{"id":"https://openalex.org/W4412886651","doi":"https://doi.org/10.18653/v1/2025.acl-long.192","title":"Establishing Trustworthy LLM Evaluation via Shortcut Neuron Analysis","display_name":"Establishing Trustworthy LLM Evaluation via Shortcut Neuron Analysis","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4412886651","doi":"https://doi.org/10.18653/v1/2025.acl-long.192"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.acl-long.192","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.192","pdf_url":"https://aclanthology.org/2025.acl-long.192.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.acl-long.192.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":null,"display_name":"Kejian Zhu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kejian Zhu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5070578353","display_name":"Shangqing Tu","orcid":"https://orcid.org/0009-0008-0640-3413"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shangqing Tu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Zhuoran Jin","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhuoran Jin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5060498828","display_name":"Lei Hou","orcid":"https://orcid.org/0000-0002-8907-3526"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lei Hou","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5003324011","display_name":"Juanzi Li","orcid":"https://orcid.org/0000-0002-6244-0664"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Juanzi Li","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5036868704","display_name":"Jun Zhao","orcid":"https://orcid.org/0000-0003-0725-6204"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jun Zhao","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":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.14083764,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"3809","last_page":"3822"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11424","display_name":"Security and Verification in Computing","score":0.6510999798774719,"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/T11424","display_name":"Security and Verification in Computing","score":0.6510999798774719,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.6291000247001648,"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/T12495","display_name":"Electrostatic Discharge in Electronics","score":0.5817000269889832,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/trustworthiness","display_name":"Trustworthiness","score":0.8614798784255981},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.7144896984100342},{"id":"https://openalex.org/keywords/computer-security","display_name":"Computer security","score":0.36019736528396606}],"concepts":[{"id":"https://openalex.org/C153701036","wikidata":"https://www.wikidata.org/wiki/Q659974","display_name":"Trustworthiness","level":2,"score":0.8614798784255981},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7144896984100342},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.36019736528396606}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.acl-long.192","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.192","pdf_url":"https://aclanthology.org/2025.acl-long.192.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.acl-long.192","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.192","pdf_url":"https://aclanthology.org/2025.acl-long.192.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G5208670511","display_name":null,"funder_award_id":"U24A20335","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"}],"funders":[{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4412886651.pdf","grobid_xml":"https://content.openalex.org/works/W4412886651.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W2076536433","https://openalex.org/W2390279801","https://openalex.org/W4391913857","https://openalex.org/W90316445","https://openalex.org/W2358668433","https://openalex.org/W4396701345","https://openalex.org/W2376932109"],"abstract_inverted_index":{"The":[0],"development":[1],"of":[2,58,69,114,156,169],"large":[3],"language":[4],"models":[5,60,71,158],"(LLMs)":[6],"depends":[7],"on":[8,15,32,96],"trustworthy":[9,137],"evaluation.However,":[10],"most":[11],"current":[12],"evaluations":[13],"rely":[14],"public":[16],"benchmarks,":[17],"which":[18],"are":[19],"prone":[20],"to":[21,36,51,75,107,165],"data":[22],"contamination":[23,53],"issues":[24],"that":[25,66,149],"significantly":[26],"compromise":[27],"fairness.Previous":[28],"researches":[29],"have":[30],"focused":[31],"constructing":[33],"dynamic":[34],"benchmarks":[35,42,174],"address":[37],"contamination.However,":[38],"continuously":[39],"building":[40],"new":[41],"is":[43,72,160],"costly":[44],"and":[45,93,159,175],"cyclical.In":[46],"this":[47],"work,":[48],"we":[49,64,98],"aim":[50],"tackle":[52],"by":[54],"analyzing":[55],"the":[56,67,112,157,167],"mechanisms":[57],"contaminated":[59,70],"themselves.Through":[61],"our":[62,115,120,150,170],"experiments,":[63],"discover":[65],"overestimation":[68],"likely":[73],"due":[74],"parameters":[76],"acquiring":[77],"shortcut":[78,89,104,109],"solutions":[79],"in":[80,117],"training.We":[81],"further":[82,163],"propose":[83],"a":[84,124,134,140],"novel":[85],"method":[86,102,151,171],"for":[87],"identifying":[88],"neurons":[90],"through":[91],"comparative":[92],"causal":[94],"analysis.Building":[95],"this,":[97],"introduce":[99],"an":[100],"evaluation":[101,121],"called":[103],"neuron":[105],"patching":[106],"suppress":[108],"neurons.Experiments":[110],"validate":[111],"effectiveness":[113],"approach":[116],"mitigating":[118],"contamination.Additionally,":[119],"results":[122],"exhibit":[123],"strong":[125],"linear":[126],"correlation":[127,147],"with":[128],"MixEval":[129],"(Ni":[130],"et":[131],"al.,":[132],"2024),":[133],"recently":[135],"released":[136],"benchmark,":[138],"achieving":[139],"Spearman":[141],"coefficient":[142],"()":[143],"exceeding":[144],"0.95.This":[145],"high":[146],"indicates":[148],"closely":[152],"reveals":[153],"true":[154],"capabilities":[155],"trustworthy.We":[161],"conduct":[162],"experiments":[164],"demonstrate":[166],"generalizability":[168],"across":[172],"various":[173],"hyperparameter":[176],"settings.":[177]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2025-10-10T00:00:00"}