{"id":"W7082282613","doi":"10.48448/zr7a-0a35","title":"SafeRoute: Adaptive Model Selection for Efficient and Accurate Safety Guardrails in Large Language Models","year":2025,"lang":"en","type":"other","venue":"Underline Science Inc.","topic":"Adversarial Robustness in Machine Learning","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Kootenay Association for Science & Technology","funders":"","keywords":"Guard (computer science); Router; Benchmark (surveying); Block (permutation group theory); Heuristics; Model selection; Selection (genetic algorithm)","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":[0.001566184,0.0003792858,0.0004461528,0.001141248,0.0003546222,0.0002270869,0.001389158,0.0002519173,0.0000166647],"category_scores_gemma":[0.0002823654,0.0003629543,0.00005970763,0.001517221,0.0002583318,0.0004973015,0.0008806889,0.0005054583,0.000005868492],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003616948,"about_ca_system_score_gemma":0.001161338,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0003464179,"about_ca_topic_score_gemma":0.0006821661,"domain_scores_codex":[0.9969453,0.00008920703,0.0003986925,0.001244585,0.0005682161,0.0007539933],"domain_scores_gemma":[0.9987041,0.0001837174,0.0002840969,0.0005445484,0.0001640601,0.0001194314],"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":[0.00002316836,0.0000538487,0.00001587154,0.00003628524,0.0000115637,0.000002558023,0.0007838092,0.7725331,0.00004713438,0.2164881,0.0007608932,0.00924368],"study_design_scores_gemma":[0.0008140112,0.00004917041,0.00001736641,0.0001754889,0.00001388032,0.000002967535,0.000176141,0.9927184,0.00002546004,0.004968031,0.0006725797,0.0003665151],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.00005647599,0.0002238213,0.955087,0.0002923347,0.000379802,0.0008055327,0.00006836985,0.0002957924,0.04279086],"genre_scores_gemma":[0.2328539,0.00006848104,0.6924516,0.0005321956,0.0002365764,0.0000966576,0.00003822111,0.0001874896,0.07353491],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.2626354,"threshold_uncertainty_score":0.9998822,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01736622173029912,"score_gpt":0.2995262720487557,"score_spread":0.2821600503184566,"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."}}