{"id":"https://openalex.org/W7160991731","doi":"https://doi.org/10.48550/arxiv.2605.12056","title":"OmniRefine: Alignment-Aware Cooperative Compression for Efficient Omnimodal Large Language Models","display_name":"OmniRefine: Alignment-Aware Cooperative Compression for Efficient Omnimodal Large Language Models","publication_year":2026,"publication_date":"2026-05-12","ids":{"openalex":"https://openalex.org/W7160991731","doi":"https://doi.org/10.48550/arxiv.2605.12056"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.12056","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.12056","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.2605.12056","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5136026693","display_name":"Yuchen Deng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Deng, Yuchen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136016720","display_name":"Zidang Cai","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Cai, Zidang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136050486","display_name":"Hai-Tao Zheng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zheng, Hai-Tao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5136046895","display_name":"Jie Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Jie","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5040168310","display_name":"Feidiao Yang","orcid":"https://orcid.org/0000-0001-7637-2562"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yang, Feidiao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5136087148","display_name":"Yuxing Han","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Han, Yuxing","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/T10860","display_name":"Speech and Audio Processing","score":0.3971000015735626,"subfield":{"id":"https://openalex.org/subfields/1711","display_name":"Signal Processing"},"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/T10860","display_name":"Speech and Audio Processing","score":0.3971000015735626,"subfield":{"id":"https://openalex.org/subfields/1711","display_name":"Signal Processing"},"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.2948000133037567,"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/T11309","display_name":"Music and Audio Processing","score":0.12359999865293503,"subfield":{"id":"https://openalex.org/subfields/1711","display_name":"Signal Processing"},"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":0.6409000158309937},{"id":"https://openalex.org/keywords/data-compression","display_name":"Data compression","score":0.6366999745368958},{"id":"https://openalex.org/keywords/redundancy","display_name":"Redundancy (engineering)","score":0.6029999852180481},{"id":"https://openalex.org/keywords/compression","display_name":"Compression (physics)","score":0.534500002861023},{"id":"https://openalex.org/keywords/encoding","display_name":"Encoding (memory)","score":0.45089998841285706},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.44920000433921814},{"id":"https://openalex.org/keywords/matching","display_name":"Matching (statistics)","score":0.40049999952316284},{"id":"https://openalex.org/keywords/code","display_name":"Code (set theory)","score":0.39959999918937683}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8632000088691711},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.6409000158309937},{"id":"https://openalex.org/C78548338","wikidata":"https://www.wikidata.org/wiki/Q2493","display_name":"Data compression","level":2,"score":0.6366999745368958},{"id":"https://openalex.org/C152124472","wikidata":"https://www.wikidata.org/wiki/Q1204361","display_name":"Redundancy (engineering)","level":2,"score":0.6029999852180481},{"id":"https://openalex.org/C180016635","wikidata":"https://www.wikidata.org/wiki/Q2712821","display_name":"Compression (physics)","level":2,"score":0.534500002861023},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.45089998841285706},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.44920000433921814},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.40049999952316284},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.39959999918937683},{"id":"https://openalex.org/C105339364","wikidata":"https://www.wikidata.org/wiki/Q2297740","display_name":"Software deployment","level":2,"score":0.37139999866485596},{"id":"https://openalex.org/C150178126","wikidata":"https://www.wikidata.org/wiki/Q18433212","display_name":"Dynamic range compression","level":2,"score":0.36489999294281006},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.34459999203681946},{"id":"https://openalex.org/C25797200","wikidata":"https://www.wikidata.org/wiki/Q828137","display_name":"Compression ratio","level":3,"score":0.3400999903678894},{"id":"https://openalex.org/C103278499","wikidata":"https://www.wikidata.org/wiki/Q254465","display_name":"Similarity (geometry)","level":3,"score":0.33820000290870667},{"id":"https://openalex.org/C43126263","wikidata":"https://www.wikidata.org/wiki/Q128751","display_name":"Source code","level":2,"score":0.33489999175071716},{"id":"https://openalex.org/C81917197","wikidata":"https://www.wikidata.org/wiki/Q628760","display_name":"Selection (genetic algorithm)","level":2,"score":0.32739999890327454},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.314300000667572},{"id":"https://openalex.org/C81081738","wikidata":"https://www.wikidata.org/wiki/Q55542","display_name":"Lossless compression","level":3,"score":0.30559998750686646},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.29339998960494995},{"id":"https://openalex.org/C94835093","wikidata":"https://www.wikidata.org/wiki/Q3113333","display_name":"Data compression ratio","level":5,"score":0.29319998621940613},{"id":"https://openalex.org/C79403827","wikidata":"https://www.wikidata.org/wiki/Q3988","display_name":"Real-time computing","level":1,"score":0.2851000130176544},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.2842999994754791},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.28130000829696655},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.26510000228881836},{"id":"https://openalex.org/C13481523","wikidata":"https://www.wikidata.org/wiki/Q412438","display_name":"Image compression","level":4,"score":0.26420000195503235},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.25859999656677246}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.12056","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.12056","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.2605.12056","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.12056","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":{"Omnimodal":[0],"large":[1],"language":[2],"models":[3],"(Omni-LLMs)":[4],"show":[5,128],"strong":[6,137],"capability":[7],"in":[8,84],"audio-video":[9,56],"understanding,":[10],"but":[11],"their":[12],"practical":[13],"deployment":[14],"remains":[15],"limited":[16],"by":[17],"high":[18],"inference":[19,63],"cost":[20],"of":[21],"long":[22],"video":[23,111],"streams":[24],"and":[25,50,102,112,139,167],"dense":[26],"audio":[27,113],"sequences.":[28],"Despite":[29],"recent":[30],"progress,":[31],"existing":[32],"compression":[33,43,83,97,145],"methods":[34],"for":[35,55,79],"Omni-LLMs":[36],"typically":[37],"rely":[38],"on":[39],"fixed":[40],"or":[41],"native":[42,91],"units,":[44],"which":[45],"can":[46],"disrupt":[47],"cross-modal":[48],"correspondence":[49],"the":[51,162],"complementary":[52],"information":[53],"required":[54],"reasoning,":[57],"making":[58],"it":[59,149],"difficult":[60],"to":[61,119,172],"improve":[62],"efficiency":[64],"while":[65,122],"stably":[66],"preserving":[67,123],"performance.":[68],"To":[69],"address":[70],"this,":[71],"we":[72],"propose":[73],"OmniRefine,":[74],"a":[75,132,155],"training-free":[76],"two-stage":[77],"framework":[78],"efficient":[80],"audio-visual":[81],"token":[82,157],"Omni-LLMs.":[85],"First,":[86],"Correspondence-Preserving":[87],"Chunk":[88],"Refinement":[89],"refines":[90],"chunk":[92],"boundaries":[93],"into":[94],"cross-modally":[95],"aligned":[96],"units":[98],"through":[99],"frame-audio":[100],"similarity":[101],"dynamic":[103],"programming.":[104],"Second,":[105],"Modality-Aware":[106],"Cooperative":[107],"Compression":[108],"jointly":[109],"compresses":[110],"tokens":[114],"within":[115],"each":[116],"refined":[117],"unit":[118],"reduce":[120],"redundancy":[121],"critical":[124],"evidence.":[125],"Extensive":[126],"experiments":[127],"that":[129],"OmniRefine":[130],"achieves":[131],"better":[133],"efficiency-performance":[134],"trade-off":[135],"than":[136],"baselines":[138],"maintains":[140],"stable":[141],"performance":[142],"under":[143],"lower":[144],"ratios.":[146],"On":[147],"WorldSense,":[148],"still":[150],"reaches":[151],"46.7%":[152],"accuracy":[153],"at":[154],"44%":[156],"retention":[158],"ratio,":[159],"nearly":[160],"matching":[161],"full-token":[163],"baseline.":[164],"The":[165],"code":[166],"interface":[168],"will":[169],"be":[170],"released":[171],"facilitate":[173],"further":[174],"research.":[175]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-05-14T00:00:00"}