F4320320300","display_name":"European Commission","ror":"https://ror.org/00k4n6c32"},{"id":"https://openalex.org/F4320322327","display_name":"Knut och Alice Wallenbergs Stiftelse","ror":"https://ror.org/004hzzk67"},{"id":"https://openalex.org/F4320322581","display_name":"Vetenskapsr\u00e5det","ror":"https://ror.org/03zttf063"},{"id":"https://openalex.org/F4320334322","display_name":"HORIZON EUROPE Framework Programme","ror":null},{"id":"https://openalex.org/F4320336380","display_name":"National Supercomputer Centre, Link\u00f6pings Universitet","ror":"https://ror.org/054wejp28"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4414809307.pdf","grobid_xml":"https://content.openalex.org/works/W4414809307.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W7009588143","https://openalex.org/W7015924724","https://openalex.org/W4412889440","https://openalex.org/W7166513614","https://openalex.org/W4404993448","https://openalex.org/W4407569751","https://openalex.org/W7140081364","https://openalex.org/W7164192949","https://openalex.org/W7164553850","https://openalex.org/W4388650715"],"abstract_inverted_index":{"We":[0,40],"investigate":[1],"strategies":[2],"for":[3,79,84],"adapting":[4],"small,":[5],"efficient":[6],"language":[7,54,112,170],"models":[8],"to":[9],"Faroese,":[10],"a":[11],"low-resource":[12],"North":[13],"Germanic":[14],"language.":[15],"Starting":[16],"from":[17,103],"English-pretrained":[18],"models,":[19],"we":[20,71],"apply":[21],"continued":[22],"pre-training":[23],"on":[24,38,52,131],"related":[25,104,160],"Scandinavian":[26],"languages":[27,105,161],"--":[28,35],"individually":[29],"or":[30],"combined":[31],"via":[32,47],"model":[33],"merging":[34],"before":[36],"fine-tuning":[37,43,148,164],"Faroese.":[39],"compare":[41],"full":[42,147,163],"with":[44,90],"parameter-efficient":[45],"adaptation":[46,127],"LoRA,":[48],"assessing":[49],"their":[50],"effects":[51],"general":[53,169],"modeling":[55],"performance,":[56],"linguistic":[57,80,117,138,177],"accuracy,":[58,118],"and":[59,82,87,140,153,179],"text":[60,85],"comprehension.":[61,123],"To":[62],"address":[63],"the":[64,109,132,176],"lack":[65],"of":[66,126],"existing":[67],"Faroese":[68,96],"evaluation":[69,144],"resources,":[70],"construct":[72],"two":[73],"new":[74],"minimal-pair":[75],"probing":[76],"benchmarks,":[77],"one":[78,83],"acceptability":[81,139,178],"comprehension,":[86],"complement":[88],"them":[89],"human":[91,143],"evaluations":[92],"conducted":[93],"by":[94],"native":[95],"linguists.":[97],"Our":[98],"results":[99],"show":[100],"that":[101],"transfer":[102],"is":[106,113],"essential,":[107],"but":[108],"optimal":[110],"source":[111],"task-dependent:":[114],"Icelandic":[115],"improves":[116,168],"while":[119],"Danish":[120],"boosts":[121],"reading":[122],"The":[124],"choice":[125],"method":[128],"likewise":[129],"depends":[130],"target":[133],"task:":[134],"LoRA":[135],"yields":[136],"stronger":[137],"marginally":[141],"higher":[142],"scores,":[145],"whereas":[146],"produces":[149],"better":[150],"comprehension":[151,180],"performance":[152],"more":[154],"robust":[155],"downstream":[156],"fine-tuning.":[157],"Merging":[158],"multiple":[159],"under":[162],"(but":[165],"not":[166],"LoRA)":[167],"modeling,":[171],"though":[172],"its":[173],"benefits":[174],"in":[175],"probes":[181],"are":[182],"less":[183],"consistent.":[184]},"counts_by_year":[],"updated_date":"2026-09-30T09:27:27.532063","created_date":"2025-10-10T00:00:00"}