{"id":"https://openalex.org/W7165661666","doi":"https://doi.org/10.48550/arxiv.2606.22511","title":"Breaking the Likelihood Trap: Variance-Calibrated Modulation for Large Language Model Decoding","display_name":"Breaking the Likelihood Trap: Variance-Calibrated Modulation for Large Language Model Decoding","publication_year":2026,"publication_date":"2026-06-21","ids":{"openalex":"https://openalex.org/W7165661666","doi":"https://doi.org/10.48550/arxiv.2606.22511"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.22511","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.22511","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,"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":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2606.22511","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5139136778","display_name":"Yuanhao Ding","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ding, Yuanhao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5113478345","display_name":"Meimingwei Li","orcid":"https://orcid.org/0009-0003-1400-562X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Meimingwei","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5114076222","display_name":"Esteban Garces Arias","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Arias, Esteban Garces","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139213785","display_name":"Matthias A\u00dfenmacher","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"A\u00dfenmacher, Matthias","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135809939","display_name":"Christian Heumann","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Heumann, Christian","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5091464412","display_name":"Chongsheng Zhang","orcid":"https://orcid.org/0000-0003-1632-7238"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Chongsheng","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":0.646399974822998,"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":0.646399974822998,"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":0.09189999848604202,"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/T13629","display_name":"Text Readability and Simplification","score":0.06120000034570694,"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/decoding-methods","display_name":"Decoding methods","score":0.6571000218391418},{"id":"https://openalex.org/keywords/truncation","display_name":"Truncation (statistics)","score":0.5608000159263611},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.5277000069618225},{"id":"https://openalex.org/keywords/vocabulary","display_name":"Vocabulary","score":0.4805000126361847},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.43220001459121704},{"id":"https://openalex.org/keywords/probability-distribution","display_name":"Probability distribution","score":0.37049999833106995},{"id":"https://openalex.org/keywords/joint-probability-distribution","display_name":"Joint probability distribution","score":0.3571000099182129}],"concepts":[{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.6571000218391418},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5849000215530396},{"id":"https://openalex.org/C106195933","wikidata":"https://www.wikidata.org/wiki/Q7847935","display_name":"Truncation (statistics)","level":2,"score":0.5608000159263611},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.5277000069618225},{"id":"https://openalex.org/C2777601683","wikidata":"https://www.wikidata.org/wiki/Q6499736","display_name":"Vocabulary","level":2,"score":0.4805000126361847},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.43709999322891235},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.43220001459121704},{"id":"https://openalex.org/C149441793","wikidata":"https://www.wikidata.org/wiki/Q200726","display_name":"Probability distribution","level":2,"score":0.37049999833106995},{"id":"https://openalex.org/C18653775","wikidata":"https://www.wikidata.org/wiki/Q1333358","display_name":"Joint probability distribution","level":2,"score":0.3571000099182129},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3564999997615814},{"id":"https://openalex.org/C2778334786","wikidata":"https://www.wikidata.org/wiki/Q1586270","display_name":"Variation (astronomy)","level":2,"score":0.3546000123023987},{"id":"https://openalex.org/C140779682","wikidata":"https://www.wikidata.org/wiki/Q210868","display_name":"Sampling (signal processing)","level":3,"score":0.3400000035762787},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.33469998836517334},{"id":"https://openalex.org/C12713177","wikidata":"https://www.wikidata.org/wiki/Q1900281","display_name":"Perspective (graphical)","level":2,"score":0.33059999346733093},{"id":"https://openalex.org/C52740198","wikidata":"https://www.wikidata.org/wiki/Q1539564","display_name":"Importance sampling","level":3,"score":0.32829999923706055},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.266400009393692},{"id":"https://openalex.org/C2778755073","wikidata":"https://www.wikidata.org/wiki/Q10858537","display_name":"Scale (ratio)","level":2,"score":0.26330000162124634},{"id":"https://openalex.org/C2779899686","wikidata":"https://www.wikidata.org/wiki/Q571227","display_name":"Implicature","level":3,"score":0.2547000050544739},{"id":"https://openalex.org/C51167844","wikidata":"https://www.wikidata.org/wiki/Q4422623","display_name":"Latent variable","level":2,"score":0.2535000145435333},{"id":"https://openalex.org/C105795698","wikidata":"https://www.wikidata.org/wiki/Q12483","display_name":"Statistics","level":1,"score":0.25290000438690186}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.22511","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.22511","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,"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":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2606.22511","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.22511","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,"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":"Preprint"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/4","display_name":"Quality Education","score":0.6274456977844238}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"In":[0],"open-ended":[1,120],"generation,":[2,121],"LLMs":[3],"frequently":[4],"fall":[5],"into":[6],"the":[7,35,42,84,130],"\"likelihood":[8],"trap\",":[9],"marked":[10],"by":[11],"repetitive":[12],"degeneration":[13],"and":[14,22,45,106,124],"vocabulary":[15],"dullness,":[16],"creating":[17],"a":[18,78],"discrepancy":[19],"between":[20],"machine-generated":[21],"human-written":[23],"text.":[24],"While":[25],"post-hoc":[26],"tail":[27],"truncation":[28,88],"(e.g.,":[29],"Top-$p$,":[30],"Min-$p$)":[31],"avoids":[32],"sampling":[33],"from":[34,41],"unreliable":[36],"tail,":[37],"it":[38],"can":[39],"over-sample":[40],"uncalibrated":[43],"head":[44],"misalign":[46],"generation":[47],"with":[48,139],"human":[49],"lexical":[50],"preferences;":[51],"fixed":[52],"scalar":[53],"repetition":[54],"penalties":[55],"likewise":[56],"ignore":[57],"variation":[58],"in":[59],"logit":[60,113],"scale":[61],"across":[62],"inference":[63],"steps,":[64],"potentially":[65],"disrupting":[66],"semantic":[67],"coherence.":[68],"To":[69],"address":[70],"both":[71],"limitations,":[72],"we":[73],"propose":[74],"Variance-Calibrated":[75],"Modulation":[76],"(VCM),":[77],"training-free":[79],"pre-decoding":[80],"intervention":[81],"that":[82],"reshapes":[83],"probability":[85],"distribution":[86],"before":[87],"through":[89],"two":[90],"dynamic":[91],"mechanisms:":[92],"(1)":[93],"Contextual":[94],"Searchlight":[95],"via":[96],"PMI,":[97],"which":[98,110],"suppresses":[99],"global":[100],"stopwords":[101],"while":[102],"elevating":[103],"context-evoked":[104],"tokens,":[105],"(2)":[107],"Adaptive":[108],"Self-Debiasing,":[109],"uses":[111],"real-time":[112],"standard":[114],"deviation":[115],"for":[116],"scale-invariant":[117],"penalization.":[118],"Across":[119],"factual":[122],"QA,":[123],"mathematical":[125],"reasoning,":[126],"VCM":[127,137],"consistently":[128],"mitigates":[129],"likelihood":[131],"trap.":[132],"With":[133],"negligible":[134],"computational":[135],"overhead,":[136],"integrates":[138],"existing":[140],"decoding":[141,150],"strategies,":[142],"improving":[143],"diversity,":[144],"coherence,":[145],"and,":[146],"particularly":[147],"at":[148],"higher":[149],"temperatures,":[151],"reasoning":[152],"accuracy.":[153]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-06-24T00:00:00"}