{"id":"W6948203318","doi":"10.48448/pnm1-gx20","title":"Self-Consistency of Large Language Models under Ambiguity","year":2023,"lang":"en","type":"other","venue":"Open MIND","topic":"Medicinal Plant Pharmacodynamics Research","field":"Pharmacology, Toxicology and Pharmaceutics","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Dalhousie University","funders":"","keywords":"Ambiguity; Language model; Nonparametric statistics; Robustness (evolution); Consistency (knowledge bases); Suite; Probability distribution; Series (stratigraphy)","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.02250328,0.001031619,0.001222245,0.001102041,0.000834504,0.002950239,0.002285627,0.001650367,0.001438359],"category_scores_gemma":[0.1224456,0.000949888,0.00135581,0.0006701711,0.002665639,0.006396674,0.003171363,0.003430297,0.0005452504],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001566242,"about_ca_system_score_gemma":0.001842203,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002909997,"about_ca_topic_score_gemma":0.003415599,"domain_scores_codex":[0.9882082,0.007150242,0.0007960389,0.002165274,0.001301693,0.0003785013],"domain_scores_gemma":[0.8510522,0.1181597,0.007558038,0.01857073,0.003138047,0.001521254],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.001784618,0.0005433717,0.03855414,0.0004296104,0.0007262589,0.0005491592,0.00297885,0.834662,0.01241671,0.02272114,0.002239602,0.08239451],"study_design_scores_gemma":[0.00007715647,0.0002329995,0.002058379,0.0000330231,0.00005300965,0.0001178633,0.0001814166,0.9478769,0.004126676,0.04464907,0.0005496083,0.00004384465],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.6851617,0.00035864,0.305658,0.001229591,0.000055841,0.0001811909,0.0004932203,0.003614263,0.003247498],"genre_scores_gemma":[0.9555119,0.00006519126,0.04257225,0.0002637349,0.00002324543,0.0001209427,0.0006175145,0.0003479601,0.0004772563],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.02250328,"threshold_uncertainty_score":0.1190102,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.2207518045997466,"score_gpt":0.5200152132505748,"score_spread":0.2992634086508282,"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."}}