{"id":"W3037828233","doi":"10.48550/arxiv.1904.11439","title":"META-Learning State-based Eligibility Traces for More Sample-Efficient\\n Policy Evaluation","year":2019,"lang":"","type":"preprint","venue":"arXiv (Cornell University)","topic":"Reinforcement Learning in Robotics","field":"Computer Science","cited_by":7,"is_retracted":false,"has_abstract":true,"ca_institutions":"Mila - Quebec Artificial Intelligence Institute; McGill University","funders":"","keywords":"Reinforcement learning; Computer science; Robustness (evolution); Machine learning; Temporal difference learning; Artificial intelligence; Q-learning; Sample (material); TRACE (psycholinguistics)","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":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.005610837,0.001270061,0.001623821,0.001273039,0.0009623637,0.0007600882,0.003784883,0.0006821725,0.0003501939],"category_scores_gemma":[0.002634945,0.001448617,0.001763171,0.002314891,0.0005593536,0.000732894,0.001883825,0.00169623,0.0002129556],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002097534,"about_ca_system_score_gemma":0.003921904,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0007045917,"about_ca_topic_score_gemma":0.00001613287,"domain_scores_codex":[0.9907823,0.001648456,0.001192914,0.003829886,0.0009953522,0.001551155],"domain_scores_gemma":[0.9883607,0.003204662,0.002218086,0.003401602,0.002319476,0.0004954783],"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":[0.0002489312,0.000220572,0.003253281,0.0005335612,0.001364194,0.00001009851,0.001172496,0.9748872,0.00002376835,0.01714876,0.00003053433,0.001106597],"study_design_scores_gemma":[0.002581093,0.000488231,0.001432939,0.0001118628,0.004003098,0.000001241455,0.0002481149,0.9852003,0.0002386545,0.003096771,0.00129683,0.001300839],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1070036,0.0001114311,0.8861336,0.0005257711,0.001064158,0.004275801,0.00009422439,0.0003254403,0.0004660066],"genre_scores_gemma":[0.9876084,0.00008989858,0.008359835,0.000209302,0.0001411431,0.00002399512,0.000245711,0.00008940477,0.003232346],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.8806047,"threshold_uncertainty_score":0.9987963,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.2079463326522307,"score_gpt":0.2850733927490826,"score_spread":0.07712706009685186,"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."}}