{"id":"W6947826500","doi":"10.48448/5786-9s75","title":"STOP! Benchmarking Large Language Models with Sensitivity Testing on Offensive Progressions","year":2024,"lang":"en","type":"other","venue":"Underline Science Inc.","topic":"Species Distribution and Climate Change","field":"Environmental Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Brock University","funders":"","keywords":"Offensive; Benchmarking; Context (archaeology); Language model; Sensitivity (control systems); Natural language; Focus (optics); Natural language understanding","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":[0.009265216,0.003349174,0.001062744,0.001998152,0.0009919226,0.002600481,0.003100148,0.002597379,0.007566205],"category_scores_gemma":[0.04687266,0.0007360574,0.00209622,0.001432162,0.001233903,0.004875332,0.003043604,0.004138989,0.005071722],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001427827,"about_ca_system_score_gemma":0.002117118,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01711483,"about_ca_topic_score_gemma":0.02402117,"domain_scores_codex":[0.9923536,0.004077052,0.0005322369,0.001695519,0.0009629145,0.0003787249],"domain_scores_gemma":[0.9749123,0.01859151,0.0005456344,0.003400522,0.001982338,0.0005676683],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.002368009,0.001620757,0.04103163,0.00192406,0.001700354,0.0009257396,0.001220137,0.4795515,0.008367402,0.008326399,0.18359,0.2693741],"study_design_scores_gemma":[0.0002714697,0.0003124987,0.003415517,0.000139966,0.0001199688,0.0001985594,0.0003526105,0.9637898,0.00502406,0.01214984,0.01412771,0.0000978883],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"methods","genre_scores_codex":[0.5773118,0.00654579,0.2139199,0.005010961,0.00279144,0.001171853,0.05913835,0.1082373,0.02587264],"genre_scores_gemma":[0.7695427,0.000637094,0.1050437,0.002111974,0.0003404325,0.0007737649,0.1129961,0.004251995,0.00430227],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.01711483,"threshold_uncertainty_score":0.04899973,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03526844744257849,"score_gpt":0.2887160014226878,"score_spread":0.2534475539801093,"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."}}