{"id":"W6947887187","doi":"10.48448/0fpe-gv41","title":"Confidence Under the Hood: An Investigation into the Confidence-Probability Alignment in Large Language Models","year":2024,"lang":"en","type":"other","venue":"Underline Science Inc.","topic":"Species Distribution and Climate Change","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Brock University","funders":"","keywords":"Reliability (semiconductor); Range (aeronautics); Language model; Work (physics); Language understanding; Confidence interval","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":["sts","insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.002386895,0.0003192999,0.0002222495,0.0001140207,0.0003034928,0.0003313677,0.001533117,0.0001706371,0.02359549],"category_scores_gemma":[0.00007571634,0.0001883961,0.00005860775,0.001148546,0.003214817,0.0003977961,0.0006646987,0.0004355924,0.001909882],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001214734,"about_ca_system_score_gemma":0.0001549507,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.01221,"about_ca_topic_score_gemma":0.07281323,"domain_scores_codex":[0.9967296,0.0001757726,0.0003666889,0.0009310461,0.001205659,0.0005912905],"domain_scores_gemma":[0.9984841,0.00007924056,0.0001772256,0.001053798,0.00002269175,0.0001828878],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.00001358832,0.0003268581,0.001120592,0.00009243839,0.00002799253,0.00002618901,0.02819696,0.00139994,0.004276699,0.8167242,0.1459407,0.001853747],"study_design_scores_gemma":[0.001158215,0.0003089962,0.008537668,0.0005787648,0.0001538244,0.00004272154,0.2105983,0.07003402,0.001853505,0.590655,0.1138549,0.002224177],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.1751219,0.003889179,0.00734964,0.02724812,0.002748156,0.004868637,0.0006269206,0.0009540735,0.7771934],"genre_scores_gemma":[0.9423445,0.0002298535,0.00025891,0.003807484,0.0001376395,0.0001478231,0.0001191222,0.0001243448,0.05283033],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.7672226,"threshold_uncertainty_score":0.9994979,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03851609725265768,"score_gpt":0.297084339083671,"score_spread":0.2585682418310133,"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."}}