{"id":"https://openalex.org/W7138434327","doi":"https://doi.org/10.48550/arxiv.2603.14493","title":"Fine-tuning MLLMs Without Forgetting Is Easier Than You Think","display_name":"Fine-tuning MLLMs Without Forgetting Is Easier Than You Think","publication_year":2026,"publication_date":"2026-03-15","ids":{"openalex":"https://openalex.org/W7138434327","doi":"https://doi.org/10.48550/arxiv.2603.14493"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2603.14493","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.14493","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.2603.14493","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5129747217","display_name":"He Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, He","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129688424","display_name":"Yuhui Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Yuhui","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129691050","display_name":"Xiaohan Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Xiaohan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129684686","display_name":"Kaifeng Lyu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lyu, Kaifeng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5129679077","display_name":"Serena Yeung-Levy","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yeung-Levy, Serena","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":0.8891000151634216,"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":0.8891000151634216,"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":0.07249999791383743,"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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.006500000134110451,"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/forgetting","display_name":"Forgetting","score":0.9100000262260437},{"id":"https://openalex.org/keywords/overfitting","display_name":"Overfitting","score":0.838699996471405},{"id":"https://openalex.org/keywords/robustness","display_name":"Robustness (evolution)","score":0.6324999928474426},{"id":"https://openalex.org/keywords/simple","display_name":"Simple (philosophy)","score":0.4307999908924103},{"id":"https://openalex.org/keywords/generalization","display_name":"Generalization","score":0.3815999925136566}],"concepts":[{"id":"https://openalex.org/C7149132","wikidata":"https://www.wikidata.org/wiki/Q1377840","display_name":"Forgetting","level":2,"score":0.9100000262260437},{"id":"https://openalex.org/C22019652","wikidata":"https://www.wikidata.org/wiki/Q331309","display_name":"Overfitting","level":3,"score":0.838699996471405},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7228000164031982},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.6324999928474426},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5952000021934509},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.453900009393692},{"id":"https://openalex.org/C2780586882","wikidata":"https://www.wikidata.org/wiki/Q7520643","display_name":"Simple (philosophy)","level":2,"score":0.4307999908924103},{"id":"https://openalex.org/C177148314","wikidata":"https://www.wikidata.org/wiki/Q170084","display_name":"Generalization","level":2,"score":0.3815999925136566},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.26339998841285706},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.251800000667572}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2603.14493","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.14493","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.2603.14493","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.14493","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":{"The":[0],"paper":[1],"demonstrate":[2,112],"that":[3,47,105,113],"simple":[4],"adjustments":[5],"of":[6,10,55,78,141],"the":[7,53,133,138],"fine-tuning":[8],"recipes":[9],"multimodal":[11],"large":[12],"language":[13],"models":[14],"(MLLM)":[15],"are":[16],"sufficient":[17],"to":[18,32,118],"mitigate":[19],"catastrophic":[20],"forgetting.":[21],"On":[22],"visual":[23],"question":[24],"answering,":[25],"we":[26,73,111],"design":[27],"a":[28,60,75,101],"2x2":[29],"experimental":[30],"framework":[31],"assess":[33],"model":[34],"performance":[35],"across":[36],"in-distribution":[37,83],"and":[38,41,85,95,108,143],"out-of-distribution":[39,70,86],"image":[40],"text":[42],"inputs.":[43],"Our":[44],"results":[45],"show":[46],"appropriate":[48],"regularization,":[49],"such":[50],"as":[51,92],"constraining":[52],"number":[54],"trainable":[56],"parameters":[57],"or":[58],"adopting":[59],"low":[61],"learning":[62],"rate,":[63],"effectively":[64],"prevents":[65],"forgetting":[66,79,91],"when":[67],"dealing":[68],"with":[69,82,124],"images.":[71],"However,":[72],"uncover":[74],"distinct":[76],"form":[77],"in":[80],"settings":[81],"images":[84],"text.":[87],"We":[88],"attribute":[89],"this":[90,97,114],"task-specific":[93],"overfitting":[94],"address":[96],"issue":[98],"by":[99,136],"introducing":[100],"data-hybrid":[102],"training":[103],"strategy":[104],"combines":[106],"datasets":[107],"tasks.":[109],"Finally,":[110],"approach":[115],"naturally":[116],"extends":[117],"continual":[119],"learning,":[120],"outperforming":[121],"existing":[122],"methods":[123],"complex":[125],"auxiliary":[126],"mechanisms.":[127],"In":[128],"general,":[129],"our":[130],"findings":[131],"challenge":[132],"prevailing":[134],"assumptions":[135],"highlighting":[137],"inherent":[139],"robustness":[140],"MLLMs":[142],"providing":[144],"practical":[145],"guidelines":[146],"for":[147],"adapting":[148],"them":[149],"while":[150],"preserving":[151],"their":[152],"general":[153],"capabilities.":[154]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-03-18T00:00:00"}