{"id":"https://openalex.org/W7164956703","doi":"https://doi.org/10.48550/arxiv.2606.15327","title":"Semantic DLM+: Improving Diffusion Language Models through Bias-variance Trade-off in Transition Kernel Design","display_name":"Semantic DLM+: Improving Diffusion Language Models through Bias-variance Trade-off in Transition Kernel Design","publication_year":2026,"publication_date":"2026-06-13","ids":{"openalex":"https://openalex.org/W7164956703","doi":"https://doi.org/10.48550/arxiv.2606.15327"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.15327","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.15327","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2606.15327","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5112546231","display_name":"Keyue Jiang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jiang, Keyue","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138717655","display_name":"Yuxiang Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Yuxiang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5076601802","display_name":"Yanan Zhao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhao, Yanan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138756604","display_name":"Xiang Yu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yu, Xiang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138758071","display_name":"Qifang Zhao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhao, Qifang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5107380052","display_name":"B Tang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tang, Bohan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138757456","display_name":"Baojian Zhou","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhou, Baojian","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5133363812","display_name":"Y H Xiao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xiao, Yanghua","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138732048","display_name":"Lin Qu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Qu, Lin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5138713268","display_name":"Xiaoxiao Xu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xu, Xiaoxiao","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":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/T10028","display_name":"Topic Modeling","score":0.4147999882698059,"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":0.4147999882698059,"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/T13910","display_name":"Computational and Text Analysis Methods","score":0.10409999638795853,"subfield":{"id":"https://openalex.org/subfields/3300","display_name":"General Social Sciences"},"field":{"id":"https://openalex.org/fields/33","display_name":"Social Sciences"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},{"id":"https://openalex.org/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.094200000166893,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/language-model","display_name":"Language model","score":0.6484000086784363},{"id":"https://openalex.org/keywords/scaling","display_name":"Scaling","score":0.5848000049591064},{"id":"https://openalex.org/keywords/generalization","display_name":"Generalization","score":0.5796999931335449},{"id":"https://openalex.org/keywords/diffusion","display_name":"Diffusion","score":0.4909000098705292},{"id":"https://openalex.org/keywords/kernel","display_name":"Kernel (algebra)","score":0.4867999851703644},{"id":"https://openalex.org/keywords/variance","display_name":"Variance (accounting)","score":0.4645000100135803},{"id":"https://openalex.org/keywords/masking","display_name":"Masking (illustration)","score":0.42179998755455017}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7005000114440918},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.6484000086784363},{"id":"https://openalex.org/C99844830","wikidata":"https://www.wikidata.org/wiki/Q102441924","display_name":"Scaling","level":2,"score":0.5848000049591064},{"id":"https://openalex.org/C177148314","wikidata":"https://www.wikidata.org/wiki/Q170084","display_name":"Generalization","level":2,"score":0.5796999931335449},{"id":"https://openalex.org/C69357855","wikidata":"https://www.wikidata.org/wiki/Q163214","display_name":"Diffusion","level":2,"score":0.4909000098705292},{"id":"https://openalex.org/C74193536","wikidata":"https://www.wikidata.org/wiki/Q574844","display_name":"Kernel (algebra)","level":2,"score":0.4867999851703644},{"id":"https://openalex.org/C196083921","wikidata":"https://www.wikidata.org/wiki/Q7915758","display_name":"Variance (accounting)","level":2,"score":0.4645000100135803},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.4584999978542328},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.43290001153945923},{"id":"https://openalex.org/C2777402240","wikidata":"https://www.wikidata.org/wiki/Q6783436","display_name":"Masking (illustration)","level":2,"score":0.42179998755455017},{"id":"https://openalex.org/C194232998","wikidata":"https://www.wikidata.org/wiki/Q1606712","display_name":"Transition (genetics)","level":3,"score":0.40610000491142273},{"id":"https://openalex.org/C140779682","wikidata":"https://www.wikidata.org/wiki/Q210868","display_name":"Sampling (signal processing)","level":3,"score":0.3352000117301941},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.320499986410141},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.2973000109195709},{"id":"https://openalex.org/C52740198","wikidata":"https://www.wikidata.org/wiki/Q1539564","display_name":"Importance sampling","level":3,"score":0.28769999742507935},{"id":"https://openalex.org/C184337299","wikidata":"https://www.wikidata.org/wiki/Q1437428","display_name":"Semantics (computer science)","level":2,"score":0.2702000141143799},{"id":"https://openalex.org/C159877910","wikidata":"https://www.wikidata.org/wiki/Q2202883","display_name":"Autoregressive model","level":2,"score":0.26179999113082886},{"id":"https://openalex.org/C2779439875","wikidata":"https://www.wikidata.org/wiki/Q1078276","display_name":"Natural language understanding","level":3,"score":0.2572999894618988},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.25609999895095825},{"id":"https://openalex.org/C2777601683","wikidata":"https://www.wikidata.org/wiki/Q6499736","display_name":"Vocabulary","level":2,"score":0.25440001487731934}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.15327","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.15327","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2606.15327","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.15327","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/4","display_name":"Quality Education","score":0.49586912989616394}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Diffusion":[0],"Language":[1],"Models":[2],"(DLMs)":[3],"have":[4],"demonstrated":[5],"strong":[6],"scaling":[7],"capacity":[8],"as":[9,140],"alternatives":[10],"to":[11,21,33,127],"autoregressive":[12],"language":[13,212],"models.":[14],"However,":[15,158],"their":[16],"performance":[17],"is":[18],"highly":[19],"sensitive":[20],"the":[22,67,122,147],"choice":[23],"of":[24,54,151],"transition":[25,87,123,191],"kernels,":[26],"and":[27,40,57,76,93,177,192,201,209,214],"poorly":[28],"designed":[29],"kernels":[30],"can":[31,138],"lead":[32],"issues":[34],"like":[35],"training":[36,207],"instability,":[37],"slow":[38],"convergence,":[39],"biased":[41],"sampling.":[42,197],"In":[43],"this":[44,48,110],"paper,":[45],"we":[46,112,159,184],"study":[47],"sensitivity":[49],"through":[50],"a":[51,114,141,165,174,189,193],"principled":[52],"analysis":[53],"generalization":[55],"error":[56],"identify":[58],"three":[59],"critical":[60],"factors:":[61],"asymptotic":[62],"bias":[63,71],"(difficulty":[64],"in":[65],"approximating":[66],"posterior":[68,148],"distribution),":[69],"exposure":[70],"(error":[72],"propagation":[73],"during":[74,196],"sampling),":[75],"optimization":[77],"variance":[78],"induced":[79],"by":[80,109,145],"kernel":[81,124],"dispersion.":[82],"We":[83],"further":[84],"compare":[85],"different":[86],"kernels:":[88],"masking":[89],"diffusion":[90,99,153],"yields":[91],"sparse":[92],"easier":[94],"posterior-approximation":[95],"targets,":[96],"while":[97,154],"uniform":[98,152],"provides":[100],"stronger":[101],"sampling-side":[102],"repair":[103,156],"but":[104],"induces":[105],"harder":[106],"approximation.":[107],"Motivated":[108],"trade-off,":[111],"revisit":[113],"previously":[115],"overlooked":[116],"variant,":[117],"semantic":[118,166,175],"DLM":[119],"(SemDLM),":[120],"where":[121,169],"corrupts":[125],"tokens":[126],"neighborhoods":[128],"that":[129,136,161,204],"are":[130],"semantically":[131],"similar.":[132],"Our":[133],"theory":[134],"suggests":[135],"SemDLM":[137,162],"serve":[139],"plausible":[142],"middle":[143],"ground":[144],"reducing":[146],"approximation":[149],"difficulty":[150],"retaining":[155],"ability.":[157],"find":[160],"suffers":[163],"from":[164],"basin":[167],"problem,":[168],"sampling":[170],"repeatedly":[171],"stays":[172],"within":[173],"region":[176],"produces":[178],"low-diversity":[179],"text.":[180],"To":[181],"address":[182],"this,":[183],"propose":[185],"SemDLM+,":[186],"which":[187],"adds":[188],"global":[190],"semantic-frequency":[194],"penalty":[195],"Experiments":[198],"on":[199],"LM1B":[200],"OpenWebText":[202],"show":[203],"SemDLM+":[205],"improves":[206],"dynamics":[208],"achieves":[210],"competitive":[211],"modeling":[213],"generation":[215],"quality":[216],"with":[217],"satisfactory":[218],"diversity.":[219]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-06-17T00:00:00"}