{"id":"W4412945738","doi":"10.18653/v1/2025.acl-long.276","title":"Hallucination Detox: Sensitivity Dropout (SenD) for Large Language Model Training","year":2025,"lang":"en","type":"article","venue":"","topic":"Brain Tumor Detection and Classification","field":"Neuroscience","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Compute Canada; Canadian Institute for Advanced Research","keywords":"Dropout (neural networks); Computer science; Sensitivity (control systems); Training (meteorology); Language model; Artificial intelligence; Machine learning; Engineering","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":[0.005156425,0.002409986,0.001888401,0.001126288,0.001097105,0.001450953,0.00358999,0.002571712,0.01624426],"category_scores_gemma":[0.01545655,0.001271868,0.001520293,0.0008512877,0.0008068621,0.00294261,0.004240476,0.00485571,0.006562804],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009720456,"about_ca_system_score_gemma":0.002800782,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006027129,"about_ca_topic_score_gemma":0.01630851,"domain_scores_codex":[0.9982278,0.0008497396,0.0001380388,0.0002746141,0.0003158086,0.00019404],"domain_scores_gemma":[0.9953476,0.002832797,0.0001431019,0.0008421483,0.0005855992,0.0002488626],"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":[0.002379611,0.0009772406,0.002689454,0.0004925405,0.0007645262,0.0005012251,0.0003565444,0.1107668,0.006931912,0.00958976,0.1000163,0.764534],"study_design_scores_gemma":[0.0003274196,0.0002536116,0.0003147665,0.00003068022,0.00005560798,0.00006292028,0.00007854234,0.9794481,0.005206462,0.01049502,0.003697334,0.00002950542],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.03056349,0.001410192,0.8945657,0.001337063,0.0005869703,0.00032378,0.001293714,0.06701824,0.002900831],"genre_scores_gemma":[0.3449608,0.0005086769,0.6260412,0.001276372,0.0003752816,0.001139897,0.007730137,0.005694458,0.01227325],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01624426,"threshold_uncertainty_score":0.05434245,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.05352404748894912,"score_gpt":0.3223530593030851,"score_spread":0.268829011814136,"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."}}