{"id":"W4412889548","doi":"10.18653/v1/2025.acl-short.15","title":"Improving the Calibration of Confidence Scores in Text Generation Using the Output Distribution’s Characteristics","year":2025,"lang":"en","type":"article","venue":"","topic":"Advanced Text Analysis Techniques","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Institut de Valorisation des Données; Nvidia","keywords":"Calibration; Computer science; Confidence interval; Statistics; Distribution (mathematics); Mathematics","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003619334,0.00007033561,0.0001024023,0.00005540679,0.0001321425,0.0001247815,0.0005062921,0.00003175186,0.000002353124],"category_scores_gemma":[0.0001737209,0.00004088519,0.000033091,0.0005262332,0.0000709425,0.0004147112,0.0001774618,0.00008693907,2.752731e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0000682873,"about_ca_system_score_gemma":0.00008879408,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0003411067,"about_ca_topic_score_gemma":0.0001588022,"domain_scores_codex":[0.9992077,0.00008396445,0.000295756,0.0001711332,0.0001401915,0.000101242],"domain_scores_gemma":[0.9991804,0.0001132925,0.0001536304,0.0004341714,0.0001096487,0.00000883316],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.000003834978,0.00004104517,0.007456771,0.00001781532,0.0000158563,0.000001229294,0.0003893457,0.001888548,0.1306869,0.797408,0.0003555497,0.06173508],"study_design_scores_gemma":[0.00003463309,0.000006207811,0.003668769,0.00001704015,0.00000703503,6.819764e-7,0.00002865069,0.9188461,0.07458135,0.002713025,0.00005042368,0.00004610376],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.04085714,0.00004115918,0.9579692,0.0008232846,0.00006431899,0.0001558425,0.000002397365,0.00004534491,0.00004126145],"genre_scores_gemma":[0.9854705,0.00001080557,0.01416757,0.0002064657,0.00002363642,0.00001168506,0.000007848165,0.00000190123,0.00009959844],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9446133,"threshold_uncertainty_score":0.166725,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02464011149066286,"score_gpt":0.2866724729756874,"score_spread":0.2620323614850246,"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."}}