{"id":"W7130611624","doi":"10.1109/fllm67465.2025.11391199","title":"Enhancing the Reliability of Large Language Models in Specialized Domains by Balancing Internal and External Knowledge","year":2025,"lang":"","type":"article","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University","funders":"","keywords":"Reliability (semiconductor); Focus (optics); Measure (data warehouse); Language model; Training (meteorology); Training set","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":[0.00585818,0.002885231,0.002472498,0.002415742,0.0009802851,0.002558939,0.0036268,0.00282812,0.002088706],"category_scores_gemma":[0.02679968,0.001091852,0.001906303,0.001461072,0.001353324,0.004887776,0.003361,0.003630689,0.002580218],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001168784,"about_ca_system_score_gemma":0.001796903,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00433445,"about_ca_topic_score_gemma":0.006679205,"domain_scores_codex":[0.9971153,0.001245025,0.0001992,0.000817874,0.0004410387,0.0001815823],"domain_scores_gemma":[0.983614,0.01177702,0.0007200992,0.002452566,0.001077555,0.0003588556],"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":[0.0009363004,0.0004344744,0.005302254,0.0009000723,0.0005298572,0.0006212128,0.0008900813,0.2950593,0.02104017,0.005393347,0.01315774,0.6557353],"study_design_scores_gemma":[0.0000543272,0.00008969032,0.0005009318,0.00004016195,0.00008214772,0.0001705044,0.0001077702,0.9842378,0.005178542,0.008015716,0.001490369,0.00003194742],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.05430211,0.003615438,0.9284152,0.0008154659,0.0001491361,0.0002265381,0.0004187916,0.00954047,0.002516877],"genre_scores_gemma":[0.6664761,0.001396355,0.3222165,0.001015565,0.0004278644,0.0003278799,0.002973249,0.00121347,0.003953076],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.00585818,"threshold_uncertainty_score":0.03098142,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.00894258193979811,"score_gpt":0.2789350916978804,"score_spread":0.2699925097580823,"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."}}