{"id":"W3082716386","doi":"10.48550/arxiv.2008.13773","title":"Beyond variance reduction: Understanding the true impact of baselines on policy optimization","year":2020,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Advanced Bandit Algorithms Research","field":"Decision Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University","funders":"","keywords":"Reduction (mathematics); Variance reduction; Variance (accounting); Econometrics; Economics; Computer science; Environmental science; Mathematics","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":[],"consensus_categories":[],"category_scores_codex":[0.001127262,0.0003034977,0.000461302,0.0007345846,0.0003173341,0.0001678456,0.001732296,0.000228923,0.0003519463],"category_scores_gemma":[0.00225486,0.0002219631,0.0004145095,0.003108988,0.0004244723,0.0003630004,0.0008768512,0.000740669,0.00005935245],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001017638,"about_ca_system_score_gemma":0.0008451728,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001823802,"about_ca_topic_score_gemma":0.000005172486,"domain_scores_codex":[0.9970043,0.0004900653,0.0004493984,0.001107494,0.0006107429,0.0003380158],"domain_scores_gemma":[0.9963268,0.000988684,0.0006282122,0.001327116,0.0005354091,0.0001937558],"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":[0.0002268452,0.00004153176,0.0001336427,0.000009995124,0.00007823712,0.0000314657,0.0002414161,0.9764067,0.00002313661,0.02180655,0.0006471776,0.000353321],"study_design_scores_gemma":[0.0003522634,0.0001347629,0.0002228799,0.00003859147,0.00002798709,0.000004766994,0.0009604284,0.7740712,0.0000893562,0.2238521,0.00004909221,0.0001966514],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.00649439,0.00003337699,0.9888022,0.001250425,0.0004087012,0.0004225985,0.0001308358,0.00005403894,0.002403439],"genre_scores_gemma":[0.996856,0.0002244901,0.00103521,0.00003841919,0.0004813338,9.770741e-7,0.00002188499,0.0000254317,0.001316211],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9903616,"threshold_uncertainty_score":0.9051392,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.3346032301983551,"score_gpt":0.3524518645244382,"score_spread":0.01784863432608302,"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."}}