{"id":"W4296182512","doi":"10.31234/osf.io/z8yrv","title":"Action chunking as conditional policy compression","year":2022,"lang":"en","type":"preprint","venue":"","topic":"Reinforcement Learning in Robotics","field":"Computer Science","cited_by":10,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University","funders":"Office of Naval Research; National Science Foundation","keywords":"Chunking (psychology); Action (physics); Compression (physics); Computer science; Econometrics; Natural language processing; Artificial intelligence; Economics","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":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0002328116,0.0002155275,0.0001949426,0.0003367802,0.0003417966,0.0003710244,0.001534688,0.0001466388,0.001786445],"category_scores_gemma":[0.00007780298,0.0002215199,0.0001220217,0.0002364535,0.00003290221,0.0002751541,0.005300696,0.0009855522,0.0001994872],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003851485,"about_ca_system_score_gemma":0.0004982696,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002613462,"about_ca_topic_score_gemma":0.000001066585,"domain_scores_codex":[0.9979367,0.000126763,0.0003008433,0.0005350529,0.0008327972,0.0002678249],"domain_scores_gemma":[0.9985977,0.0001063768,0.0003069512,0.0008339304,0.00007118459,0.00008379379],"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":[0.000001972365,0.00001341174,0.00009471386,0.00002816828,0.00002025622,0.000006134413,0.0001356387,0.8193392,0.0001244639,0.1764147,0.002651761,0.001169597],"study_design_scores_gemma":[0.0002375773,0.00006054838,0.001952458,0.00006208791,0.00000867974,0.00002787641,0.00004575165,0.9081709,0.0005078515,0.02949802,0.05899166,0.0004365267],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.0005790429,0.00001586068,0.9118388,0.001729992,0.001426329,0.0002177464,0.000002949933,0.0004828128,0.08370646],"genre_scores_gemma":[0.8842698,0.00007955931,0.08190945,0.002766287,0.0008721317,0.000127714,0.0006081457,0.00004136766,0.0293256],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.8836907,"threshold_uncertainty_score":0.9991261,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.05507145678160796,"score_gpt":0.3614701003934595,"score_spread":0.3063986436118515,"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."}}