{"id":"W3130843035","doi":"","title":"Conservative Safety Critics for Exploration","year":2021,"lang":"en","type":"article","venue":"International Conference on Learning Representations","topic":"Reinforcement Learning in Robotics","field":"Computer Science","cited_by":12,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto","funders":"","keywords":"Reinforcement learning; Computer science; Convergence (economics); Task (project management); Suite; Upper and lower bounds; Artificial intelligence; Machine learning; Mathematical optimization; Engineering; Mathematics; Law","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":[0.0002116543,0.0001302834,0.000129035,0.0001253811,0.0003106437,0.0004787887,0.0005230504,0.00005692994,0.0002231388],"category_scores_gemma":[0.002070685,0.0001476235,0.00008642254,0.0002454846,0.00007975584,0.0008343218,0.0001477986,0.0002632783,0.0001068828],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00009041642,"about_ca_system_score_gemma":0.0002562981,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0000154374,"about_ca_topic_score_gemma":0.00001086767,"domain_scores_codex":[0.9984492,0.0001425669,0.0003423723,0.0004255035,0.0004452711,0.0001950974],"domain_scores_gemma":[0.997215,0.0007006772,0.0001719606,0.0003388001,0.001506478,0.00006711167],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00001489497,0.00003472692,0.000645059,0.000005101281,0.00004670355,0.000009367834,0.001216881,0.2514834,0.0006253039,0.7428561,0.000745949,0.00231654],"study_design_scores_gemma":[0.0006166069,0.0001128152,0.001109663,0.00005639896,0.0000094078,0.00001123181,0.001614878,0.9587225,0.002720407,0.0152824,0.01953899,0.0002046517],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.0003861848,0.000005540693,0.9230995,0.02075612,0.0007723424,0.0001768135,0.00000884677,0.0001654453,0.05462917],"genre_scores_gemma":[0.9106532,0.00007504399,0.06302591,0.0008563285,0.0001613198,0.0001261145,0.0003716069,0.00001858234,0.02471186],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.9102671,"threshold_uncertainty_score":0.601991,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.117497022015405,"score_gpt":0.379141317792146,"score_spread":0.261644295776741,"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."}}