{"id":"https://openalex.org/W7165383970","doi":"https://doi.org/10.48550/arxiv.2606.20246","title":"Finetuning Vision-Language-Action Models Requires Fewer Layers Than You Think","display_name":"Finetuning Vision-Language-Action Models Requires Fewer Layers Than You Think","publication_year":2026,"publication_date":"2026-06-18","ids":{"openalex":"https://openalex.org/W7165383970","doi":"https://doi.org/10.48550/arxiv.2606.20246"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.20246","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.20246","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":"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.2606.20246","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5139017266","display_name":"Gia-Binh Nguyen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Nguyen, Gia-Binh","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101944191","display_name":"Trong-Bao Ho","orcid":"https://orcid.org/0009-0002-1794-2814"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ho, Trong-Bao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139007748","display_name":"Thien-Loc Ha","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ha, Thien-Loc","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139010456","display_name":"Khoa Vo","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Vo, Khoa","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5104217175","display_name":"Philip Lund M\u00f8ller","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"M\u00f8ller, Philip Lund","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139000052","display_name":"Quang T. Nguyen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Nguyen, Quang T.","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5113069976","display_name":"Long Dinh","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Dinh, Long","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5061282536","display_name":"Tuan Dam","orcid":"https://orcid.org/0000-0001-7422-139X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Luu, Tung M.","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139020707","display_name":"Vu Duong","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Dam, Tuan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138393001","display_name":"Tung M. Luu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Duong, Vu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138990618","display_name":"Trung Le","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Le, Trung","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5124124270","display_name":"Tran Nguyen Le","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Bui, Nghi D. Q.","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138981190","display_name":"Minh Vu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Vu, Minh","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102888876","display_name":"An T. Le","orcid":"https://orcid.org/0000-0003-0929-3316"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Le, Tran Nguyen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139000982","display_name":"Ngan Le","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Le, An Thai","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5069432042","display_name":"Daniel Sonntag","orcid":"https://orcid.org/0000-0002-8857-8709"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Le, Ngan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138974337","display_name":"James Zou","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sonntag, Daniel","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138964124","display_name":"Jan Peters","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zou, James","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5050539353","display_name":"Duy M. H. Nguyen","orcid":"https://orcid.org/0009-0008-6007-7786"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Peters, Jan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5043775193","display_name":"Ngo Anh Vien","orcid":"https://orcid.org/0000-0001-9646-267X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Nguyen, Duy M. H.","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":null,"display_name":"Vien, Ngo Anh","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Vien, Ngo Anh","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":0.7419000267982483,"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":0.7419000267982483,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.04560000076889992,"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/T10462","display_name":"Reinforcement Learning in Robotics","score":0.041200000792741776,"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/pipeline","display_name":"Pipeline (software)","score":0.6700000166893005},{"id":"https://openalex.org/keywords/exploit","display_name":"Exploit","score":0.6450999975204468},{"id":"https://openalex.org/keywords/scalability","display_name":"Scalability","score":0.5845999717712402},{"id":"https://openalex.org/keywords/dependency","display_name":"Dependency (UML)","score":0.5149999856948853},{"id":"https://openalex.org/keywords/layer","display_name":"Layer (electronics)","score":0.4726000130176544},{"id":"https://openalex.org/keywords/reduction","display_name":"Reduction (mathematics)","score":0.41999998688697815},{"id":"https://openalex.org/keywords/kernel","display_name":"Kernel (algebra)","score":0.413100004196167},{"id":"https://openalex.org/keywords/downstream","display_name":"Downstream (manufacturing)","score":0.4007999897003174}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7692999839782715},{"id":"https://openalex.org/C43521106","wikidata":"https://www.wikidata.org/wiki/Q2165493","display_name":"Pipeline (software)","level":2,"score":0.6700000166893005},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.6450999975204468},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.5845999717712402},{"id":"https://openalex.org/C19768560","wikidata":"https://www.wikidata.org/wiki/Q320727","display_name":"Dependency (UML)","level":2,"score":0.5149999856948853},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.48649999499320984},{"id":"https://openalex.org/C2779227376","wikidata":"https://www.wikidata.org/wiki/Q6505497","display_name":"Layer (electronics)","level":2,"score":0.4726000130176544},{"id":"https://openalex.org/C111335779","wikidata":"https://www.wikidata.org/wiki/Q3454686","display_name":"Reduction (mathematics)","level":2,"score":0.41999998688697815},{"id":"https://openalex.org/C74193536","wikidata":"https://www.wikidata.org/wiki/Q574844","display_name":"Kernel (algebra)","level":2,"score":0.413100004196167},{"id":"https://openalex.org/C2776207758","wikidata":"https://www.wikidata.org/wiki/Q5303302","display_name":"Downstream (manufacturing)","level":2,"score":0.4007999897003174},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.40070000290870667},{"id":"https://openalex.org/C90509273","wikidata":"https://www.wikidata.org/wiki/Q11012","display_name":"Robot","level":2,"score":0.3991999924182892},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.3644999861717224},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.364300012588501},{"id":"https://openalex.org/C2775924081","wikidata":"https://www.wikidata.org/wiki/Q55608371","display_name":"Control (management)","level":2,"score":0.33009999990463257},{"id":"https://openalex.org/C123657996","wikidata":"https://www.wikidata.org/wiki/Q12271","display_name":"Architecture","level":2,"score":0.31150001287460327},{"id":"https://openalex.org/C144559511","wikidata":"https://www.wikidata.org/wiki/Q2986279","display_name":"Principal (computer security)","level":2,"score":0.3086000084877014},{"id":"https://openalex.org/C2780980858","wikidata":"https://www.wikidata.org/wiki/Q110022","display_name":"Dual (grammatical number)","level":2,"score":0.3043000102043152},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.29600000381469727},{"id":"https://openalex.org/C42058472","wikidata":"https://www.wikidata.org/wiki/Q810214","display_name":"Base (topology)","level":2,"score":0.2955999970436096},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.2556000053882599},{"id":"https://openalex.org/C34413123","wikidata":"https://www.wikidata.org/wiki/Q170978","display_name":"Robotics","level":3,"score":0.2522999942302704}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.20246","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.20246","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":"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.2606.20246","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.20246","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":"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":{"Vision-Language-Action":[0],"(VLA)":[1],"models":[2,81],"pre-trained":[3],"on":[4,49],"massive":[5],"video-robot":[6],"datasets":[7],"have":[8],"revolutionized":[9],"robotic":[10,185],"manipulation,":[11],"yet":[12],"their":[13],"multi-billion":[14],"parameter":[15],"architectures":[16],"impose":[17],"prohibitive":[18],"computational":[19],"burdens":[20],"during":[21],"downstream":[22],"fine-tuning":[23,133],"and":[24,126,149,176],"real-time":[25,154],"inference.":[26],"In":[27],"this":[28,135],"work,":[29],"we":[30,62,107],"reveal":[31],"a":[32,64,94,139,143,201],"highly":[33,202],"non-trivial":[34],"architectural":[35],"characteristic":[36],"of":[37,75,134],"these":[38],"continuous":[39,128],"control":[40,129],"foundation":[41],"policies":[42],"(e.g.,":[43],"pi_0,":[44],"GR00T-N1.5):":[45],"despite":[46],"being":[47],"trained":[48],"diverse":[50,178],"physical":[51],"trajectories,":[52],"they":[53],"exhibit":[54],"severe":[55],"layer-wise":[56],"representational":[57],"redundancy.":[58],"To":[59],"exploit":[60],"this,":[61],"introduce":[63],"structural":[65],"compression":[66],"pipeline":[67],"that":[68,190],"is":[69],"entirely":[70],"training-free,":[71],"bypassing":[72],"the":[73,114,123,127],"need":[74],"existing":[76],"methods":[77],"to":[78,82,102,111,119,151],"load":[79],"full-scale":[80,160],"learn":[83],"optimized":[84],"token":[85],"reductions":[86],"or":[87,158],"dynamic":[88],"layer":[89,105],"selectors.":[90],"Instead,":[91],"using":[92],"only":[93],"single":[95],"forward":[96],"pass":[97],"via":[98],"Centered":[99],"Kernel":[100],"Alignment":[101],"identify":[103],"redundant":[104],"features,":[106],"remove":[108],"twin":[109],"layers":[110,196],"permanently":[112],"compress":[113],"model":[115,162],"depth":[116],"by":[117],"up":[118,150],"50%":[120],"across":[121,169,182],"both":[122],"VLM":[124],"backbone":[125],"policy":[130],"head.":[131],"Downstream":[132],"streamlined":[136],"architecture":[137],"yields":[138],"dual":[140],"acceleration":[141],"benefit:":[142],"40-50%":[144],"reduction":[145],"in":[146],"training":[147],"time":[148],"30%":[152],"faster":[153],"inference,":[155],"while":[156],"matching":[157],"exceeding":[159],"base":[161],"performance.":[163],"We":[164],"comprehensively":[165],"validate":[166],"our":[167],"method":[168],"three":[170],"simulation":[171],"benchmarks":[172],"(LIBERO,":[173],"RoboCasa,":[174],"SimplerEnv)":[175],"10":[177],"real-world":[179],"manipulation":[180],"tasks":[181],"4":[183],"unique":[184],"embodiments.":[186],"These":[187],"results":[188],"prove":[189],"advanced":[191],"VLAs":[192],"require":[193],"significantly":[194],"fewer":[195],"than":[197],"previously":[198],"assumed,":[199],"offering":[200],"compute-efficient":[203],"paradigm":[204],"for":[205],"scalable":[206],"robot":[207],"learning.":[208]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-06-20T00:00:00"}