{"id":"https://openalex.org/W7225405642","doi":"https://doi.org/10.48550/arxiv.2611.07948","title":"Circe: Training-Free LLM Self-Evaluation via Component Interaction Consistency","display_name":"Circe: Training-Free LLM Self-Evaluation via Component Interaction Consistency","publication_year":2026,"publication_date":"2026-11-05","ids":{"openalex":"https://openalex.org/W7225405642","doi":"https://doi.org/10.48550/arxiv.2611.07948"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2611.07948","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2611.07948","pdf_url":"https://arxiv.org/pdf/2611.07948","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,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2611.07948","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5015257902","display_name":"Shuyi Miao","orcid":"https://orcid.org/0009-0007-1382-3662","observed_orcids":["https://orcid.org/0009-0007-1382-3662"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Miao, Shuyi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5161044524","display_name":"Zheng Li","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Zheng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5157716839","display_name":"Xu Xie","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xie, Xu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5161090601","display_name":"Linrui Xu","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xu, Linrui","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128002960","display_name":"Fei Shen","orcid":"https://orcid.org/0000-0001-9749-0234","observed_orcids":["https://orcid.org/0000-0001-9749-0234"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shen, Fei","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5160992476","display_name":"Xiaoyu Du","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Du, Xiaoyu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5089404640","display_name":"Tat\u2010Seng Chua","orcid":"https://orcid.org/0000-0001-6097-7807","observed_orcids":["https://orcid.org/0000-0001-6097-7807"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chua, Tat-Seng","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/T10028","display_name":"Topic Modeling","score":0.7538999915122986,"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/T10028","display_name":"Topic Modeling","score":0.7538999915122986,"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":0.12600000202655792,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.04650000110268593,"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/large-language-models","display_name":"large language models","score":0.8870000243186951},{"id":"https://openalex.org/keywords/reference-free-evaluation","display_name":"reference-free evaluation","score":0.6209999918937683},{"id":"https://openalex.org/keywords/representational-geometry","display_name":"representational geometry","score":0.9769999980926514}],"concepts":[{"id":"https://openalex.org/C55439883","wikidata":"https://www.wikidata.org/wiki/Q360812","display_name":"Correctness","level":2,"score":0.9215999841690063},{"id":"https://openalex.org/C2776436953","wikidata":"https://www.wikidata.org/wiki/Q5163215","display_name":"Consistency (knowledge bases)","level":2,"score":0.7358999848365784},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6858999729156494},{"id":"https://openalex.org/C2776359362","wikidata":"https://www.wikidata.org/wiki/Q2145286","display_name":"Representation (politics)","level":3,"score":0.6247000098228455},{"id":"https://openalex.org/C168167062","wikidata":"https://www.wikidata.org/wiki/Q1117970","display_name":"Component (thermodynamics)","level":2,"score":0.6067000031471252},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.5968000292778015},{"id":"https://openalex.org/C34388435","wikidata":"https://www.wikidata.org/wiki/Q2267362","display_name":"Bounded function","level":2,"score":0.572700023651123},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.5561000108718872},{"id":"https://openalex.org/C107551265","wikidata":"https://www.wikidata.org/wiki/Q1458245","display_name":"Displacement (psychology)","level":2,"score":0.4648999869823456},{"id":"https://openalex.org/C141353440","wikidata":"https://www.wikidata.org/wiki/Q182221","display_name":"Fuse (electrical)","level":2,"score":0.421099990606308},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.41679999232292175},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.39070001244544983},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3564000129699707},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.35510000586509705},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.3345000147819519},{"id":"https://openalex.org/C3018868096","wikidata":"https://www.wikidata.org/wiki/Q2693233","display_name":"Internal consistency","level":3,"score":0.32519999146461487},{"id":"https://openalex.org/C189950617","wikidata":"https://www.wikidata.org/wiki/Q937228","display_name":"Property (philosophy)","level":2,"score":0.32440000772476196},{"id":"https://openalex.org/C122377713","wikidata":"https://www.wikidata.org/wiki/Q4422799","display_name":"Weak consistency","level":4,"score":0.3237999975681305},{"id":"https://openalex.org/C45493050","wikidata":"https://www.wikidata.org/wiki/Q7884934","display_name":"Unified Model","level":2,"score":0.29159998893737793},{"id":"https://openalex.org/C4679612","wikidata":"https://www.wikidata.org/wiki/Q866298","display_name":"Aggregate (composite)","level":2,"score":0.2700999975204468},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.2655999958515167}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2611.07948","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2611.07948","pdf_url":"https://arxiv.org/pdf/2611.07948","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,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2611.07948","doi":"https://doi.org/10.48550/arxiv.2611.07948","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2611.07948","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,"listed_in":[],"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"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2611.07948","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2611.07948","pdf_url":"https://arxiv.org/pdf/2611.07948","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,"listed_in":[],"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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4403579965","https://openalex.org/W7117255712","https://openalex.org/W7161119554","https://openalex.org/W4412888217","https://openalex.org/W4412886705","https://openalex.org/W7225190570","https://openalex.org/W4414959296","https://openalex.org/W4407806944","https://openalex.org/W4402670736","https://openalex.org/W7120287864"],"abstract_inverted_index":{"As":[0],"large":[1],"language":[2],"models":[3],"(LLMs)":[4],"are":[5,30],"increasingly":[6],"deployed":[7],"in":[8,33,142,150],"real-world":[9],"applications,":[10],"evaluating":[11],"the":[12,61,85,136,171,176],"correctness":[13,67,200],"of":[14,88,180],"their":[15],"responses":[16,50,220],"is":[17,25],"crucial":[18],"for":[19,65],"reliable":[20],"use.":[21],"However,":[22],"such":[23],"evaluation":[24],"challenging":[26],"because":[27],"reference":[28,119],"answers":[29,120],"rarely":[31],"available":[32],"practice.":[34],"Existing":[35],"reference-free":[36],"methods":[37],"mainly":[38,76],"rely":[39],"on":[40,203],"output":[41,148],"signals,":[42,117,126],"which":[43,82,134],"may":[44],"assign":[45],"high":[46],"scores":[47],"to":[48,159,228],"incorrect":[49,219],"and":[51,90,94,115,145,163,195,206,226,231],"thus":[52],"do":[53],"not":[54],"necessarily":[55],"reflect":[56],"correctness.":[57],"This":[58],"motivates":[59],"exploring":[60],"model's":[62],"internal":[63,72,114,125],"computation":[64],"complementary":[66],"signals.":[68],"Recent":[69],"studies":[70],"exploit":[71],"states,":[73],"but":[74],"they":[75],"analyze":[77],"hidden-state":[78,168],"trajectories":[79],"across":[80,209],"layers,":[81],"record":[83],"only":[84],"combined":[86],"updates":[87,141],"attention":[89,140],"feed-forward":[91],"networks":[92],"(FFNs)":[93],"overlook":[95],"how":[96],"these":[97],"components":[98],"interact.":[99],"We":[100,153],"therefore":[101],"propose":[102],"Circe,":[103],"a":[104,110,184,198],"training-free":[105],"self-evaluation":[106],"method":[107,215],"that":[108,213],"assesses":[109],"response":[111],"using":[112],"both":[113],"external":[116,172],"without":[118],"or":[121],"repeated":[122],"sampling.":[123],"For":[124,170],"we":[127,174,191],"first":[128],"introduce":[129],"component":[130],"interaction":[131],"consistency":[132],"(CIC),":[133],"measures":[135],"directional":[137],"alignment":[138],"between":[139],"earlier":[143],"layers":[144],"attention-induced":[146],"FFN":[147],"changes":[149,165],"later":[151],"layers.":[152],"further":[154],"incorporate":[155],"representation":[156],"geometry":[157],"(RG)":[158],"characterize":[160],"normalized":[161],"displacement":[162],"angular":[164],"along":[166],"layerwise":[167],"trajectories.":[169],"signal,":[173],"convert":[175],"mean":[177],"negative":[178],"log-likelihood":[179],"generated":[181],"tokens":[182],"into":[183,197],"bounded":[185],"likelihood":[186],"confidence":[187],"(LC)":[188],"score.":[189,201],"Finally,":[190],"fuse":[192],"CIC,":[193],"RG,":[194],"LC":[196],"unified":[199],"Experiments":[202],"five":[204],"LLMs":[205],"six":[207],"benchmarks":[208],"four":[210],"domains":[211],"show":[212],"our":[214],"distinguishes":[216],"correct":[217],"from":[218],"more":[221],"effectively":[222],"than":[223],"existing":[224],"baselines,":[225],"generalizes":[227],"multilingual,":[229],"multimodal,":[230],"open-world":[232],"settings.":[233]},"counts_by_year":[],"updated_date":"2026-11-08T12:12:28.789561","created_date":"2026-11-07T00:00:00"}