{"id":"W4417036181","doi":"10.48550/arxiv.2512.04072","title":"SkillFactory: Self-Distillation For Learning Cognitive Behaviors","year":2025,"lang":"","type":"preprint","venue":"arXiv (Cornell University)","topic":"Multimodal Machine Learning Applications","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada; Alfred P. Sloan Foundation; Institute for Information and Communications Technology Promotion; Open Philanthropy Project; National Science Foundation","keywords":"Leverage (statistics); Reinforcement learning; Cognition; Initialization; Task (project management); Cognitive model; Priming (agriculture)","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001525751,0.001635417,0.000759988,0.0004678038,0.0003869907,0.00107183,0.002949913,0.001445245,0.006945706],"category_scores_gemma":[0.00634076,0.0009850455,0.001321878,0.0003418538,0.001318434,0.002474927,0.002418761,0.003327798,0.001910838],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009323588,"about_ca_system_score_gemma":0.001415761,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004057874,"about_ca_topic_score_gemma":0.00768312,"domain_scores_codex":[0.9994634,0.0001531979,0.00003112671,0.0001791726,0.0001113376,0.00006193735],"domain_scores_gemma":[0.9983719,0.0008922056,0.0001225699,0.0003988822,0.0001290912,0.00008534087],"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":[0.0004902826,0.0003367373,0.002948282,0.0003930625,0.0001943594,0.0001717458,0.0002392622,0.6530398,0.0113033,0.02478127,0.01050051,0.2956014],"study_design_scores_gemma":[0.00002284895,0.0000406307,0.00006515115,0.00001108571,0.00000740456,0.00001135503,0.000004963948,0.9854876,0.00217199,0.01125386,0.0009150984,0.000007909583],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.03144128,0.0003260382,0.9392928,0.0004863211,0.0001370045,0.0001636786,0.0004869948,0.02467614,0.002989764],"genre_scores_gemma":[0.492708,0.0002332792,0.4965483,0.0008468965,0.00007468815,0.0005251569,0.001414211,0.001728991,0.005920385],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.006945706,"threshold_uncertainty_score":0.02323568,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.05909441059544795,"score_gpt":0.23556282801554,"score_spread":0.176468417420092,"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."}}