{"id":"W2591984255","doi":"10.17863/cam.21343","title":"Sequence Tutor: Conservative Fine-Tuning of Sequence Generation Models with KL-control","year":2017,"lang":"en","type":"article","venue":"Apollo (University of Cambridge)","topic":"Music and Audio Processing","field":"Computer Science","cited_by":76,"is_retracted":false,"has_abstract":true,"ca_institutions":"Université de Montréal; Google (Canada)","funders":"Engineering and Physical Sciences Research Council; Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Recurrent neural network; Sequence (biology); Reinforcement learning; Security token; Artificial intelligence; Machine learning; Artificial neural network","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002813745,0.001104185,0.0009655235,0.0005245131,0.0004309006,0.00091765,0.002022626,0.001137228,0.003388411],"category_scores_gemma":[0.009267523,0.0005511149,0.0005067873,0.0002765758,0.001269797,0.001453674,0.001970183,0.001970778,0.000722109],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008912724,"about_ca_system_score_gemma":0.001236474,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002760936,"about_ca_topic_score_gemma":0.003236106,"domain_scores_codex":[0.9989681,0.0003393711,0.00005196088,0.0002269354,0.0003183266,0.00009527774],"domain_scores_gemma":[0.9969459,0.001947869,0.0002724176,0.0003518652,0.0003441104,0.0001379097],"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":[0.0001115335,0.00009353199,0.0005315164,0.00006668696,0.00003161135,0.00006094513,0.0000865272,0.9096749,0.004675353,0.01015799,0.0009673316,0.07354204],"study_design_scores_gemma":[0.000007864462,0.00001981654,0.00002226867,0.000002851407,0.000001879105,0.000005986083,0.000002153412,0.9975029,0.0006366239,0.001604012,0.0001908908,0.000002815917],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.01092962,0.0001133709,0.9860982,0.00008431326,0.0000390987,0.00004471779,0.00001719666,0.001056753,0.001616814],"genre_scores_gemma":[0.7473419,0.0001040898,0.2474133,0.0003119206,0.00008203053,0.0002400753,0.0001350266,0.0005788886,0.003792728],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.003388411,"threshold_uncertainty_score":0.01488066,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.06863170625603472,"score_gpt":0.239950264465594,"score_spread":0.1713185582095593,"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."}}