{"id":"https://openalex.org/W7161039533","doi":"https://doi.org/10.48550/arxiv.2605.10980","title":"LEAP: Unlocking dLLM Parallelism via Lookahead Early-Convergence Token Detection","display_name":"LEAP: Unlocking dLLM Parallelism via Lookahead Early-Convergence Token Detection","publication_year":2026,"publication_date":"2026-05-09","ids":{"openalex":"https://openalex.org/W7161039533","doi":"https://doi.org/10.48550/arxiv.2605.10980"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.10980","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.10980","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":"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.2605.10980","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5136080769","display_name":"Haohui Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Haohui","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136045616","display_name":"Zhiye Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Zhiye","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5085791658","display_name":"Xiaoying Gan","orcid":"https://orcid.org/0000-0001-5200-1409"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Gan, Xiaoying","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136076171","display_name":"Xinbing Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Xinbing","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5136083494","display_name":"Bo Jiang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jiang, Bo","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/T10201","display_name":"Speech Recognition and Synthesis","score":0.32749998569488525,"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/T10201","display_name":"Speech Recognition and Synthesis","score":0.32749998569488525,"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":0.17890000343322754,"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":0.15330000221729279,"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/security-token","display_name":"Security token","score":0.7379000186920166},{"id":"https://openalex.org/keywords/scalability","display_name":"Scalability","score":0.6676999926567078},{"id":"https://openalex.org/keywords/decoding-methods","display_name":"Decoding methods","score":0.593500018119812},{"id":"https://openalex.org/keywords/context","display_name":"Context (archaeology)","score":0.5515999794006348},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.5295000076293945},{"id":"https://openalex.org/keywords/benchmarking","display_name":"Benchmarking","score":0.43140000104904175},{"id":"https://openalex.org/keywords/process","display_name":"Process (computing)","score":0.40119999647140503},{"id":"https://openalex.org/keywords/independence","display_name":"Independence (probability theory)","score":0.3978999853134155},{"id":"https://openalex.org/keywords/early-stopping","display_name":"Early stopping","score":0.3806999921798706}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7807000279426575},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.7379000186920166},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.6676999926567078},{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.593500018119812},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.5515999794006348},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.5295000076293945},{"id":"https://openalex.org/C86251818","wikidata":"https://www.wikidata.org/wiki/Q816754","display_name":"Benchmarking","level":2,"score":0.43140000104904175},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.40119999647140503},{"id":"https://openalex.org/C35651441","wikidata":"https://www.wikidata.org/wiki/Q625303","display_name":"Independence (probability theory)","level":2,"score":0.3978999853134155},{"id":"https://openalex.org/C5465570","wikidata":"https://www.wikidata.org/wiki/Q5326898","display_name":"Early stopping","level":3,"score":0.3806999921798706},{"id":"https://openalex.org/C46637626","wikidata":"https://www.wikidata.org/wiki/Q6693015","display_name":"Low latency (capital markets)","level":2,"score":0.37229999899864197},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.3668999969959259},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3352000117301941},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.32919999957084656},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3276999890804291},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.31709998846054077},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.30959999561309814},{"id":"https://openalex.org/C111335779","wikidata":"https://www.wikidata.org/wiki/Q3454686","display_name":"Reduction (mathematics)","level":2,"score":0.3059999942779541},{"id":"https://openalex.org/C2777303404","wikidata":"https://www.wikidata.org/wiki/Q759757","display_name":"Convergence (economics)","level":2,"score":0.2955000102519989},{"id":"https://openalex.org/C134261354","wikidata":"https://www.wikidata.org/wiki/Q938438","display_name":"Statistical inference","level":2,"score":0.2770000100135803},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.26759999990463257},{"id":"https://openalex.org/C163294075","wikidata":"https://www.wikidata.org/wiki/Q581861","display_name":"Noise reduction","level":2,"score":0.26759999990463257},{"id":"https://openalex.org/C61326573","wikidata":"https://www.wikidata.org/wiki/Q1496376","display_name":"Gaussian process","level":3,"score":0.2669999897480011},{"id":"https://openalex.org/C175291020","wikidata":"https://www.wikidata.org/wiki/Q1156822","display_name":"Offset (computer science)","level":2,"score":0.2615000009536743},{"id":"https://openalex.org/C206688291","wikidata":"https://www.wikidata.org/wiki/Q7617819","display_name":"Stochastic gradient descent","level":3,"score":0.25699999928474426},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.25529998540878296},{"id":"https://openalex.org/C114289077","wikidata":"https://www.wikidata.org/wiki/Q3284399","display_name":"Statistical model","level":2,"score":0.2506999969482422}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.10980","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.10980","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":"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.2605.10980","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.10980","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":"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":[{"score":0.4450241029262543,"display_name":"Reduced inequalities","id":"https://metadata.un.org/sdg/10"}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Diffusion":[0],"Language":[1],"Models":[2],"(dLLMs)":[3],"have":[4],"garnered":[5],"significant":[6],"attention":[7],"for":[8,104,203],"their":[9,72],"potential":[10],"in":[11,76],"highly":[12],"parallel":[13,16,204],"processing.":[14],"The":[15],"capabilities":[17],"of":[18,25,55,68,140,164],"existing":[19],"dLLMs":[20],"stem":[21],"from":[22],"the":[23,37,43,53,77,128,161,173,194],"assumption":[24],"conditional":[26],"independence":[27],"at":[28],"high":[29],"confidence":[30,45,85],"levels,":[31],"which":[32],"ensures":[33],"negligible":[34],"discrepancy":[35],"between":[36,130],"marginal":[38],"and":[39,119,133,154],"joint":[40],"distributions.":[41],"However,":[42],"stringent":[44],"thresholds":[46],"required":[47],"to":[48,71,82,122,158,182],"preserve":[49],"accuracy":[50],"severely":[51],"constrain":[52],"scalability":[54],"parallelism.":[56],"Through":[57],"systematic":[58],"token-level":[59],"statistical":[60],"analysis,":[61],"we":[62,97,135],"reveal":[63],"that":[64,88,114,148],"a":[65,110,200],"substantial":[66],"proportion":[67],"tokens":[69,184],"converge":[70],"correct":[73],"predictions":[74],"early":[75,131,138],"denoising":[78,165],"process":[79],"yet":[80],"fail":[81],"reach":[83],"standard":[84],"thresholds,":[86],"confirming":[87],"current":[89],"confidence-based":[90,159],"criteria":[91],"are":[92],"overly":[93],"conservative.":[94],"In":[95],"response,":[96],"introduce":[98],"LEAP":[99,108,149,177,191],"(Lookahead":[100],"Early-Convergence":[101],"Token":[102],"Detection":[103],"Accelerated":[105],"Parallel":[106],"Decoding).":[107],"is":[109,167],"training-free,":[111],"plug-and-play":[112],"method":[113],"leverages":[115],"future":[116],"context":[117],"filtering":[118],"multi-sequence":[120],"superposition":[121],"detect":[123],"early-converging":[124],"tokens.":[125,142],"By":[126],"validating":[127],"alignment":[129],"convergence":[132],"correctness,":[134],"enable":[136],"reliable":[137],"decoding":[139,155,181],"these":[141],"Benchmarking":[143],"across":[144],"diverse":[145],"domains":[146],"demonstrates":[147],"significantly":[150],"lowers":[151],"inference":[152],"latency":[153],"steps.":[156],"Compared":[157],"decoding,":[160],"average":[162],"number":[163],"steps":[166],"reduced":[168],"by":[169],"about":[170],"30%.":[171],"On":[172],"GSM8K":[174],"dataset,":[175],"combining":[176],"with":[178],"dParallel":[179],"accelerates":[180],"7.2":[183],"per":[185],"step":[186],"while":[187],"preserving":[188],"model":[189],"precision.":[190],"effectively":[192],"breaks":[193],"reliance":[195],"on":[196],"high-confidence":[197],"priors,":[198],"offering":[199],"novel":[201],"paradigm":[202],"decoding.":[205]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-05-14T00:00:00"}