{"id":"https://openalex.org/W7156796032","doi":"https://doi.org/10.48550/arxiv.2604.23987","title":"Continual Calibration: Coverage Can Collapse Before Accuracy in Lifelong LLM Fine-Tuning","display_name":"Continual Calibration: Coverage Can Collapse Before Accuracy in Lifelong LLM Fine-Tuning","publication_year":2026,"publication_date":"2026-04-27","ids":{"openalex":"https://openalex.org/W7156796032","doi":"https://doi.org/10.48550/arxiv.2604.23987"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2604.23987","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.23987","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":"article"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2604.23987","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5134780877","display_name":"Ibne Farabi Shihab","orcid":null},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Shihab, Ibne Farabi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5119832027","display_name":"Sanjeda Akter","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Akter, Sanjeda","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5134767452","display_name":"Anuj Sharma","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sharma, Anuj","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":3,"corresponding_author_ids":["https://openalex.org/A5134780877"],"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.44449999928474426,"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"}},"topics":[{"id":"https://openalex.org/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.44449999928474426,"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/T10028","display_name":"Topic Modeling","score":0.1168999969959259,"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/T12072","display_name":"Machine Learning and Algorithms","score":0.0738999992609024,"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/calibration","display_name":"Calibration","score":0.6693000197410583},{"id":"https://openalex.org/keywords/reliability","display_name":"Reliability (semiconductor)","score":0.6047000288963318},{"id":"https://openalex.org/keywords/perspective","display_name":"Perspective (graphical)","score":0.5677000284194946},{"id":"https://openalex.org/keywords/conformal-map","display_name":"Conformal map","score":0.4577000141143799},{"id":"https://openalex.org/keywords/task","display_name":"Task (project management)","score":0.34619998931884766},{"id":"https://openalex.org/keywords/channel","display_name":"Channel (broadcasting)","score":0.3361000120639801}],"concepts":[{"id":"https://openalex.org/C165838908","wikidata":"https://www.wikidata.org/wiki/Q736777","display_name":"Calibration","level":2,"score":0.6693000197410583},{"id":"https://openalex.org/C43214815","wikidata":"https://www.wikidata.org/wiki/Q7310987","display_name":"Reliability (semiconductor)","level":3,"score":0.6047000288963318},{"id":"https://openalex.org/C12713177","wikidata":"https://www.wikidata.org/wiki/Q1900281","display_name":"Perspective (graphical)","level":2,"score":0.5677000284194946},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5339000225067139},{"id":"https://openalex.org/C98214594","wikidata":"https://www.wikidata.org/wiki/Q850275","display_name":"Conformal map","level":2,"score":0.4577000141143799},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.4172999858856201},{"id":"https://openalex.org/C149782125","wikidata":"https://www.wikidata.org/wiki/Q160039","display_name":"Econometrics","level":1,"score":0.400299996137619},{"id":"https://openalex.org/C105795698","wikidata":"https://www.wikidata.org/wiki/Q12483","display_name":"Statistics","level":1,"score":0.35580000281333923},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.34619998931884766},{"id":"https://openalex.org/C127162648","wikidata":"https://www.wikidata.org/wiki/Q16858953","display_name":"Channel (broadcasting)","level":2,"score":0.3361000120639801},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.32359999418258667},{"id":"https://openalex.org/C120936955","wikidata":"https://www.wikidata.org/wiki/Q2155640","display_name":"Empirical research","level":2,"score":0.323199987411499},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.30300000309944153},{"id":"https://openalex.org/C12725497","wikidata":"https://www.wikidata.org/wiki/Q810247","display_name":"Baseline (sea)","level":2,"score":0.2847999930381775},{"id":"https://openalex.org/C34559072","wikidata":"https://www.wikidata.org/wiki/Q2334061","display_name":"Design of experiments","level":2,"score":0.2809999883174896},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.2784999907016754},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.27730000019073486},{"id":"https://openalex.org/C2777027219","wikidata":"https://www.wikidata.org/wiki/Q1284190","display_name":"Constant (computer programming)","level":2,"score":0.272599995136261},{"id":"https://openalex.org/C200601418","wikidata":"https://www.wikidata.org/wiki/Q2193887","display_name":"Reliability engineering","level":1,"score":0.2612000107765198},{"id":"https://openalex.org/C2777152325","wikidata":"https://www.wikidata.org/wiki/Q108163","display_name":"Proposition","level":2,"score":0.2563000023365021},{"id":"https://openalex.org/C6571938","wikidata":"https://www.wikidata.org/wiki/Q3274486","display_name":"Omitted-variable bias","level":2,"score":0.2515999972820282}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2604.23987","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.23987","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":"article"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2604.23987","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.23987","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":"article"},"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":{"Continual":[0],"learning":[1],"for":[2,222],"large":[3],"language":[4],"models":[5,48],"is":[6,20],"typically":[7,173],"evaluated":[8],"through":[9],"accuracy":[10,73,99,111],"retention":[11],"under":[12,147,205],"sequential":[13],"fine-tuning.":[14],"We":[15,34,127,188],"argue":[16],"that":[17,109,135],"this":[18,36],"perspective":[19],"incomplete,":[21],"because":[22],"uncertainty":[23],"reliability":[24],"can":[25],"degrade":[26],"earlier":[27],"and":[28,42,53,61,117,141,172,207],"more":[29],"sharply":[30],"than":[31,162],"top-1":[32],"performance.":[33],"study":[35,192],"empirically":[37],"by":[38,75],"measuring":[39],"conformal":[40,145,203],"coverage":[41,70,92,175],"calibration":[43,119,129],"error":[44],"on":[45,83],"sequentially":[46],"fine-tuned":[47],"across":[49,85],"three":[50,102],"model":[51,150],"families":[52],"eight":[54],"task":[55],"sequences":[56],"drawn":[57],"primarily":[58],"from":[59,94],"classification":[60],"multiple-choice":[62],"benchmarks.":[63],"Across":[64],"the":[65,88,125,148,166,190],"classification-style":[66,223],"settings":[67],"we":[68],"study,":[69],"loss":[71,74],"exceeds":[72],"a":[76,131,137,143,194,197,208],"factor":[77],"of":[78,104,124,165,168,180],"roughly":[79],"\\(3.4\\times":[80],"\\pm":[81],"0.5\\times\\)":[82],"average":[84],"seeds;":[86],"in":[87],"most":[89],"pronounced":[90],"case,":[91],"drops":[93],"\\(0.92\\)":[95],"to":[96,176,229],"\\(0.61\\),":[97],"while":[98],"remains":[100],"within":[101,177],"points":[103,179],"baseline.":[105],"Standard":[106],"continual-learning":[107],"methods":[108],"preserve":[110,115],"do":[112,215],"not":[113,216],"automatically":[114],"coverage,":[116],"naive":[118],"baselines":[120],"recover":[121],"only":[122],"part":[123],"gap.":[126],"propose":[128],"replay,":[130,171],"lightweight":[132],"post-hoc":[133],"procedure":[134],"maintains":[136],"task-specific":[138,144,226],"held-out":[139],"buffer":[140,183],"refits":[142],"threshold":[146],"current":[149],"after":[151],"each":[152],"update.":[153],"It":[154],"adds":[155],"no":[156],"training-time":[157],"gradient":[158],"cost,":[159],"uses":[160],"less":[161],"one":[163],"percent":[164],"memory":[167],"ordinary":[169],"experience":[170],"restores":[174],"two":[178],"nominal":[181],"at":[182],"size":[184],"\\(m":[185],"=":[186],"200\\).":[187],"accompany":[189],"empirical":[191],"with":[193,225],"drift":[195],"decomposition,":[196],"finite-sample":[198],"recovery":[199],"theorem":[200],"showing":[201],"exact":[202],"validity":[204],"exchangeability,":[206],"mixture-validity":[209],"proposition":[210],"explaining":[211],"why":[212],"pooled":[213],"thresholds":[214],"suffice.":[217],"Our":[218],"guarantees":[219],"are":[220,232],"stated":[221],"tasks":[224],"buffers;":[227],"extensions":[228],"open-ended":[230],"generation":[231],"exploratory.":[233]},"counts_by_year":[],"updated_date":"2026-04-29T06:16:36.941037","created_date":"2026-04-29T00:00:00"}