{"id":"https://openalex.org/W7155105059","doi":"https://doi.org/10.48550/arxiv.2604.18471","title":"NI Sampling: Accelerating Discrete Diffusion Sampling by Token Order Optimization","display_name":"NI Sampling: Accelerating Discrete Diffusion Sampling by Token Order Optimization","publication_year":2026,"publication_date":"2026-04-20","ids":{"openalex":"https://openalex.org/W7155105059","doi":"https://doi.org/10.48550/arxiv.2604.18471"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2604.18471","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.18471","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.2604.18471","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5134198516","display_name":"Enshu Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Enshu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134112627","display_name":"Xuefei Ning","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ning, Xuefei","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100445217","display_name":"Yu Wang","orcid":"https://orcid.org/0000-0002-4788-8655"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Yu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5134178206","display_name":"Zinan Lin","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lin, Zinan","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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.3043000102043152,"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"}},"topics":[{"id":"https://openalex.org/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.3043000102043152,"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"}},{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.15459999442100525,"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":0.07339999824762344,"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/sampling","display_name":"Sampling (signal processing)","score":0.7542999982833862},{"id":"https://openalex.org/keywords/security-token","display_name":"Security token","score":0.6039999723434448},{"id":"https://openalex.org/keywords/heuristic","display_name":"Heuristic","score":0.5001999735832214},{"id":"https://openalex.org/keywords/autoregressive-model","display_name":"Autoregressive model","score":0.4715000092983246},{"id":"https://openalex.org/keywords/flexibility","display_name":"Flexibility (engineering)","score":0.4706000089645386},{"id":"https://openalex.org/keywords/code","display_name":"Code (set theory)","score":0.4036000072956085},{"id":"https://openalex.org/keywords/sample","display_name":"Sample (material)","score":0.397599995136261},{"id":"https://openalex.org/keywords/acceleration","display_name":"Acceleration","score":0.3806999921798706}],"concepts":[{"id":"https://openalex.org/C140779682","wikidata":"https://www.wikidata.org/wiki/Q210868","display_name":"Sampling (signal processing)","level":3,"score":0.7542999982833862},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7236999869346619},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.6039999723434448},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.5202999711036682},{"id":"https://openalex.org/C173801870","wikidata":"https://www.wikidata.org/wiki/Q201413","display_name":"Heuristic","level":2,"score":0.5001999735832214},{"id":"https://openalex.org/C159877910","wikidata":"https://www.wikidata.org/wiki/Q2202883","display_name":"Autoregressive model","level":2,"score":0.4715000092983246},{"id":"https://openalex.org/C2780598303","wikidata":"https://www.wikidata.org/wiki/Q65921492","display_name":"Flexibility (engineering)","level":2,"score":0.4706000089645386},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.4036000072956085},{"id":"https://openalex.org/C198531522","wikidata":"https://www.wikidata.org/wiki/Q485146","display_name":"Sample (material)","level":2,"score":0.397599995136261},{"id":"https://openalex.org/C117896860","wikidata":"https://www.wikidata.org/wiki/Q11376","display_name":"Acceleration","level":2,"score":0.3806999921798706},{"id":"https://openalex.org/C52740198","wikidata":"https://www.wikidata.org/wiki/Q1539564","display_name":"Importance sampling","level":3,"score":0.38040000200271606},{"id":"https://openalex.org/C137836250","wikidata":"https://www.wikidata.org/wiki/Q984063","display_name":"Optimization problem","level":2,"score":0.36910000443458557},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.3513999879360199},{"id":"https://openalex.org/C2781395549","wikidata":"https://www.wikidata.org/wiki/Q4680762","display_name":"Adaptive sampling","level":3,"score":0.33000001311302185},{"id":"https://openalex.org/C182306322","wikidata":"https://www.wikidata.org/wiki/Q1779371","display_name":"Order (exchange)","level":2,"score":0.3140999972820282},{"id":"https://openalex.org/C2986012078","wikidata":"https://www.wikidata.org/wiki/Q210868","display_name":"Sampling interval","level":2,"score":0.3046000003814697},{"id":"https://openalex.org/C69357855","wikidata":"https://www.wikidata.org/wiki/Q163214","display_name":"Diffusion","level":2,"score":0.28299999237060547},{"id":"https://openalex.org/C170593435","wikidata":"https://www.wikidata.org/wiki/Q4128565","display_name":"Slice sampling","level":4,"score":0.2782000005245209},{"id":"https://openalex.org/C177148314","wikidata":"https://www.wikidata.org/wiki/Q170084","display_name":"Generalization","level":2,"score":0.27799999713897705},{"id":"https://openalex.org/C145671259","wikidata":"https://www.wikidata.org/wiki/Q1493786","display_name":"Discrete optimization","level":3,"score":0.2777999937534332},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.26100000739097595},{"id":"https://openalex.org/C129848803","wikidata":"https://www.wikidata.org/wiki/Q2564360","display_name":"Sample size determination","level":2,"score":0.26089999079704285},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.25279998779296875},{"id":"https://openalex.org/C2778067643","wikidata":"https://www.wikidata.org/wiki/Q166507","display_name":"Interval (graph theory)","level":2,"score":0.25209999084472656}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2604.18471","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.18471","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.2604.18471","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.18471","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":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Discrete":[0],"diffusion":[1],"language":[2],"models":[3,148],"(dLLMs)":[4],"have":[5],"recently":[6],"emerged":[7],"as":[8],"a":[9,41,111,119,135],"promising":[10],"alternative":[11],"to":[12,19,46,122,139,158],"traditional":[13],"autoregressive":[14],"approaches,":[15],"offering":[16],"the":[17,26,61,88,141,175],"flexibility":[18],"generate":[20],"tokens":[21,45,125],"in":[22,174],"arbitrary":[23],"orders":[24],"and":[25,68,146,168],"potential":[27,72],"of":[28,44,63,90,96],"parallel":[29],"decoding.":[30],"However,":[31],"existing":[32],"heuristic":[33],"sampling":[34,65,91,113,163,173],"strategies":[35],"remain":[36],"inefficient:":[37],"they":[38],"choose":[39],"only":[40],"small":[42],"part":[43],"sample":[47],"at":[48,83,129,181],"each":[49,84,130],"step,":[50],"leaving":[51],"substantial":[52],"room":[53],"for":[54,73],"improvement.":[55],"In":[56],"this":[57],"work,":[58],"we":[59,76,104],"study":[60],"problem":[62],"token":[64],"order":[66,95,114],"optimization":[67,115],"demonstrate":[69],"its":[70],"significant":[71],"acceleration.":[74],"Specifically,":[75],"find":[77],"that":[78,117,153],"fully":[79],"leveraging":[80],"correct":[81],"predictions":[82],"step":[85],"can":[86],"reduce":[87],"number":[89],"iterations":[92],"by":[93],"an":[94],"magnitude":[97],"without":[98],"compromising":[99],"accuracy.":[100],"Based":[101],"on":[102,144],"this,":[103],"propose":[105,134],"Neural":[106],"Indicator":[107],"Sampling":[108],"(NI":[109],"Sampling),":[110],"general":[112],"framework":[116],"utilize":[118],"neural":[120],"indicator":[121],"decide":[123],"which":[124],"should":[126],"be":[127],"sampled":[128],"step.":[131],"We":[132],"further":[133],"novel":[136],"trajectory-preserving":[137],"objective":[138],"train":[140],"indicator.":[142],"Experiments":[143],"LLaDA":[145],"Dream":[147],"across":[149],"multiple":[150],"benchmarks":[151],"show":[152],"our":[154],"method":[155],"achieves":[156],"up":[157],"14.3$\\times$":[159],"acceleration":[160],"over":[161],"full-step":[162],"with":[164],"negligible":[165],"performance":[166],"drop,":[167],"consistently":[169],"outperforms":[170],"confidence":[171],"threshold":[172],"accuracy-step":[176],"trade-off.":[177],"Code":[178],"is":[179],"available":[180],"https://github.com/imagination-research/NI-Sampling.":[182]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-04-22T00:00:00"}