{"id":"W7082262134","doi":"10.48448/2tta-p709","title":"Fine-Tuning on Diverse Reasoning Chains Drives Within-Inference CoT Refinement in LLMs","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":"Queen's University","funders":"","keywords":"Inference; Boosting (machine learning); Set (abstract data type); Automated reasoning; Baseline (sea); Chain (unit)","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"],"consensus_categories":[],"category_scores_codex":[0.001014021,0.0004026656,0.0003920956,0.000831686,0.0003117959,0.0002436926,0.002700832,0.0002150184,0.0002735191],"category_scores_gemma":[0.0009852847,0.0003637937,0.00005436137,0.001687427,0.0005261499,0.0002037826,0.001280568,0.0005869024,0.00007082379],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000179442,"about_ca_system_score_gemma":0.0009320258,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0004828837,"about_ca_topic_score_gemma":0.0009397633,"domain_scores_codex":[0.9968005,0.00006330321,0.0004009447,0.001286923,0.0007527438,0.0006956197],"domain_scores_gemma":[0.9980589,0.0001518094,0.0003611283,0.001124023,0.0001472227,0.0001569812],"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":[0.00001654575,0.0004304211,0.003782746,0.0002781188,0.0000528335,0.0003199062,0.002440961,0.01309692,0.002149333,0.8863454,0.04046186,0.05062494],"study_design_scores_gemma":[0.002015906,0.0005228036,0.0016256,0.007910755,0.0000381477,0.0000247891,0.00123223,0.7042772,0.004290182,0.01101535,0.2642842,0.002762866],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"other","genre_gemma":"other","genre_scores_codex":[0.00031473,0.000112677,0.0751988,0.002728249,0.001140723,0.0003451466,0.00001289594,0.0004492804,0.9196975],"genre_scores_gemma":[0.2168576,0.0000663225,0.0673603,0.0009938196,0.0002546714,0.00004312994,0.00003276378,0.00003810206,0.7143533],"genre_candidate":"other","genre_consensus":"other","teacher_disagreement_score":0.87533,"threshold_uncertainty_score":0.9998814,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02212347389358385,"score_gpt":0.2788930897980723,"score_spread":0.2567696159044884,"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."}}