{"id":"https://openalex.org/W7141171901","doi":"https://doi.org/10.48550/arxiv.2603.24941","title":"Beyond Attention Magnitude: Leveraging Inter-layer Rank Consistency for Efficient Vision-Language-Action Models","display_name":"Beyond Attention Magnitude: Leveraging Inter-layer Rank Consistency for Efficient Vision-Language-Action Models","publication_year":2026,"publication_date":"2026-03-26","ids":{"openalex":"https://openalex.org/W7141171901","doi":"https://doi.org/10.48550/arxiv.2603.24941"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2603.24941","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.24941","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.2603.24941","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5130721809","display_name":"Peiju Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Peiju","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5130807219","display_name":"Jinming Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Jinming","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5130731994","display_name":"Xipeng Qiu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Qiu, Xipeng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5130808210","display_name":"Xuanjing Huang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Huang, Xuanjing","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":0.6819999814033508,"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":0.6819999814033508,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.07580000162124634,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.04639999940991402,"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/security-token","display_name":"Security token","score":0.8052999973297119},{"id":"https://openalex.org/keywords/ranking","display_name":"Ranking (information retrieval)","score":0.6090999841690063},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.579800009727478},{"id":"https://openalex.org/keywords/consistency","display_name":"Consistency (knowledge bases)","score":0.5575000047683716},{"id":"https://openalex.org/keywords/generalization","display_name":"Generalization","score":0.548799991607666},{"id":"https://openalex.org/keywords/rank","display_name":"Rank (graph theory)","score":0.4681999981403351},{"id":"https://openalex.org/keywords/latency","display_name":"Latency (audio)","score":0.4659000039100647}],"concepts":[{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.8052999973297119},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7894999980926514},{"id":"https://openalex.org/C189430467","wikidata":"https://www.wikidata.org/wiki/Q7293293","display_name":"Ranking (information retrieval)","level":2,"score":0.6090999841690063},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.579800009727478},{"id":"https://openalex.org/C2776436953","wikidata":"https://www.wikidata.org/wiki/Q5163215","display_name":"Consistency (knowledge bases)","level":2,"score":0.5575000047683716},{"id":"https://openalex.org/C177148314","wikidata":"https://www.wikidata.org/wiki/Q170084","display_name":"Generalization","level":2,"score":0.548799991607666},{"id":"https://openalex.org/C164226766","wikidata":"https://www.wikidata.org/wiki/Q7293202","display_name":"Rank (graph theory)","level":2,"score":0.4681999981403351},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.4659000039100647},{"id":"https://openalex.org/C81917197","wikidata":"https://www.wikidata.org/wiki/Q628760","display_name":"Selection (genetic algorithm)","level":2,"score":0.43549999594688416},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.41769999265670776},{"id":"https://openalex.org/C111335779","wikidata":"https://www.wikidata.org/wiki/Q3454686","display_name":"Reduction (mathematics)","level":2,"score":0.41659998893737793},{"id":"https://openalex.org/C86037889","wikidata":"https://www.wikidata.org/wiki/Q4330127","display_name":"Learning to rank","level":3,"score":0.34220001101493835},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3368000090122223},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.31290000677108765},{"id":"https://openalex.org/C115067241","wikidata":"https://www.wikidata.org/wiki/Q1639854","display_name":"Token passing","level":3,"score":0.2856999933719635},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.2712000012397766},{"id":"https://openalex.org/C2777131613","wikidata":"https://www.wikidata.org/wiki/Q18394147","display_name":"Retard","level":2,"score":0.26440000534057617},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.2615000009536743}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2603.24941","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.24941","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.2603.24941","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.24941","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":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Vision-Language-Action":[0],"(VLA)":[1],"models":[2],"excel":[3],"in":[4],"robotic":[5],"manipulation":[6],"but":[7],"suffer":[8],"from":[9],"significant":[10],"inference":[11],"latency":[12],"due":[13],"to":[14],"processing":[15],"dense":[16],"visual":[17],"tokens.":[18],"Existing":[19],"token":[20,67,81,102],"reduction":[21],"methods":[22],"predominantly":[23],"rely":[24],"on":[25],"attention":[26,73],"magnitude":[27,74],"as":[28],"a":[29,61],"static":[30],"selection.":[31],"In":[32],"this":[33,37],"work,":[34],"we":[35,54],"challenge":[36],"assumption,":[38],"revealing":[39],"that":[40],"high-attention":[41],"tokens":[42],"are":[43],"task-dependent":[44],"and":[45,106,113],"can":[46],"even":[47],"degrade":[48],"policy":[49],"performance.":[50],"To":[51],"address":[52],"this,":[53],"introduce":[55],"\\textbf{TIES}":[56],"(\\textbf{T}au-guided":[57],"\\textbf{I}nter-layer":[58],"\\textbf{E}fficient":[59],"\\textbf{S}election),":[60],"dynamic":[62],"framework":[63],"guided":[64],"by":[65,98,104],"inter-layer":[66],"ranking":[68,76],"consistency.":[69],"By":[70],"adaptively":[71],"balancing":[72],"with":[75],"consistency,":[77],"TIES":[78,93],"ensures":[79],"robust":[80],"selection":[82],"without":[83],"requiring":[84],"additional":[85],"training.":[86],"On":[87],"the":[88],"CogACT":[89],"+":[90],"SIMPLER":[91],"benchmark,":[92],"improves":[94],"average":[95],"success":[96],"rates":[97],"6\\%":[99],"while":[100],"reducing":[101],"usage":[103],"78\\%,":[105],"demonstrate":[107],"strong":[108],"generalization":[109],"across":[110],"diverse":[111],"decoders":[112],"benchmarks.":[114]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-03-28T00:00:00"}