{"id":"https://openalex.org/W7128006584","doi":"https://doi.org/10.48550/arxiv.2602.06036","title":"DFlash: Block Diffusion for Flash Speculative Decoding","display_name":"DFlash: Block Diffusion for Flash Speculative Decoding","publication_year":2026,"publication_date":"2026-02-05","ids":{"openalex":"https://openalex.org/W7128006584","doi":"https://doi.org/10.48550/arxiv.2602.06036"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2602.06036","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2602.06036","pdf_url":"https://arxiv.org/pdf/2602.06036","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","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/2602.06036","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5125115803","display_name":"Jian Chen","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen, Jian","raw_affiliation_strings":["https:// - lab.ai/projects/dflash"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"https:// - lab.ai/projects/dflash","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125208830","display_name":"Yesheng Liang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liang, Yesheng","raw_affiliation_strings":["https:// - lab.ai/projects/dflash"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"https:// - lab.ai/projects/dflash","institution_ids":[]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5100750264","display_name":"Zhijian Liu","orcid":"https://orcid.org/0000-0003-3632-9986","observed_orcids":["https://orcid.org/0000-0003-3632-9986"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Zhijian","raw_affiliation_strings":["https:// - lab.ai/projects/dflash"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"https:// - lab.ai/projects/dflash","institution_ids":[]}]}],"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":true,"cited_by_count":138,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.8734999895095825,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.8734999895095825,"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":0.1071000024676323,"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/T10201","display_name":"Speech Recognition and Synthesis","score":0.0038999998942017555,"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/speculative-decoding","display_name":"speculative decoding","score":0.9860000014305115},{"id":"https://openalex.org/keywords/inference-latency","display_name":"inference latency","score":0.5600000023841858},{"id":"https://openalex.org/keywords/gpu-utilization","display_name":"GPU utilization","score":0.6489999890327454},{"id":"https://openalex.org/keywords/contextual-features","display_name":"contextual features","score":0.5109999775886536},{"id":"https://openalex.org/keywords/acceptance-rate","display_name":"acceptance rate","score":0.5149999856948853}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7888000011444092},{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.7098000049591064},{"id":"https://openalex.org/C159877910","wikidata":"https://www.wikidata.org/wiki/Q2202883","display_name":"Autoregressive model","level":2,"score":0.6732000112533569},{"id":"https://openalex.org/C68339613","wikidata":"https://www.wikidata.org/wiki/Q1549489","display_name":"Speedup","level":2,"score":0.5960999727249146},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.5342000126838684},{"id":"https://openalex.org/C2777210771","wikidata":"https://www.wikidata.org/wiki/Q4927124","display_name":"Block (permutation group theory)","level":2,"score":0.5042999982833862},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.48829999566078186},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.47850000858306885},{"id":"https://openalex.org/C2780513914","wikidata":"https://www.wikidata.org/wiki/Q18210350","display_name":"Bottleneck","level":2,"score":0.41019999980926514},{"id":"https://openalex.org/C43521106","wikidata":"https://www.wikidata.org/wiki/Q2165493","display_name":"Pipeline (software)","level":2,"score":0.4049000144004822},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.3955000042915344},{"id":"https://openalex.org/C69357855","wikidata":"https://www.wikidata.org/wiki/Q163214","display_name":"Diffusion","level":2,"score":0.36820000410079956},{"id":"https://openalex.org/C81081738","wikidata":"https://www.wikidata.org/wiki/Q55542","display_name":"Lossless compression","level":3,"score":0.35510000586509705},{"id":"https://openalex.org/C204323151","wikidata":"https://www.wikidata.org/wiki/Q905424","display_name":"Range (aeronautics)","level":2,"score":0.3418999910354614},{"id":"https://openalex.org/C46637626","wikidata":"https://www.wikidata.org/wiki/Q6693015","display_name":"Low latency (capital markets)","level":2,"score":0.335099995136261},{"id":"https://openalex.org/C169590947","wikidata":"https://www.wikidata.org/wiki/Q47506","display_name":"Compiler","level":2,"score":0.3181999921798706},{"id":"https://openalex.org/C70437156","wikidata":"https://www.wikidata.org/wiki/Q7228652","display_name":"Pooling","level":2,"score":0.29170000553131104},{"id":"https://openalex.org/C117896860","wikidata":"https://www.wikidata.org/wiki/Q11376","display_name":"Acceleration","level":2,"score":0.26579999923706055},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.26409998536109924}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2602.06036","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2602.06036","pdf_url":"https://arxiv.org/pdf/2602.06036","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2602.06036","doi":"https://doi.org/10.48550/arxiv.2602.06036","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2602.06036","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":"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"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2602.06036","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2602.06036","pdf_url":"https://arxiv.org/pdf/2602.06036","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[{"id":"https://openalex.org/F4320314997","display_name":"Strong","ror":"https://ror.org/041vyzr56"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7128006584.pdf","grobid_xml":"https://content.openalex.org/works/W7128006584.grobid-xml"},"referenced_works_count":19,"referenced_works":["https://openalex.org/W3177813494","https://openalex.org/W3195140297","https://openalex.org/W3210904070","https://openalex.org/W4310561894","https://openalex.org/W4379089709","https://openalex.org/W4389519291","https://openalex.org/W4391124766","https://openalex.org/W4391335474","https://openalex.org/W4392822190","https://openalex.org/W4401306886","https://openalex.org/W4402696815","https://openalex.org/W4404782770","https://openalex.org/W4405766390","https://openalex.org/W4414985005","https://openalex.org/W4415097124","https://openalex.org/W4415797413","https://openalex.org/W4416235692","https://openalex.org/W4416368080","https://openalex.org/W7084115179"],"related_works":["https://openalex.org/W7221306765","https://openalex.org/W7214540227","https://openalex.org/W7163207792","https://openalex.org/W4416368080","https://openalex.org/W7125976435","https://openalex.org/W7163995200","https://openalex.org/W7170224939","https://openalex.org/W7223027316","https://openalex.org/W4415938238","https://openalex.org/W7166476568"],"abstract_inverted_index":{"Autoregressive":[0],"large":[1],"language":[2],"models":[3,71,144],"(LLMs)":[4],"deliver":[5],"strong":[6],"performance":[7],"but":[8,68],"require":[9],"inherently":[10],"sequential":[11,53],"decoding,":[12],"leading":[13],"to":[14,149],"high":[15],"inference":[16],"latency":[17],"and":[18,54,107,127,145],"poor":[19],"GPU":[20],"utilization.":[21],"Speculative":[22],"decoding":[23,86,157],"mitigates":[24],"this":[25,79],"bottleneck":[26],"by":[27,39,64],"using":[28],"a":[29,61,84,90,103,141],"fast":[30],"draft":[31,100,110],"model":[32,94,111],"whose":[33],"outputs":[34,126],"are":[35],"verified":[36],"in":[37,102],"parallel":[38,66,96],"the":[40,109,117,154],"target":[41,118],"LLM;":[42],"however,":[43],"existing":[44],"methods":[45],"still":[46],"rely":[47],"on":[48,112],"autoregressive":[49,76],"drafting,":[50],"which":[51],"remains":[52],"limits":[55],"practical":[56],"speedups.":[57],"Diffusion":[58],"LLMs":[59],"offer":[60],"promising":[62],"alternative":[63],"enabling":[65],"generation,":[67],"current":[69],"diffusion":[70,93],"typically":[72],"underperform":[73],"compared":[74],"with":[75,124],"models.":[77],"In":[78],"paper,":[80],"we":[81],"introduce":[82],"DFlash,":[83],"speculative":[85,156],"framework":[87],"that":[88,133],"employs":[89],"lightweight":[91],"block":[92],"for":[95],"drafting.":[97],"By":[98],"generating":[99],"tokens":[101],"single":[104],"forward":[105],"pass":[106],"conditioning":[108],"context":[113],"features":[114],"extracted":[115],"from":[116],"model,":[119],"DFlash":[120,134],"enables":[121],"efficient":[122],"drafting":[123],"high-quality":[125],"higher":[128,151],"acceptance":[129],"rates.":[130],"Experiments":[131],"show":[132],"achieves":[135],"over":[136],"6x":[137],"lossless":[138],"acceleration":[139],"across":[140],"range":[142],"of":[143],"tasks,":[146],"delivering":[147],"up":[148],"2.5x":[150],"speedup":[152],"than":[153],"state-of-the-art":[155],"method":[158],"EAGLE-3.":[159]},"counts_by_year":[{"year":2026,"cited_by_count":136},{"year":2025,"cited_by_count":2}],"updated_date":"2026-11-12T12:28:57.059296","created_date":"2026-02-07T00:00:00"}