{"id":"https://openalex.org/W7166848404","doi":"https://doi.org/10.18653/v1/2026.findings-acl.2068","title":"MorphBPE: Morphology-Aware Tokenization for Efficient LLM Training","display_name":"MorphBPE: Morphology-Aware Tokenization for Efficient LLM Training","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7166848404","doi":"https://doi.org/10.18653/v1/2026.findings-acl.2068"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2026.findings-acl.2068","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.2068","pdf_url":"https://aclanthology.org/2026.findings-acl.2068.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":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2026.findings-acl.2068.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5014316659","display_name":"Ehsaneddin Asgari","orcid":"https://orcid.org/0000-0002-6518-7238"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ehsaneddin Asgari","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139827771","display_name":"Yassine El Kheir","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yassine El Kheir","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5115772360","display_name":"MohammadAli SadraeiJavaheri","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"MohammadAli SadraeiJavaheri","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5139718220","display_name":"Ali Nazari","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ali Nazari","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":0.83817427,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"41610","last_page":"41621"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.5412999987602234,"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":0.5412999987602234,"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/T12535","display_name":"Machine Learning and Data Classification","score":0.08389999717473984,"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":0.0560000017285347,"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","display_name":"Training (meteorology)","score":0.41830000281333923},{"id":"https://openalex.org/keywords/troubleshooting","display_name":"Troubleshooting","score":0.29280000925064087},{"id":"https://openalex.org/keywords/training-set","display_name":"Training set","score":0.27709999680519104},{"id":"https://openalex.org/keywords/feature","display_name":"Feature (linguistics)","score":0.25519999861717224},{"id":"https://openalex.org/keywords/lexical-analysis","display_name":"Lexical analysis","score":0.24230000376701355}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6942999958992004},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5134000182151794},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.4336000084877014},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.41830000281333923},{"id":"https://openalex.org/C147494362","wikidata":"https://www.wikidata.org/wiki/Q2078905","display_name":"Troubleshooting","level":2,"score":0.29280000925064087},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.27709999680519104},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.25519999861717224},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.243599995970726},{"id":"https://openalex.org/C176982825","wikidata":"https://www.wikidata.org/wiki/Q835922","display_name":"Lexical analysis","level":2,"score":0.24230000376701355},{"id":"https://openalex.org/C2780791683","wikidata":"https://www.wikidata.org/wiki/Q846785","display_name":"Action (physics)","level":2,"score":0.23229999840259552}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2026.findings-acl.2068","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.2068","pdf_url":"https://aclanthology.org/2026.findings-acl.2068.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"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.findings-acl.2068","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.2068","pdf_url":"https://aclanthology.org/2026.findings-acl.2068.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":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7166848404.pdf","grobid_xml":"https://content.openalex.org/works/W7166848404.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Tokenization":[0],"fundamentally":[1],"shapes":[2],"NLP":[3],"performance,":[4],"affecting":[5],"both":[6],"efficiency":[7],"and":[8,30,60,87,100,115,121,132,170],"linguistic":[9],"fidelity.While":[10],"Byte":[11],"Pair":[12],"Encoding":[13],"(BPE)":[14],"underpins":[15],"most":[16],"Large":[17],"Language":[18],"Models":[19],"(LLMs),":[20],"its":[21],"frequency-driven":[22],"merges":[23],"often":[24],"disregard":[25],"morpheme":[26,54,95],"boundaries,":[27,55],"yielding":[28],"inconsistent":[29],"semantically":[31],"opaque":[32],"segmentations":[33],"in":[34,163],"morphologically":[35,164],"rich":[36,165],"languages.We":[37],"introduce":[38],"MorphBPE,":[39],"a":[40],"simple":[41],"extension":[42],"of":[43],"BPE":[44],"that":[45,142],"constrains":[46],"merge":[47],"operations":[48],"during":[49],"tokenizer":[50],"training":[51,122],"to":[52,148],"respect":[53],"while":[56],"leaving":[57],"inference":[58],"unchanged":[59],"fully":[61],"compatible":[62],"with":[63,94],"existing":[64],"LLM":[65],"pipelines.We":[66],"evaluate":[67],"tokenization":[68],"quality":[69],"using":[70],"two":[71],"intrinsic":[72,129],"metrics:":[73],"Morphological":[74,88],"Consistency":[75],"F1,":[76],"which":[77,91],"measures":[78],"whether":[79],"shared":[80],"morphemes":[81],"are":[82,145],"assigned":[83],"consistent":[84],"token":[85,138],"representations,":[86],"Edit":[89],"Distance,":[90],"quantifies":[92],"alignment":[93],"boundaries.We":[96],"then":[97],"train":[98],"300M":[99],"1B":[101],"parameter":[102],"decoderonly":[103],"LMs":[104],"from":[105],"scratch":[106],"across":[107],"four":[108],"typologically":[109],"diverse":[110],"languages:":[111],"English,":[112],"Russian,":[113],"Hungarian,":[114],"Arabic,":[116],"under":[117],"identical":[118],"vocabulary":[119],"sizes":[120],"settings.Across":[123],"all":[124],"languages,":[125],"Mor-phBPE":[126],"consistently":[127],"improves":[128],"morphological":[130],"coherence":[131],"reduces":[133],"language":[134],"model":[135],"cross-entropy;":[136],"moreover,":[137],"length":[139],"statistics":[140],"indicate":[141],"these":[143],"gains":[144],"not":[146],"attributable":[147],"materially":[149],"shorter":[150],"tokens.Finally,":[151],"on":[152],"the":[153],"Belebele":[154],"multilingual":[155],"reading":[156],"comprehension":[157],"benchmark,":[158],"MorphBPE":[159],"yields":[160],"significant":[161],"improvements":[162],"languages":[166],"such":[167],"as":[168],"Russian":[169],"Arabic.":[171]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2026-07-02T00:00:00"}