{"id":"https://openalex.org/W7138942527","doi":"https://doi.org/10.48550/arxiv.2603.17541","title":"Temporal Gains, Spatial Costs: Revisiting Video Fine-Tuning in Multimodal Large Language Models","display_name":"Temporal Gains, Spatial Costs: Revisiting Video Fine-Tuning in Multimodal Large Language Models","publication_year":2026,"publication_date":"2026-03-18","ids":{"openalex":"https://openalex.org/W7138942527","doi":"https://doi.org/10.48550/arxiv.2603.17541"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2603.17541","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.17541","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":"article"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2603.17541","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5129956897","display_name":"Linghao Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Zhang, Linghao","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129930813","display_name":"Jungang Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Jungang","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5119938674","display_name":"Yonghua Hei","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hei, Yonghua","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Tao, Sicheng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tao, Sicheng","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100824896","display_name":"Song Dai","orcid":"https://orcid.org/0009-0003-5413-7635"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Dai, Song","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5130096111","display_name":"Yibo Yan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yan, Yibo","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5130041279","display_name":"Zihao Dongfang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Dongfang, Zihao","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5124130589","display_name":"Weiting Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Weiting","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129940803","display_name":"Chenxi Qin","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Qin, Chenxi","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5130049429","display_name":"Hanqian Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Hanqian","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129971638","display_name":"Xin Zou","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zou, Xin","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5130171071","display_name":"Jiahao Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Jiahao","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129866100","display_name":"Shuhang Xun","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xun, Shuhang","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129854418","display_name":"Haiyun Jiang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jiang, Haiyun","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5129859983","display_name":"Xuming Hu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hu, Xuming","raw_affiliation_strings":[],"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":15,"corresponding_author_ids":["https://openalex.org/A5129956897"],"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":0.8073999881744385,"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":0.8073999881744385,"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":0.05889999866485596,"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":0.05260000005364418,"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/frame","display_name":"Frame (networking)","score":0.5796999931335449},{"id":"https://openalex.org/keywords/joint","display_name":"Joint (building)","score":0.5631999969482422},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.4666000008583069},{"id":"https://openalex.org/keywords/sampling","display_name":"Sampling (signal processing)","score":0.4296000003814697},{"id":"https://openalex.org/keywords/frame-rate","display_name":"Frame rate","score":0.3391000032424927},{"id":"https://openalex.org/keywords/spatial-analysis","display_name":"Spatial analysis","score":0.3192000091075897},{"id":"https://openalex.org/keywords/image","display_name":"Image (mathematics)","score":0.3190000057220459}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8059999942779541},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.6571000218391418},{"id":"https://openalex.org/C126042441","wikidata":"https://www.wikidata.org/wiki/Q1324888","display_name":"Frame (networking)","level":2,"score":0.5796999931335449},{"id":"https://openalex.org/C18555067","wikidata":"https://www.wikidata.org/wiki/Q8375051","display_name":"Joint (building)","level":2,"score":0.5631999969482422},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.5126000046730042},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.4666000008583069},{"id":"https://openalex.org/C140779682","wikidata":"https://www.wikidata.org/wiki/Q210868","display_name":"Sampling (signal processing)","level":3,"score":0.4296000003814697},{"id":"https://openalex.org/C3261483","wikidata":"https://www.wikidata.org/wiki/Q119565","display_name":"Frame rate","level":2,"score":0.3391000032424927},{"id":"https://openalex.org/C159620131","wikidata":"https://www.wikidata.org/wiki/Q1938983","display_name":"Spatial analysis","level":2,"score":0.3192000091075897},{"id":"https://openalex.org/C115961682","wikidata":"https://www.wikidata.org/wiki/Q860623","display_name":"Image (mathematics)","level":2,"score":0.3190000057220459},{"id":"https://openalex.org/C168031717","wikidata":"https://www.wikidata.org/wiki/Q1530280","display_name":"Balance (ability)","level":2,"score":0.3118000030517578},{"id":"https://openalex.org/C160086991","wikidata":"https://www.wikidata.org/wiki/Q5939193","display_name":"Human visual system model","level":3,"score":0.29269999265670776},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.28929999470710754},{"id":"https://openalex.org/C160633673","wikidata":"https://www.wikidata.org/wiki/Q355198","display_name":"Pixel","level":2,"score":0.28130000829696655},{"id":"https://openalex.org/C2780139006","wikidata":"https://www.wikidata.org/wiki/Q1493902","display_name":"Key frame","level":3,"score":0.2752000093460083},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.2720000147819519},{"id":"https://openalex.org/C2986089797","wikidata":"https://www.wikidata.org/wiki/Q6501338","display_name":"Visual attention","level":3,"score":0.25519999861717224},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.25049999356269836}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2603.17541","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.17541","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":"article"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2603.17541","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.17541","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":"article"},"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":{"Multimodal":[0],"large":[1],"language":[2],"models":[3],"(MLLMs)":[4],"are":[5],"typically":[6],"trained":[7],"in":[8,57,160],"multiple":[9],"stages,":[10],"with":[11],"video-based":[12],"supervised":[13],"fine-tuning":[14],"(Video-SFT)":[15],"serving":[16],"as":[17],"a":[18,69,147,157],"key":[19],"step":[20],"for":[21,150],"improving":[22],"visual":[23,33,55],"understanding.":[24],"Yet":[25],"its":[26],"effect":[27],"on":[28,85],"the":[29,36,102,137],"fine-grained":[30],"evolution":[31],"of":[32,104],"capabilities,":[34],"particularly":[35],"balance":[37],"between":[38],"spatial":[39,154],"and":[40,63,134,152],"temporal":[41,99],"understanding,":[42],"remains":[43,156],"poorly":[44],"understood.":[45],"In":[46],"this":[47,93,121],"paper,":[48],"we":[49,67,123],"systematically":[50],"study":[51,124],"how":[52],"Video-SFT":[53,72,144],"reshapes":[54],"capabilities":[56],"MLLMs.":[58],"Across":[59],"architectures,":[60],"parameter":[61],"scales,":[62],"frame":[64,132],"sampling":[65],"settings,":[66],"observe":[68],"consistent":[70],"pattern:":[71],"reliably":[73,114],"improves":[74,108],"video":[75,109],"performance,":[76,110],"but":[77,111],"often":[78],"yields":[79],"limited":[80],"gains":[81],"or":[82],"even":[83],"degradation":[84],"static":[86,116],"image":[87,117],"benchmarks.":[88],"We":[89],"further":[90],"show":[91],"that":[92,129,143],"trade-off":[94],"is":[95,145],"closely":[96],"tied":[97],"to":[98],"budget:":[100],"increasing":[101],"number":[103],"sampled":[105],"frames":[106],"generally":[107],"does":[112],"not":[113,146],"improve":[115],"performance.":[118],"Motivated":[119],"by":[120],"finding,":[122],"an":[125],"instruction-aware":[126],"Hybrid-Frame":[127],"strategy":[128],"adaptively":[130],"allocates":[131],"counts":[133],"partially":[135],"mitigates":[136],"image-video":[138,162],"trade-off.":[139],"Our":[140],"results":[141],"indicate":[142],"free":[148],"lunch":[149],"MLLMs,":[151],"preserving":[153],"understanding":[155],"central":[158],"challenge":[159],"joint":[161],"training.":[163]},"counts_by_year":[],"updated_date":"2026-04-21T08:09:41.155169","created_date":"2026-03-20T00:00:00"}