{"id":"W4385015322","doi":"10.48550/arxiv.2307.10236","title":"Look Before You Leap: An Exploratory Study of Uncertainty Measurement for Large Language Models","year":2023,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Software Engineering Research","field":"Computer Science","cited_by":17,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"JST-Mirai Program; Japan Society for the Promotion of Science; Natural Sciences and Engineering Research Council of Canada; Canadian Institute for Advanced Research","keywords":"Trustworthiness; Estimation; Computer science; Computer security; Economics","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.03146103,0.0008810477,0.00065809,0.002112694,0.001174183,0.002506672,0.001964262,0.001586894,0.000938327],"category_scores_gemma":[0.219968,0.0005745706,0.001058935,0.001806406,0.002432334,0.005738533,0.003189286,0.004074125,0.0002694992],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001873664,"about_ca_system_score_gemma":0.001319542,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004405742,"about_ca_topic_score_gemma":0.005308107,"domain_scores_codex":[0.9693649,0.02379973,0.001114213,0.002042368,0.003270688,0.0004079697],"domain_scores_gemma":[0.5086489,0.4625421,0.008553146,0.01374199,0.005538207,0.0009755762],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.002709149,0.002107351,0.2498857,0.002681288,0.001228835,0.002866813,0.04629505,0.2892392,0.01608579,0.05885544,0.01554423,0.3125012],"study_design_scores_gemma":[0.000133675,0.0008075609,0.02632644,0.0002712032,0.0001303665,0.00085912,0.004904356,0.9008011,0.009294544,0.04730482,0.008980416,0.0001862689],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8503106,0.001319664,0.1405707,0.002275387,0.00005736082,0.00043354,0.00119429,0.00118215,0.002656388],"genre_scores_gemma":[0.9274985,0.0001715715,0.06997803,0.0002904627,0.00003765351,0.0002500008,0.001186579,0.0002425167,0.0003446054],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.03146103,"threshold_uncertainty_score":0.1663838,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1721652541498987,"score_gpt":0.244593140594098,"score_spread":0.07242788644419926,"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."}}