{"id":"W4403407106","doi":"10.48550/arxiv.2410.10397","title":"Tighter Risk Bounds for Mixtures of Experts","year":2024,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Psychology of Social Influence","field":"Social Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Université Laval","funders":"Natural Sciences and Engineering Research Council of Canada; Consortium de Recherche et d’innovation en Aérospatiale au Québec","keywords":"Econometrics; Computer science; Mathematics","routes":{"ca_aff":true,"ca_fund":true,"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":[],"consensus_categories":[],"category_scores_codex":[0.0005090848,0.0001977647,0.0003150309,0.0001763576,0.0003572758,0.00005312708,0.0008101581,0.0006175825,0.0001236868],"category_scores_gemma":[0.0001632545,0.00022934,0.0003448273,0.0002957472,0.001096901,0.00009629418,0.0004209165,0.000424462,0.0000444094],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001988601,"about_ca_system_score_gemma":0.0003402356,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002893666,"about_ca_topic_score_gemma":0.0009907675,"domain_scores_codex":[0.9984977,0.0001977497,0.0001817702,0.0006779597,0.00009856514,0.0003462129],"domain_scores_gemma":[0.998775,0.0002468796,0.0002550312,0.000405983,0.0001918379,0.0001252797],"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":[0.0003250244,0.0003088572,0.02013712,0.0004667287,0.0009901834,0.0001807023,0.05002126,0.00312205,0.0003799815,0.8509569,0.07108638,0.002024786],"study_design_scores_gemma":[0.0004152753,0.00006429242,0.001416372,0.0002127457,0.0003282689,3.568693e-7,0.003426044,0.0005227865,0.0002319634,0.9416363,0.05120205,0.0005435162],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9435818,0.0006716195,0.002257284,0.0005059994,0.002459531,0.0005812576,0.0001634813,0.0001843375,0.04959476],"genre_scores_gemma":[0.9910097,0.0006322093,0.0002387002,0.0001163118,0.0003483944,0.000003451829,0.000007142689,0.00002017192,0.007623959],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.09067939,"threshold_uncertainty_score":0.9352214,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1062782288353468,"score_gpt":0.2710699552333071,"score_spread":0.1647917263979602,"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."}}