{"id":"https://openalex.org/W7127329259","doi":"https://doi.org/10.48550/arxiv.2602.02159","title":"Focus-dLLM: Accelerating Long-Context Diffusion LLM Inference via Confidence-Guided Context Focusing","display_name":"Focus-dLLM: Accelerating Long-Context Diffusion LLM Inference via Confidence-Guided Context Focusing","publication_year":2026,"publication_date":"2026-02-02","ids":{"openalex":"https://openalex.org/W7127329259","doi":"https://doi.org/10.48550/arxiv.2602.02159"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2602.02159","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2602.02159","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.2602.02159","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5124931058","display_name":"Lingkun Long","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Long, Lingkun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5108256718","display_name":"Yushi Huang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Huang, Yushi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5124946193","display_name":"Shihao Bai","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Bai, Shihao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5063549968","display_name":"Ruihao Gong","orcid":"https://orcid.org/0000-0002-0081-5395"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Gong, Ruihao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5124930101","display_name":"Jun Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Jun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5124878812","display_name":"Ao Zhou","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhou, Ao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5124958926","display_name":"Jianlei Yang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yang, Jianlei","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":0.38510000705718994,"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":0.38510000705718994,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.07450000196695328,"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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.07109999656677246,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.7541999816894531},{"id":"https://openalex.org/keywords/security-token","display_name":"Security token","score":0.6909999847412109},{"id":"https://openalex.org/keywords/speedup","display_name":"Speedup","score":0.5871000289916992},{"id":"https://openalex.org/keywords/context","display_name":"Context (archaeology)","score":0.5690000057220459},{"id":"https://openalex.org/keywords/decoding-methods","display_name":"Decoding methods","score":0.5386000275611877},{"id":"https://openalex.org/keywords/pruning","display_name":"Pruning","score":0.5264000296592712},{"id":"https://openalex.org/keywords/lossless-compression","display_name":"Lossless compression","score":0.46480000019073486}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7954999804496765},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.7541999816894531},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.6909999847412109},{"id":"https://openalex.org/C68339613","wikidata":"https://www.wikidata.org/wiki/Q1549489","display_name":"Speedup","level":2,"score":0.5871000289916992},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.5690000057220459},{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.5386000275611877},{"id":"https://openalex.org/C108010975","wikidata":"https://www.wikidata.org/wiki/Q500094","display_name":"Pruning","level":2,"score":0.5264000296592712},{"id":"https://openalex.org/C81081738","wikidata":"https://www.wikidata.org/wiki/Q55542","display_name":"Lossless compression","level":3,"score":0.46480000019073486},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.44670000672340393},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.439300000667572},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4287000000476837},{"id":"https://openalex.org/C43126263","wikidata":"https://www.wikidata.org/wiki/Q128751","display_name":"Source code","level":2,"score":0.382099986076355},{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.3790999948978424},{"id":"https://openalex.org/C147764199","wikidata":"https://www.wikidata.org/wiki/Q6865248","display_name":"Minification","level":2,"score":0.37279999256134033},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.3686000108718872},{"id":"https://openalex.org/C183322885","wikidata":"https://www.wikidata.org/wiki/Q17007702","display_name":"Context model","level":3,"score":0.35580000281333923},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3422999978065491},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.2930999994277954},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.2842000126838684},{"id":"https://openalex.org/C2777472644","wikidata":"https://www.wikidata.org/wiki/Q16968992","display_name":"Approximate inference","level":3,"score":0.2685999870300293},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.2535000145435333}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2602.02159","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2602.02159","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.2602.02159","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2602.02159","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":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Diffusion":[0],"Large":[1],"Language":[2],"Models":[3],"(dLLMs)":[4],"deliver":[5],"strong":[6],"long-context":[7,78],"processing":[8],"capability":[9],"in":[10],"a":[11,67,96,109],"non-autoregressive":[12],"decoding":[13],"paradigm.":[14],"However,":[15],"the":[16,25,40,53,83,140],"considerable":[17],"computational":[18],"cost":[19],"of":[20],"bidirectional":[21],"full":[22],"attention":[23,30,44,70,119,125],"limits":[24],"inference":[26],"efficiency.":[27],"Although":[28],"sparse":[29],"is":[31,162],"promising,":[32],"existing":[33],"methods":[34],"remain":[35],"ineffective.":[36],"This":[37],"stems":[38],"from":[39],"need":[41],"to":[42,49,100,113],"estimate":[43,115],"importance":[45],"for":[46,74],"tokens":[47],"yet":[48],"be":[50],"decoded,":[51],"while":[52,121],"unmasked":[54,102],"token":[55,86],"positions":[56],"are":[57],"unknown":[58],"during":[59],"diffusion.":[60],"In":[61],"this":[62,131],"paper,":[63],"we":[64,93,107],"present":[65],"Focus-dLLM,":[66],"novel":[68],"training-free":[69],"sparsification":[71],"framework":[72],"tailored":[73],"accurate":[75],"and":[76,116],"efficient":[77],"dLLM":[79],"inference.":[80],"Based":[81],"on":[82],"finding":[84],"that":[85,147],"confidence":[87],"strongly":[88],"correlates":[89],"across":[90,137],"adjacent":[91],"steps,":[92],"first":[94],"design":[95],"past":[97],"confidence-guided":[98],"indicator":[99],"predict":[101],"regions.":[103],"Built":[104],"upon":[105],"this,":[106],"propose":[108],"sink-aware":[110],"pruning":[111],"strategy":[112,132],"accurately":[114],"remove":[117],"redundant":[118],"computation,":[120],"preserving":[122],"highly":[123],"influential":[124],"sinks.":[126],"To":[127],"further":[128],"reduce":[129],"overhead,":[130],"reuses":[133],"identified":[134],"sink":[135],"locations":[136],"layers,":[138],"leveraging":[139],"observed":[141],"cross-layer":[142],"consistency.":[143],"Experimental":[144],"results":[145],"show":[146],"our":[148],"method":[149],"offers":[150],"more":[151],"than":[152],"$29\\times$":[153],"lossless":[154],"speedup":[155],"under":[156],"$32K$":[157],"context":[158],"length.":[159],"The":[160],"code":[161],"publicly":[163],"available":[164],"at:":[165],"https://github.com/Longxmas/Focus-dLLM":[166]},"counts_by_year":[],"updated_date":"2026-08-16T07:02:28.622633","created_date":"2026-02-04T00:00:00"}