{"id":"https://openalex.org/W7160632553","doi":"https://doi.org/10.48550/arxiv.2605.05973","title":"Towards Reliable LLM Evaluation: Correcting the Winner's Curse in Adaptive Benchmarking","display_name":"Towards Reliable LLM Evaluation: Correcting the Winner's Curse in Adaptive Benchmarking","publication_year":2026,"publication_date":"2026-05-07","ids":{"openalex":"https://openalex.org/W7160632553","doi":"https://doi.org/10.48550/arxiv.2605.05973"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.05973","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.05973","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.2605.05973","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5135679813","display_name":"Yang Xu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xu, Yang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135699492","display_name":"Jiefu Zhang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Jiefu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135723795","display_name":"Haixiang Sun","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sun, Haixiang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135724521","display_name":"Zihan Zhou","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhou, Zihan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135726413","display_name":"Tianyu Cao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Cao, Tianyu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5135661058","display_name":"Vaneet Aggarwal","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Aggarwal, Vaneet","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/T10286","display_name":"Information Retrieval and Search Behavior","score":0.13899999856948853,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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/T10286","display_name":"Information Retrieval and Search Behavior","score":0.13899999856948853,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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":0.09279999881982803,"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/T12127","display_name":"Software System Performance and Reliability","score":0.09239999949932098,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"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/benchmarking","display_name":"Benchmarking","score":0.7184000015258789},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.6345999836921692},{"id":"https://openalex.org/keywords/benchmark","display_name":"Benchmark (surveying)","score":0.608299970626831},{"id":"https://openalex.org/keywords/estimator","display_name":"Estimator","score":0.6010000109672546},{"id":"https://openalex.org/keywords/protocol","display_name":"Protocol (science)","score":0.5360999703407288},{"id":"https://openalex.org/keywords/multiplier","display_name":"Multiplier (economics)","score":0.3937000036239624},{"id":"https://openalex.org/keywords/gaussian-process","display_name":"Gaussian process","score":0.3765999972820282},{"id":"https://openalex.org/keywords/gaussian","display_name":"Gaussian","score":0.3741999864578247}],"concepts":[{"id":"https://openalex.org/C86251818","wikidata":"https://www.wikidata.org/wiki/Q816754","display_name":"Benchmarking","level":2,"score":0.7184000015258789},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6345999836921692},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6194999814033508},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.608299970626831},{"id":"https://openalex.org/C185429906","wikidata":"https://www.wikidata.org/wiki/Q1130160","display_name":"Estimator","level":2,"score":0.6010000109672546},{"id":"https://openalex.org/C2780385302","wikidata":"https://www.wikidata.org/wiki/Q367158","display_name":"Protocol (science)","level":3,"score":0.5360999703407288},{"id":"https://openalex.org/C124584101","wikidata":"https://www.wikidata.org/wiki/Q1053266","display_name":"Multiplier (economics)","level":2,"score":0.3937000036239624},{"id":"https://openalex.org/C61326573","wikidata":"https://www.wikidata.org/wiki/Q1496376","display_name":"Gaussian process","level":3,"score":0.3765999972820282},{"id":"https://openalex.org/C163716315","wikidata":"https://www.wikidata.org/wiki/Q901177","display_name":"Gaussian","level":2,"score":0.3741999864578247},{"id":"https://openalex.org/C105339364","wikidata":"https://www.wikidata.org/wiki/Q2297740","display_name":"Software deployment","level":2,"score":0.36640000343322754},{"id":"https://openalex.org/C44249647","wikidata":"https://www.wikidata.org/wiki/Q208498","display_name":"Confidence interval","level":2,"score":0.36239999532699585},{"id":"https://openalex.org/C81917197","wikidata":"https://www.wikidata.org/wiki/Q628760","display_name":"Selection (genetic algorithm)","level":2,"score":0.3368000090122223},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.33180001378059387},{"id":"https://openalex.org/C162376815","wikidata":"https://www.wikidata.org/wiki/Q2158281","display_name":"Frequentist inference","level":4,"score":0.32519999146461487},{"id":"https://openalex.org/C149782125","wikidata":"https://www.wikidata.org/wiki/Q160039","display_name":"Econometrics","level":1,"score":0.32350000739097595},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.3172000050544739},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.3027999997138977},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.28529998660087585},{"id":"https://openalex.org/C46686674","wikidata":"https://www.wikidata.org/wiki/Q466303","display_name":"Boosting (machine learning)","level":2,"score":0.2825999855995178},{"id":"https://openalex.org/C174348530","wikidata":"https://www.wikidata.org/wiki/Q188635","display_name":"Bridging (networking)","level":2,"score":0.27619999647140503},{"id":"https://openalex.org/C187691185","wikidata":"https://www.wikidata.org/wiki/Q2020720","display_name":"Grid","level":2,"score":0.2703999876976013},{"id":"https://openalex.org/C95167961","wikidata":"https://www.wikidata.org/wiki/Q4483495","display_name":"Fiducial inference","level":5,"score":0.257999986410141},{"id":"https://openalex.org/C52740198","wikidata":"https://www.wikidata.org/wiki/Q1539564","display_name":"Importance sampling","level":3,"score":0.2533999979496002}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.05973","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.05973","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.2605.05973","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.05973","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":{"Adaptive":[0],"prompt":[1],"and":[2,61,86,105,108,113,124],"program":[3],"search":[4],"makes":[5],"LLM":[6],"evaluation":[7],"selection-sensitive.":[8],"Once":[9],"benchmark":[10],"items":[11],"are":[12],"reused":[13],"inside":[14],"tuning,":[15],"the":[16,23,27,52,79,87,134],"observed":[17],"winner's":[18],"score":[19],"need":[20],"not":[21],"estimate":[22],"fresh-data":[24],"performance":[25],"of":[26],"full":[28],"tune-then-deploy":[29],"procedure.":[30],"We":[31,42],"study":[32],"inference":[33,92],"for":[34,68,102],"this":[35],"procedure-level":[36],"target":[37],"under":[38],"explicit":[39],"tuning":[40,115],"budgets.":[41],"propose":[43],"SIREN,":[44],"a":[45,72,82,94],"selection-aware":[46],"repeated-split":[47],"reporting":[48,120,136],"protocol":[49],"that":[50,118],"freezes":[51],"post-search":[53],"shortlist,":[54],"separates":[55],"splitwise":[56],"selection":[57],"from":[58],"held-out":[59],"evaluation,":[60],"uses":[62],"an":[63],"item-level":[64,84],"Gaussian":[65],"multiplier":[66],"bootstrap":[67,88],"uncertainty":[69],"quantification.":[70],"In":[71],"fixed-shortlist":[73],"regime":[74],"with":[75],"smooth":[76],"stabilized":[77],"selection,":[78],"estimator":[80],"admits":[81],"first-order":[83],"representation,":[85],"yields":[89],"valid":[90],"simultaneous":[91],"on":[93],"finite":[95],"budget":[96],"grid.":[97],"This":[98],"supports":[99],"confidence":[100],"intervals":[101],"procedure-performance":[103],"curves":[104],"pre-specified":[106],"equal-budget":[107],"cross-budget":[109],"comparisons.":[110],"Controlled":[111],"simulations":[112],"MMLU-Pro":[114],"experiments":[116],"show":[117],"winner-based":[119],"can":[121,125],"be":[122],"optimistic":[123],"change":[126],"deployment":[127],"conclusions,":[128],"while":[129],"SIREN":[130],"remains":[131],"close":[132],"to":[133],"finite-sample":[135],"target.":[137]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-05-09T00:00:00"}