{"id":"W7082283166","doi":"10.48448/8ygs-qg81","title":"Self-Training Elicits Concise Reasoning in Large Language Models","year":2025,"lang":"en","type":"other","venue":"Underline Science Inc.","topic":"Geochemistry and Geologic Mapping","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Kootenay Association for Science & Technology","funders":"","keywords":"Inference; Language model; Sampling (signal processing); Non-monotonic logic; Data modeling; Latent variable","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002424459,0.001230678,0.0006313851,0.0005385623,0.000560394,0.001285176,0.00214192,0.001268755,0.00658462],"category_scores_gemma":[0.01714816,0.0005872236,0.001023665,0.0004547464,0.001139492,0.003380781,0.002365682,0.00268892,0.001989587],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001095767,"about_ca_system_score_gemma":0.00178831,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003572109,"about_ca_topic_score_gemma":0.01499842,"domain_scores_codex":[0.9983146,0.0007014564,0.00008298938,0.0005572159,0.0002430145,0.0001007991],"domain_scores_gemma":[0.9924543,0.005286168,0.000280368,0.001407633,0.0003726514,0.0001987971],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.000912043,0.0008359074,0.008846554,0.0009536236,0.0002959492,0.0005089412,0.0009716941,0.3003252,0.03323927,0.03174085,0.03063641,0.5907335],"study_design_scores_gemma":[0.0000601972,0.00006081874,0.0003644743,0.00002679478,0.00002366343,0.000059198,0.00006975123,0.9580549,0.007867159,0.03067798,0.002717885,0.00001715942],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.1033174,0.0005420851,0.8654656,0.001221084,0.0001639822,0.0002250677,0.001247129,0.02087728,0.006940479],"genre_scores_gemma":[0.6278163,0.0001726951,0.3617221,0.0008925037,0.00008412204,0.000225096,0.00371146,0.001048241,0.004327555],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.00658462,"threshold_uncertainty_score":0.02202773,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01684458539469767,"score_gpt":0.2691218664544908,"score_spread":0.2522772810597931,"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."}}