{"id":"https://openalex.org/W7133601568","doi":"https://doi.org/10.48550/arxiv.2603.03333","title":"Training-free Dropout Sampling for Semantic Token Acceptance in Speculative Decoding","display_name":"Training-free Dropout Sampling for Semantic Token Acceptance in Speculative Decoding","publication_year":2026,"publication_date":"2026-02-11","ids":{"openalex":"https://openalex.org/W7133601568","doi":"https://doi.org/10.48550/arxiv.2603.03333"},"language":null,"primary_location":{"id":"pmh:doi:10.48550/arxiv.2603.03333","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"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":"Article"},"type":"article","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":null,"any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5128176961","display_name":"Jeongtae Lee","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lee, Jeongtae","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128218418","display_name":"Minjung Jo","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jo, Minjung","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5109524319","display_name":"Hyeongyo Jeong","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jeong, Hyunjoon","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5011127697","display_name":"Gunho Park","orcid":"https://orcid.org/0000-0002-8078-4356"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Park, Gunho","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128171058","display_name":"Sunghyeon Woo","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Woo, Sunghyeon","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128158278","display_name":"Joonghoon Kim","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kim, Joonghoon","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128208957","display_name":"Se Jung Kwon","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kwon, Se Jung","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5128193477","display_name":"Dongsoo Lee","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lee, Dongsoo","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":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.24475494,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"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":0.3294000029563904,"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":0.3294000029563904,"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":0.16519999504089355,"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":0.053199999034404755,"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":0.8306000232696533},{"id":"https://openalex.org/keywords/security-token","display_name":"Security token","score":0.7537999749183655},{"id":"https://openalex.org/keywords/dropout","display_name":"Dropout (neural networks)","score":0.7129999995231628},{"id":"https://openalex.org/keywords/speedup","display_name":"Speedup","score":0.6035000085830688},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.5805000066757202},{"id":"https://openalex.org/keywords/task","display_name":"Task (project management)","score":0.5461000204086304},{"id":"https://openalex.org/keywords/range","display_name":"Range (aeronautics)","score":0.4156999886035919},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.40540000796318054},{"id":"https://openalex.org/keywords/sampling","display_name":"Sampling (signal processing)","score":0.3734000027179718}],"concepts":[{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.8306000232696533},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8119999766349792},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.7537999749183655},{"id":"https://openalex.org/C2776145597","wikidata":"https://www.wikidata.org/wiki/Q25339462","display_name":"Dropout (neural networks)","level":2,"score":0.7129999995231628},{"id":"https://openalex.org/C68339613","wikidata":"https://www.wikidata.org/wiki/Q1549489","display_name":"Speedup","level":2,"score":0.6035000085830688},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.5805000066757202},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.5461000204086304},{"id":"https://openalex.org/C204323151","wikidata":"https://www.wikidata.org/wiki/Q905424","display_name":"Range (aeronautics)","level":2,"score":0.4156999886035919},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.40540000796318054},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4049000144004822},{"id":"https://openalex.org/C140779682","wikidata":"https://www.wikidata.org/wiki/Q210868","display_name":"Sampling (signal processing)","level":3,"score":0.3734000027179718},{"id":"https://openalex.org/C52740198","wikidata":"https://www.wikidata.org/wiki/Q1539564","display_name":"Importance sampling","level":3,"score":0.3549000024795532},{"id":"https://openalex.org/C126780896","wikidata":"https://www.wikidata.org/wiki/Q899871","display_name":"Distortion (music)","level":4,"score":0.34700000286102295},{"id":"https://openalex.org/C149441793","wikidata":"https://www.wikidata.org/wiki/Q200726","display_name":"Probability distribution","level":2,"score":0.34360000491142273},{"id":"https://openalex.org/C19499675","wikidata":"https://www.wikidata.org/wiki/Q232207","display_name":"Monte Carlo method","level":2,"score":0.3384999930858612},{"id":"https://openalex.org/C198531522","wikidata":"https://www.wikidata.org/wiki/Q485146","display_name":"Sample (material)","level":2,"score":0.30489999055862427},{"id":"https://openalex.org/C127162648","wikidata":"https://www.wikidata.org/wiki/Q16858953","display_name":"Channel (broadcasting)","level":2,"score":0.30480000376701355},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.30329999327659607},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3012000024318695},{"id":"https://openalex.org/C193969084","wikidata":"https://www.wikidata.org/wiki/Q7452500","display_name":"Sequential decoding","level":4,"score":0.29899999499320984},{"id":"https://openalex.org/C114289077","wikidata":"https://www.wikidata.org/wiki/Q3284399","display_name":"Statistical model","level":2,"score":0.2759999930858612},{"id":"https://openalex.org/C148220186","wikidata":"https://www.wikidata.org/wiki/Q7111912","display_name":"Outcome (game theory)","level":2,"score":0.2705000042915344},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.26739999651908875},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.2639999985694885},{"id":"https://openalex.org/C134261354","wikidata":"https://www.wikidata.org/wiki/Q938438","display_name":"Statistical inference","level":2,"score":0.26249998807907104},{"id":"https://openalex.org/C28719098","wikidata":"https://www.wikidata.org/wiki/Q44946","display_name":"Point (geometry)","level":2,"score":0.2581999897956848}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:doi:10.48550/arxiv.2603.03333","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"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":"Article"},{"id":"doi:10.48550/arxiv.2603.03333","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.03333","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:doi:10.48550/arxiv.2603.03333","is_oa":true,"landing_page_url":null,"pdf_url":null,"source":{"id":"https://openalex.org/S4406922384","display_name":"Open MIND","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"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":"Article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Speculative":[0],"decoding":[1,59,88,134],"accelerates":[2],"large":[3],"language":[4],"model":[5,14,41,81,101],"inference":[6,136,156],"by":[7],"proposing":[8],"tokens":[9,33,71],"with":[10,127],"a":[11,20,27,109,128],"lightweight":[12],"draft":[13,32,70],"and":[15,112,122,135,167],"selectively":[16],"accepting":[17],"them":[18],"using":[19],"target":[21,40,100],"model.":[22],"This":[23,76],"work":[24],"introduces":[25],"DropMatch,":[26],"novel":[28],"approach":[29,146],"that":[30,144],"matches":[31],"to":[34,48,82,119,161,169],"the":[35,39,49,80,85,99,164],"predictive":[36,102],"distribution":[37,67],"of":[38,87,98,131,178],"via":[42],"Monte":[43],"Carlo":[44],"dropout":[45,93],"applied":[46,175],"exclusively":[47],"LM":[50],"head,":[51],"enabling":[52],"sampling-based":[53],"acceptance":[54,77,148],"decisions.":[55],"By":[56],"generating":[57],"multiple":[58,141],"paths,":[60],"our":[61,145],"method":[62,106],"forms":[63],"an":[64,91,170],"empirical":[65],"token":[66],"against":[68],"which":[69],"are":[72],"evaluated":[73],"for":[74],"consistency.":[75],"mechanism":[78],"enables":[79],"adaptively":[83],"control":[84],"size":[86],"paths":[89],"under":[90],"appropriate":[92],"probability,":[94],"preventing":[95],"substantial":[96],"distortion":[97],"distribution.":[103],"The":[104],"proposed":[105],"operates":[107],"in":[108],"training-free,":[110],"data-free,":[111],"calibration-free":[113],"manner,":[114],"requires":[115],"no":[116],"architectural":[117],"modification":[118],"pretrained":[120],"models,":[121],"can":[123],"be":[124],"orthogonally":[125],"integrated":[126],"wide":[129],"range":[130],"existing":[132],"speculative":[133],"acceleration":[137],"techniques.":[138],"Experiments":[139],"across":[140],"benchmarks":[142],"demonstrate":[143],"increases":[147],"length":[149],"while":[150],"maintaining":[151],"competitive":[152],"task":[153],"performance,":[154],"yielding":[155],"speedups":[157],"ranging":[158],"from":[159],"1.09x":[160,172],"1.33x":[162],"over":[163],"standard":[165],"baseline,":[166],"up":[168],"additional":[171],"speedup":[173],"when":[174],"on":[176],"top":[177],"EAGLE3.":[179]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-03-06T00:00:00"}