{"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":"codex-gemma-dda1882f352a","candidate_categories":["insufficient_payload"],"consensus_categories":["insufficient_payload"],"category_scores_codex":[0.0005360366,0.0003163439,0.0002335697,0.0001769984,0.0002778868,0.0001568556,0.0002900405,0.0001300699,0.01388869],"category_scores_gemma":[0.00009013912,0.0002300898,0.00004122363,0.00102678,0.0009078268,0.0001279631,0.000459252,0.0003619169,0.002463033],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006172978,"about_ca_system_score_gemma":0.00006560185,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0005810821,"about_ca_topic_score_gemma":0.003094033,"domain_scores_codex":[0.9973544,0.0000417126,0.0001551,0.0008574069,0.0009495044,0.000641903],"domain_scores_gemma":[0.9991065,0.00006176304,0.0001518167,0.0004511074,0.00002915611,0.0001996858],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00003757191,0.0008783571,0.002099745,0.0002029692,0.00007060204,0.001385897,0.002301891,0.001135625,0.006822639,0.01937462,0.9319108,0.03377931],"study_design_scores_gemma":[0.002643941,0.001966519,0.007758811,0.006094679,0.0003654021,0.0005764225,0.0281701,0.2527221,0.002181138,0.001565045,0.6901135,0.005842295],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.005630187,0.0001186541,0.0005584416,0.0002902516,0.0003513183,0.0004458704,0.000417331,0.0004317569,0.9917562],"genre_scores_gemma":[0.7458476,0.0000389195,0.006747006,0.001822721,0.0005559676,0.00007112983,0.000313132,0.0005397979,0.2440637],"genre_candidate":"other","genre_consensus":null,"teacher_disagreement_score":0.7476925,"threshold_uncertainty_score":0.9983137,"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."}}