{"id":"https://openalex.org/W7164927339","doi":"https://doi.org/10.48550/arxiv.2606.16511","title":"Tail-Shape Estimation in LLM Evaluation Is Fragile: A Protocol for Diagnosing False Positives","display_name":"Tail-Shape Estimation in LLM Evaluation Is Fragile: A Protocol for Diagnosing False Positives","publication_year":2026,"publication_date":"2026-06-15","ids":{"openalex":"https://openalex.org/W7164927339","doi":"https://doi.org/10.48550/arxiv.2606.16511"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.16511","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.16511","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"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.2606.16511","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5114638048","display_name":"Luca Zhou","orcid":null},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Zhou, Luca","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":["https://openalex.org/A5114638048"],"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/T12026","display_name":"Explainable Artificial Intelligence (XAI)","score":0.11819999665021896,"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/T12026","display_name":"Explainable Artificial Intelligence (XAI)","score":0.11819999665021896,"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":0.09640000015497208,"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/T11943","display_name":"Pharmacovigilance and Adverse Drug Reactions","score":0.07249999791383743,"subfield":{"id":"https://openalex.org/subfields/3005","display_name":"Toxicology"},"field":{"id":"https://openalex.org/fields/30","display_name":"Pharmacology, Toxicology and Pharmaceutics"},"domain":{"id":"https://openalex.org/domains/1","display_name":"Life Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/protocol","display_name":"Protocol (science)","score":0.7883999943733215},{"id":"https://openalex.org/keywords/false-positive-paradox","display_name":"False positive paradox","score":0.7096999883651733},{"id":"https://openalex.org/keywords/discriminative-model","display_name":"Discriminative model","score":0.6635000109672546},{"id":"https://openalex.org/keywords/headline","display_name":"Headline","score":0.5432999730110168},{"id":"https://openalex.org/keywords/false-positives-and-false-negatives","display_name":"False positives and false negatives","score":0.48260000348091125},{"id":"https://openalex.org/keywords/point","display_name":"Point (geometry)","score":0.4009000062942505},{"id":"https://openalex.org/keywords/statistic","display_name":"Statistic","score":0.3912999927997589}],"concepts":[{"id":"https://openalex.org/C2780385302","wikidata":"https://www.wikidata.org/wiki/Q367158","display_name":"Protocol (science)","level":3,"score":0.7883999943733215},{"id":"https://openalex.org/C64869954","wikidata":"https://www.wikidata.org/wiki/Q1859747","display_name":"False positive paradox","level":2,"score":0.7096999883651733},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6794000267982483},{"id":"https://openalex.org/C97931131","wikidata":"https://www.wikidata.org/wiki/Q5282087","display_name":"Discriminative model","level":2,"score":0.6635000109672546},{"id":"https://openalex.org/C2778689934","wikidata":"https://www.wikidata.org/wiki/Q1313396","display_name":"Headline","level":2,"score":0.5432999730110168},{"id":"https://openalex.org/C112789634","wikidata":"https://www.wikidata.org/wiki/Q18207010","display_name":"False positives and false negatives","level":3,"score":0.48260000348091125},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4410000145435333},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.4185999929904938},{"id":"https://openalex.org/C28719098","wikidata":"https://www.wikidata.org/wiki/Q44946","display_name":"Point (geometry)","level":2,"score":0.4009000062942505},{"id":"https://openalex.org/C89128539","wikidata":"https://www.wikidata.org/wiki/Q1949963","display_name":"Statistic","level":2,"score":0.3912999927997589},{"id":"https://openalex.org/C2989486834","wikidata":"https://www.wikidata.org/wiki/Q3808900","display_name":"True positive rate","level":2,"score":0.38019999861717224},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.35409998893737793},{"id":"https://openalex.org/C165838908","wikidata":"https://www.wikidata.org/wiki/Q736777","display_name":"Calibration","level":2,"score":0.32829999923706055},{"id":"https://openalex.org/C96250715","wikidata":"https://www.wikidata.org/wiki/Q965330","display_name":"Estimation","level":2,"score":0.31859999895095825},{"id":"https://openalex.org/C105795698","wikidata":"https://www.wikidata.org/wiki/Q12483","display_name":"Statistics","level":1,"score":0.28189998865127563},{"id":"https://openalex.org/C77618280","wikidata":"https://www.wikidata.org/wiki/Q1155772","display_name":"Scheme (mathematics)","level":2,"score":0.2797999978065491},{"id":"https://openalex.org/C90329073","wikidata":"https://www.wikidata.org/wiki/Q914232","display_name":"Ask price","level":2,"score":0.2775999903678894},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.25519999861717224},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.25119999051094055}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.16511","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.16511","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"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.2606.16511","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.16511","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"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":[{"id":"https://metadata.un.org/sdg/10","score":0.7182743549346924,"display_name":"Reduced inequalities"}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Recent":[0],"work":[1],"motivates":[2],"moving":[3],"large":[4,40],"language":[5],"model":[6],"(LLM)":[7],"evaluation":[8],"from":[9,38],"mean-based":[10],"to":[11,96],"tail-aware":[12],"metrics,":[13],"including":[14],"conditional":[15],"value-at-risk":[16],"and":[17,51,67,124,153],"tail-index":[18,29,162],"estimates":[19],"of":[20,80,90,114],"reward-model":[21],"error.":[22],"We":[23,59,133],"ask":[24],"whether":[25],"the":[26,41,49,78,83,108,126,139,149,155],"canonical":[27],"extreme-value-theory":[28],"parameter,":[30],"which":[31],"isolates":[32],"how":[33,39],"heavy":[34],"a":[35,52,61,88,97,118,158],"tail":[36,42],"is":[37,77,87,145],"mass":[43],"is,":[44],"adds":[45],"discriminative":[46],"information":[47],"beyond":[48],"mean":[50],"standard":[53,98],"tail-magnitude":[54],"statistic":[55],"in":[56,138,164],"LLM":[57,99,140],"evaluation.":[58],"pre-register":[60],"protocol":[62,76,109,156],"covering":[63],"admissibility,":[64],"goodness-of-fit,":[65],"threshold-stability,":[66],"effect-size":[68],"requirements":[69],"for":[70,161],"any":[71],"positive":[72],"tail-shape":[73,128,136],"claim.":[74],"The":[75],"contribution":[79],"this":[81],"paper;":[82],"empirical":[84],"study":[85],"below":[86],"demonstration":[89],"what":[91],"its":[92],"gates":[93],"catch.":[94],"Applied":[95],"toxicity-evaluation":[100,141],"setup":[101],"under":[102],"two":[103],"structurally":[104],"different":[105],"scorer":[106],"families,":[107],"catches":[110],"three":[111],"distinct":[112],"modes":[113],"false":[115],"positives":[116],"that":[117,135],"naive":[119],"analysis":[120],"would":[121],"have":[122],"published,":[123],"rejects":[125],"headline":[127],"claim":[129],"on":[130],"both":[131],"scorers.":[132],"conclude":[134],"estimation":[137],"setups":[142],"we":[143],"examined":[144],"more":[146],"fragile":[147],"than":[148],"recent":[150],"literature":[151],"suggests,":[152],"recommend":[154],"as":[157],"starting":[159],"point":[160],"claims":[163],"similar":[165],"setups.":[166]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-06-17T00:00:00"}