{"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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003331163,0.00198407,0.0008614673,0.0009472332,0.0006786631,0.001958264,0.003712648,0.001576835,0.006852265],"category_scores_gemma":[0.01755636,0.0009852556,0.001879,0.000535241,0.001371149,0.003526512,0.002906869,0.004028282,0.00260357],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001891871,"about_ca_system_score_gemma":0.00299415,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01078346,"about_ca_topic_score_gemma":0.02171155,"domain_scores_codex":[0.9978014,0.0006955251,0.0001332715,0.0006838746,0.0004814661,0.0002044764],"domain_scores_gemma":[0.992477,0.004440253,0.0003332532,0.001686859,0.0007443881,0.0003182344],"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":[0.001345383,0.0007589713,0.0135648,0.00112764,0.0003627826,0.0006591498,0.001127177,0.5307328,0.0388065,0.02238481,0.02514054,0.3639896],"study_design_scores_gemma":[0.0000859059,0.00007499108,0.000421823,0.0000359219,0.00004833691,0.00008533619,0.00007805651,0.9736198,0.009031507,0.0120162,0.004481523,0.00002056094],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1617844,0.001048803,0.7533576,0.001358299,0.0003633601,0.0004557635,0.001542396,0.06975648,0.01033284],"genre_scores_gemma":[0.5153032,0.0002466873,0.4737262,0.0008309003,0.00006486419,0.0002855701,0.002766897,0.002879149,0.003896635],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01078346,"threshold_uncertainty_score":0.02292311,"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."}}