{"id":"W7077905675","doi":"10.48448/geb8-4d78","title":"FaithBench: A Diverse Hallucination Benchmark for Summarization by Modern LLMs","year":2025,"lang":"en","type":"other","venue":"Underline Science Inc.","topic":"Credit Risk and Financial Regulations","field":"Economics, Econometrics and Finance","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Waterloo","funders":"","keywords":"Automatic summarization; Benchmark (surveying); Ground truth; Grammaticality; Diversity (politics); Meaning (existential)","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.0004800697,0.0002279014,0.0003619459,0.001000689,0.000333257,0.0001576398,0.0004797323,0.0002661037,0.0005714274],"category_scores_gemma":[0.0003475824,0.0002699763,0.00008731666,0.0007884445,0.0002879829,0.0002611364,0.0001084381,0.0001244084,0.00009101391],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002361695,"about_ca_system_score_gemma":0.0002602152,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0009396233,"about_ca_topic_score_gemma":0.000594384,"domain_scores_codex":[0.9983025,0.000005391414,0.0004611998,0.0007516899,0.0001168598,0.0003623299],"domain_scores_gemma":[0.9988744,0.00004682301,0.0004778011,0.00039553,0.0001208456,0.0000846238],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"not_applicable","study_design_scores_codex":[0.000007081975,0.0001424482,0.002736827,0.0000638509,0.00002773984,4.49707e-7,0.0001630442,0.000283179,0.00003809737,0.5679645,0.3840078,0.04456502],"study_design_scores_gemma":[0.0005032842,0.00005849159,0.001336087,0.00006178854,0.00002054647,3.025737e-7,0.00003236363,0.1293381,0.00001690638,0.04770377,0.8204941,0.0004342226],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"other","genre_scores_codex":[0.00007916587,0.001266844,0.5952122,0.0003674818,0.001347418,0.0007074601,0.002777618,0.0001118334,0.3981299],"genre_scores_gemma":[0.05841111,0.0008767445,0.02526475,0.0001651195,0.000787012,0.0001602683,0.002789337,0.0002211661,0.9113245],"genre_candidate":"other","genre_consensus":null,"teacher_disagreement_score":0.5699475,"threshold_uncertainty_score":0.9999753,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02244945441742962,"score_gpt":0.2514437560826722,"score_spread":0.2289943016652425,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}