{"id":"https://openalex.org/W7214923670","doi":"https://doi.org/10.48550/arxiv.2609.38066","title":"Alpha Diffusion Language Models: Factorization Alone Is Not the Problem","display_name":"Alpha Diffusion Language Models: Factorization Alone Is Not the Problem","publication_year":2026,"publication_date":"2026-09-29","ids":{"openalex":"https://openalex.org/W7214923670","doi":"https://doi.org/10.48550/arxiv.2609.38066"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2609.38066","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.38066","pdf_url":"https://arxiv.org/pdf/2609.38066","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"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2609.38066","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5085084315","display_name":"Nikita Gushchin","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Gushchin, Nikita","raw_affiliation_strings":["AXXX , Russia","Applied AI Institute , Moscow , Russia"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"AXXX , Russia","institution_ids":[]},{"raw_affiliation_string":"Applied AI Institute , Moscow , Russia","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5064657861","display_name":"Dmitry Baranchuk","orcid":"https://orcid.org/0000-0001-7660-3666","observed_orcids":["https://orcid.org/0000-0001-7660-3666"]},"institutions":[{"id":"https://openalex.org/I58957048","display_name":"Yandex (Russia)","ror":"https://ror.org/04dbch786","country_code":"RU","type":"company","lineage":["https://openalex.org/I58957048"]}],"countries":["RU"],"is_corresponding":false,"raw_author_name":"Baranchuk, Dmitry","raw_affiliation_strings":["Yandex Research"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Yandex Research","institution_ids":["https://openalex.org/I58957048"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5056141660","display_name":"Alexander Andreevich Korotin","orcid":"https://orcid.org/0000-0003-4286-925X","observed_orcids":["https://orcid.org/0000-0003-4286-925X","https://orcid.org/0000-0003-3773-342X"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Korotin, Alexander","raw_affiliation_strings":["AXXX , Russia","Applied AI Institute , Moscow , Russia"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"AXXX , Russia","institution_ids":[]},{"raw_affiliation_string":"Applied AI Institute , Moscow , Russia","institution_ids":[]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":2,"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/T10028","display_name":"Topic Modeling","score":0.7095000147819519,"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/T10028","display_name":"Topic Modeling","score":0.7095000147819519,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.2483000010251999,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.010999999940395355,"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/diffusion-language-models","display_name":"diffusion language models","score":0.8820000290870667},{"id":"https://openalex.org/keywords/code-generation","display_name":"code generation","score":0.4580000042915344},{"id":"https://openalex.org/keywords/mathematical-reasoning","display_name":"mathematical reasoning","score":0.6150000095367432}],"concepts":[{"id":"https://openalex.org/C187834632","wikidata":"https://www.wikidata.org/wiki/Q188804","display_name":"Factorization","level":2,"score":0.6814000010490417},{"id":"https://openalex.org/C64943373","wikidata":"https://www.wikidata.org/wiki/Q2651003","display_name":"Alpha (finance)","level":4,"score":0.6754999756813049},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5521000027656555},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.5241000056266785},{"id":"https://openalex.org/C151201525","wikidata":"https://www.wikidata.org/wiki/Q177239","display_name":"Limit (mathematics)","level":2,"score":0.5224999785423279},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.5013999938964844},{"id":"https://openalex.org/C69357855","wikidata":"https://www.wikidata.org/wiki/Q163214","display_name":"Diffusion","level":2,"score":0.4934000074863434},{"id":"https://openalex.org/C2780009758","wikidata":"https://www.wikidata.org/wiki/Q6804172","display_name":"Measure (data warehouse)","level":2,"score":0.43959999084472656},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.4169999957084656},{"id":"https://openalex.org/C2778755073","wikidata":"https://www.wikidata.org/wiki/Q10858537","display_name":"Scale (ratio)","level":2,"score":0.41040000319480896},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.4101000130176544},{"id":"https://openalex.org/C77618280","wikidata":"https://www.wikidata.org/wiki/Q1155772","display_name":"Scheme (mathematics)","level":2,"score":0.3549000024795532},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.3441999852657318},{"id":"https://openalex.org/C111335779","wikidata":"https://www.wikidata.org/wiki/Q3454686","display_name":"Reduction (mathematics)","level":2,"score":0.32339999079704285},{"id":"https://openalex.org/C28826006","wikidata":"https://www.wikidata.org/wiki/Q33521","display_name":"Applied mathematics","level":1,"score":0.3100999891757965},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.305400013923645},{"id":"https://openalex.org/C118505674","wikidata":"https://www.wikidata.org/wiki/Q42586063","display_name":"Encoder","level":2,"score":0.29809999465942383},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.29440000653266907},{"id":"https://openalex.org/C43126263","wikidata":"https://www.wikidata.org/wiki/Q128751","display_name":"Source code","level":2,"score":0.28519999980926514},{"id":"https://openalex.org/C118615104","wikidata":"https://www.wikidata.org/wiki/Q121416","display_name":"Discrete mathematics","level":1,"score":0.27489998936653137},{"id":"https://openalex.org/C2777299769","wikidata":"https://www.wikidata.org/wiki/Q3707858","display_name":"Type (biology)","level":2,"score":0.2615000009536743},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.2596000134944916},{"id":"https://openalex.org/C2778067643","wikidata":"https://www.wikidata.org/wiki/Q166507","display_name":"Interval (graph theory)","level":2,"score":0.25870001316070557}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2609.38066","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.38066","pdf_url":"https://arxiv.org/pdf/2609.38066","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.2609.38066","doi":"https://doi.org/10.48550/arxiv.2609.38066","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2609.38066","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:2609.38066","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.38066","pdf_url":"https://arxiv.org/pdf/2609.38066","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"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7214923670.pdf","grobid_xml":"https://content.openalex.org/works/W7214923670.grobid-xml"},"referenced_works_count":27,"referenced_works":["https://openalex.org/W2475118502","https://openalex.org/W2767206889","https://openalex.org/W3123553048","https://openalex.org/W3134642945","https://openalex.org/W3177813494","https://openalex.org/W3181640983","https://openalex.org/W3195140297","https://openalex.org/W4293052541","https://openalex.org/W4387964173","https://openalex.org/W4389821462","https://openalex.org/W4403751234","https://openalex.org/W4415331074","https://openalex.org/W4415796024","https://openalex.org/W4415800813","https://openalex.org/W4416048624","https://openalex.org/W4416125073","https://openalex.org/W4417529490","https://openalex.org/W6910504028","https://openalex.org/W7128776757","https://openalex.org/W7131391977","https://openalex.org/W7131421095","https://openalex.org/W7135156033","https://openalex.org/W7161018930","https://openalex.org/W7165810407","https://openalex.org/W7167599728","https://openalex.org/W7204465741","https://openalex.org/W7213294243"],"related_works":["https://openalex.org/W7161258637","https://openalex.org/W7220869571","https://openalex.org/W7220719280","https://openalex.org/W7133299406","https://openalex.org/W7117716201","https://openalex.org/W7160170245","https://openalex.org/W7213372004","https://openalex.org/W7160836971","https://openalex.org/W4404307915","https://openalex.org/W7083435283"],"abstract_inverted_index":{"Discrete":[0],"diffusion":[1,140],"language":[2,141],"models":[3],"can":[4,17,133],"generate":[5],"multiple":[6,87],"tokens":[7],"in":[8,52],"parallel,":[9],"but":[10],"reducing":[11],"the":[12,53,70,76,113,130,135],"number":[13],"of":[14,55,138],"denoising":[15],"steps":[16],"lead":[18],"to":[19,115],"inconsistent":[20],"predictions.":[21,35],"Standard":[22],"cross-entropy":[23,51],"training":[24,131],"fits":[25],"conditional":[26],"token":[27,93],"marginals,":[28],"whereas":[29],"parallel":[30],"generation":[31],"requires":[32],"consistent":[33],"joint":[34],"We":[36,79,110],"introduce":[37],"Alpha":[38],"Diffusion":[39],"Language":[40],"Models":[41],"(AlphaDLM),":[42],"trained":[43],"with":[44,105],"a":[45,60],"sequence-level":[46],"alpha":[47,57,64,85],"loss":[48],"that":[49,128],"recovers":[50],"limit":[54],"vanishing":[56],"and":[58,72,117,122],"has":[59],"joint-mode":[61],"optimum":[62],"at":[63],"one.":[65],"Our":[66],"analysis":[67],"characterizes":[68],"how":[69],"objective":[71,132],"factorization":[73],"jointly":[74],"determine":[75],"fitted":[77],"distribution.":[78],"identify":[80],"conditions":[81],"under":[82],"which":[83],"intermediate":[84],"preserves":[86],"valid":[88],"completions":[89],"while":[90],"excluding":[91],"invalid":[92],"combinations.":[94],"Trained":[95],"on":[96,103,120],"TinyGSM,":[97],"our":[98],"method":[99,114],"achieves":[100],"34.6%":[101],"accuracy":[102],"GSM8K":[104],"only":[106],"four":[107],"model":[108],"evaluations.":[109],"further":[111],"scale":[112],"SDAR-1.7B":[116],"evaluate":[118],"it":[119],"code":[121],"mathematics":[123],"benchmarks.":[124],"These":[125],"results":[126],"show":[127],"changing":[129],"improve":[134],"accuracy-computation":[136],"trade-off":[137],"factorized":[139],"models.":[142]},"counts_by_year":[{"year":2026,"cited_by_count":2}],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2026-10-01T00:00:00"}