{"id":"https://openalex.org/W7204940474","doi":"https://doi.org/10.48550/arxiv.2608.29291","title":"Accelerating Unified Multimodal Models with Core-Expansion Routing and Unified Computation Scheduling","display_name":"Accelerating Unified Multimodal Models with Core-Expansion Routing and Unified Computation Scheduling","publication_year":2026,"publication_date":"2026-08-29","ids":{"openalex":"https://openalex.org/W7204940474","doi":"https://doi.org/10.48550/arxiv.2608.29291"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2608.29291","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.29291","pdf_url":"https://arxiv.org/pdf/2608.29291","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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2608.29291","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5101394613","display_name":"Wengyi Zhan","orcid":"https://orcid.org/0009-0001-5059-2222","observed_orcids":["https://orcid.org/0009-0001-5059-2222"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhan, Wengyi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5048876463","display_name":"Chenqian Yan","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yan, Chenqian","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5148903687","display_name":"Songwei Liu","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Songwei","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138203376","display_name":"Mingbao Lin","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lin, Mingbao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5111335834","display_name":"Rongrong Ji","orcid":"https://orcid.org/0000-0001-9163-2932","observed_orcids":["https://orcid.org/0000-0001-9163-2932"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ji, Rongrong","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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.9526000022888184,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.9526000022888184,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.02800000086426735,"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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.011300000362098217,"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/unified-multimodal-models","display_name":"unified multimodal models","score":0.9959999918937683},{"id":"https://openalex.org/keywords/layer-skipping","display_name":"layer skipping","score":0.8560000061988831},{"id":"https://openalex.org/keywords/inference-acceleration","display_name":"inference acceleration","score":0.6769999861717224}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8151000142097473},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.7501999735832214},{"id":"https://openalex.org/C206729178","wikidata":"https://www.wikidata.org/wiki/Q2271896","display_name":"Scheduling (production processes)","level":2,"score":0.6129999756813049},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.553600013256073},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.53329998254776},{"id":"https://openalex.org/C168167062","wikidata":"https://www.wikidata.org/wiki/Q1117970","display_name":"Component (thermodynamics)","level":2,"score":0.5180000066757202},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.4986000061035156},{"id":"https://openalex.org/C74172769","wikidata":"https://www.wikidata.org/wiki/Q1446839","display_name":"Routing (electronic design automation)","level":2,"score":0.4099999964237213},{"id":"https://openalex.org/C45493050","wikidata":"https://www.wikidata.org/wiki/Q7884934","display_name":"Unified Model","level":2,"score":0.4081999957561493},{"id":"https://openalex.org/C115537543","wikidata":"https://www.wikidata.org/wiki/Q165596","display_name":"Cache","level":2,"score":0.40560001134872437},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.3799999952316284},{"id":"https://openalex.org/C2164484","wikidata":"https://www.wikidata.org/wiki/Q5170150","display_name":"Core (optical fiber)","level":2,"score":0.3580000102519989},{"id":"https://openalex.org/C2779227376","wikidata":"https://www.wikidata.org/wiki/Q6505497","display_name":"Layer (electronics)","level":2,"score":0.3506999909877777},{"id":"https://openalex.org/C124681953","wikidata":"https://www.wikidata.org/wiki/Q339062","display_name":"Decomposition","level":2,"score":0.32170000672340393},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.3077999949455261},{"id":"https://openalex.org/C78766204","wikidata":"https://www.wikidata.org/wiki/Q555032","display_name":"Multi-core processor","level":2,"score":0.2971999943256378},{"id":"https://openalex.org/C184596265","wikidata":"https://www.wikidata.org/wiki/Q2651576","display_name":"Model of computation","level":3,"score":0.296099990606308},{"id":"https://openalex.org/C2984173633","wikidata":"https://www.wikidata.org/wiki/Q22725","display_name":"Routing algorithm","level":4,"score":0.2879999876022339},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.2635999917984009},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.2517000138759613}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2608.29291","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.29291","pdf_url":"https://arxiv.org/pdf/2608.29291","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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2608.29291","doi":"https://doi.org/10.48550/arxiv.2608.29291","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2608.29291","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2608.29291","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.29291","pdf_url":"https://arxiv.org/pdf/2608.29291","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":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W7138202757","https://openalex.org/W7222955963","https://openalex.org/W7128781553","https://openalex.org/W7083169517","https://openalex.org/W7203772916","https://openalex.org/W7220963867","https://openalex.org/W7109105434","https://openalex.org/W7106789511","https://openalex.org/W4417004418","https://openalex.org/W4417123848"],"abstract_inverted_index":{"Unified":[0,79],"multimodal":[1],"models":[2],"jointly":[3],"support":[4],"understanding":[5,28,112],"and":[6,16,54,62,72,91],"generation,":[7],"but":[8,40],"incur":[9],"substantial":[10],"redundant":[11],"computation":[12,71],"across":[13,105],"tokens,":[14],"layers,":[15],"generation":[17,35,56,60],"timesteps.":[18],"Through":[19],"token-importance":[20],"probing,":[21],"we":[22],"identify":[23],"an":[24],"asymmetric":[25],"core-expansion":[26],"structure:":[27],"exhibits":[29],"a":[30,50,74,115],"stable":[31],"importance":[32],"component,":[33],"while":[34],"largely":[36],"shares":[37],"this":[38],"component":[39],"requires":[41],"progress-dependent":[42],"corrections.":[43],"We":[44],"therefore":[45],"propose":[46],"CE-Router,":[47],"which":[48,82],"uses":[49],"task-shared":[51],"core":[52,64],"scorer":[53],"progress-conditioned":[55],"expansions,":[57],"optimized":[58],"through":[59],"decomposition":[61],"cross-task":[63],"alignment.":[65],"At":[66],"inference,":[67],"CE-Router":[68],"compacts":[69],"token":[70],"supplies":[73],"learned":[75],"routing":[76],"signal":[77],"to":[78],"Computation":[80],"Scheduling,":[81],"coordinates":[83],"layer":[84],"skipping,":[85],"FFN":[86],"pruning,":[87],"diffusion-head":[88],"cache":[89],"reuse,":[90],"denoising-step":[92],"early":[93],"exit.":[94],"Experiments":[95],"on":[96],"two":[97],"representative":[98],"UMM":[99],"architectures":[100],"demonstrate":[101],"consistent":[102],"quality--efficiency":[103],"improvements":[104],"both":[106],"tasks,":[107],"retaining":[108],"98.03\\%":[109],"of":[110],"dense":[111],"performance":[113],"with":[114],"1.93$\\times$":[116],"end-to-end":[117],"inference":[118],"speedup.":[119]},"counts_by_year":[],"updated_date":"2026-10-28T10:33:24.094632","created_date":"2026-09-02T00:00:00"}