{"id":"https://openalex.org/W7133359396","doi":"https://doi.org/10.48550/arxiv.2603.02041","title":"EstLLM: Enhancing Estonian Capabilities in Multilingual LLMs via Continued Pretraining and Post-Training","display_name":"EstLLM: Enhancing Estonian Capabilities in Multilingual LLMs via Continued Pretraining and Post-Training","publication_year":2026,"publication_date":"2026-03-02","ids":{"openalex":"https://openalex.org/W7133359396","doi":"https://doi.org/10.48550/arxiv.2603.02041"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2603.02041","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.02041","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"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":false,"raw_source_name":null,"raw_type":"article"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2603.02041","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5095875312","display_name":"Aleksei Dorkin","orcid":null},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Dorkin, Aleksei","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5005106713","display_name":"Taido Purason","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Purason, Taido","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5016111855","display_name":"Emil Kalbaliyev","orcid":"https://orcid.org/0000-0003-0704-0470"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kalbaliyev, Emil","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5127965157","display_name":"Hele-Andra Kuulmets","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kuulmets, Hele-Andra","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5120470494","display_name":"Marii Ojastu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ojastu, Marii","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5063503892","display_name":"Mark Fi\u0161el","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Fi\u0161el, Mark","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5122885119","display_name":"Tanel Alum\u00e4e","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Alum\u00e4e, Tanel","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5127919413","display_name":"Eleri Aedmaa","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Aedmaa, Eleri","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5127985572","display_name":"Krister Kruusmaa","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kruusmaa, Krister","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5035358986","display_name":"Kairit Sirts","orcid":"https://orcid.org/0000-0001-7388-2583"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sirts, Kairit","raw_affiliation_strings":[],"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":10,"corresponding_author_ids":["https://openalex.org/A5095875312"],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"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":0.4300999939441681,"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":0.4300999939441681,"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":0.2167000025510788,"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/T14347","display_name":"Big Data and Digital Economy","score":0.03550000116229057,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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/estonian","display_name":"Estonian","score":0.9800999760627747},{"id":"https://openalex.org/keywords/suite","display_name":"Suite","score":0.482699990272522},{"id":"https://openalex.org/keywords/preference","display_name":"Preference","score":0.47999998927116394},{"id":"https://openalex.org/keywords/base","display_name":"Base (topology)","score":0.423799991607666},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.4002000093460083},{"id":"https://openalex.org/keywords/training-set","display_name":"Training set","score":0.39160001277923584},{"id":"https://openalex.org/keywords/translation","display_name":"Translation (biology)","score":0.3880999982357025}],"concepts":[{"id":"https://openalex.org/C2776092919","wikidata":"https://www.wikidata.org/wiki/Q9072","display_name":"Estonian","level":2,"score":0.9800999760627747},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6403999924659729},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5491999983787537},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.49639999866485596},{"id":"https://openalex.org/C79581498","wikidata":"https://www.wikidata.org/wiki/Q1367530","display_name":"Suite","level":2,"score":0.482699990272522},{"id":"https://openalex.org/C2781249084","wikidata":"https://www.wikidata.org/wiki/Q908656","display_name":"Preference","level":2,"score":0.47999998927116394},{"id":"https://openalex.org/C42058472","wikidata":"https://www.wikidata.org/wiki/Q810214","display_name":"Base (topology)","level":2,"score":0.423799991607666},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.4002000093460083},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.39160001277923584},{"id":"https://openalex.org/C41895202","wikidata":"https://www.wikidata.org/wiki/Q8162","display_name":"Linguistics","level":1,"score":0.3896999955177307},{"id":"https://openalex.org/C149364088","wikidata":"https://www.wikidata.org/wiki/Q185917","display_name":"Translation (biology)","level":4,"score":0.3880999982357025},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.3465999960899353},{"id":"https://openalex.org/C2777868144","wikidata":"https://www.wikidata.org/wiki/Q7239817","display_name":"Preference elicitation","level":3,"score":0.3248000144958496},{"id":"https://openalex.org/C189430467","wikidata":"https://www.wikidata.org/wiki/Q7293293","display_name":"Ranking (information retrieval)","level":2,"score":0.32030001282691956},{"id":"https://openalex.org/C110121322","wikidata":"https://www.wikidata.org/wiki/Q865811","display_name":"Distribution (mathematics)","level":2,"score":0.3041999936103821},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.28700000047683716},{"id":"https://openalex.org/C109359841","wikidata":"https://www.wikidata.org/wiki/Q728944","display_name":"Inclusion (mineral)","level":2,"score":0.2815999984741211},{"id":"https://openalex.org/C2987496018","wikidata":"https://www.wikidata.org/wiki/Q1860","display_name":"English language","level":2,"score":0.2648000121116638}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2603.02041","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.02041","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"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":"article"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2603.02041","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.02041","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"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":false,"raw_source_name":null,"raw_type":"article"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/4","display_name":"Quality Education","score":0.7988423705101013}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Large":[0],"language":[1],"models":[2],"(LLMs)":[3],"are":[4],"predominantly":[5],"trained":[6],"on":[7,53,95,128],"English-centric":[8],"data,":[9],"resulting":[10],"in":[11,28,105,151],"uneven":[12],"performance":[13,127],"for":[14],"smaller":[15],"languages.":[16],"We":[17,78],"study":[18],"whether":[19],"continued":[20],"pretraining":[21],"(CPT)":[22],"can":[23,146],"substantially":[24,147],"improve":[25,148],"Estonian":[26,58,100],"capabilities":[27,150],"a":[29,54,96],"pretrained":[30,152],"multilingual":[31,153],"LLM":[32],"while":[33,60,124],"preserving":[34],"its":[35,121],"English":[36,67,129],"and":[37,69,75,85,112,120],"general":[38],"reasoning":[39],"performance.":[40],"Using":[41],"Llama":[42],"3.1":[43],"8B":[44],"as":[45],"the":[46,62,70,116],"main":[47],"base":[48,118],"model,":[49],"we":[50],"perform":[51],"CPT":[52],"mixture":[55],"that":[56,134],"increases":[57],"exposure":[59],"approximating":[61],"original":[63,117],"training":[64],"distribution":[65],"through":[66],"replay":[68],"inclusion":[71],"of":[72,99],"code,":[73],"mathematics,":[74],"instruction-like":[76],"data.":[77],"subsequently":[79],"apply":[80],"supervised":[81],"fine-tuning,":[82],"preference":[83],"optimization,":[84],"chat":[86],"vector":[87],"merging":[88],"to":[89,115],"introduce":[90],"robust":[91],"instruction-following":[92,113],"behavior.":[93],"Evaluation":[94],"comprehensive":[97],"suite":[98],"benchmarks":[101],"shows":[102],"consistent":[103],"gains":[104],"linguistic":[106],"competence,":[107],"knowledge,":[108],"reasoning,":[109],"translation":[110],"quality,":[111],"compared":[114],"model":[119],"instruction-tuned":[122],"variant,":[123],"maintaining":[125],"competitive":[126],"benchmarks.":[130],"These":[131],"findings":[132],"indicate":[133],"CPT,":[135],"with":[136,143],"an":[137],"appropriately":[138],"balanced":[139],"data":[140],"mixture,":[141],"together":[142],"post-training":[144],"alignment,":[145],"single-language":[149],"LLMs.":[154]},"counts_by_year":[],"updated_date":"2026-03-04T07:09:34.246503","created_date":"2026-03-04T00:00:00"}