{"id":"W4416748856","doi":"10.1109/iros60139.2025.11246761","title":"Refined Policy Distillation: From VLA Generalists to RL Experts","year":2025,"lang":"","type":"article","venue":"","topic":"Multimodal Machine Learning Applications","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"Ministry of Education","keywords":"Reinforcement learning; Complement (music); Policy learning; Generalization; Perspective (graphical); Key (lock); Task (project management)","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","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0002139813,0.0004335218,0.0004120303,0.0005202834,0.0005678099,0.0007302684,0.001835425,0.000195992,0.0005976524],"category_scores_gemma":[0.0005326049,0.0004382784,0.0001556753,0.002797864,0.00008357569,0.0002754559,0.001341329,0.00025335,0.0009059611],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003630857,"about_ca_system_score_gemma":0.000575573,"about_ca_topic_candidate":true,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.04120293,"about_ca_topic_score_gemma":0.0006523341,"domain_scores_codex":[0.9965342,0.0002275068,0.0007360344,0.001447141,0.000449168,0.0006059797],"domain_scores_gemma":[0.9964463,0.0003367304,0.0001587021,0.002374965,0.0002643171,0.0004189623],"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":[0.00003059151,0.0002280367,0.01180502,0.00002178917,0.00009194661,0.000005842081,0.002630864,0.01212315,0.004768512,0.595974,0.03299076,0.3393295],"study_design_scores_gemma":[0.0005421186,0.00004814442,0.308431,0.00008746579,0.00002099578,0.000001809468,0.0000229356,0.5453138,0.000845143,0.006003175,0.1381327,0.0005507245],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.05754455,0.0002094196,0.739852,0.1502776,0.0007767258,0.000589313,0.00003497936,0.000422589,0.05029292],"genre_scores_gemma":[0.8391844,0.00001360885,0.1281215,0.0122079,0.0008336607,0.0001335004,0.00003451519,0.00002215648,0.01944875],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.7816398,"threshold_uncertainty_score":0.999872,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01237008961308218,"score_gpt":0.3352070523733426,"score_spread":0.3228369627602605,"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."}}