{"id":"W6929451287","doi":"10.48448/fccs-5q95","title":"FLIRT: Feedback Loop In-context Red Teaming","year":2024,"lang":"en","type":"other","venue":"Underline Science Inc.","topic":"Lipid Membrane Structure and Behavior","field":"Biochemistry, Genetics and Molecular Biology","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto","funders":"","keywords":"Adversarial system; Robustness (evolution); Feedback loop; Generative grammar; Human-in-the-loop; Key (lock)","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":[],"consensus_categories":[],"category_scores_codex":[0.0002511873,0.0002900199,0.0002438346,0.0003752211,0.00004597929,0.00008181806,0.0005771765,0.0003391684,0.0003180587],"category_scores_gemma":[0.00005577056,0.000244518,0.00007565955,0.0004324805,0.0003643849,0.000005129257,0.0002709411,0.0002634741,0.0002386095],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0000420649,"about_ca_system_score_gemma":0.0003732033,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001394263,"about_ca_topic_score_gemma":0.001262433,"domain_scores_codex":[0.9981633,0.0000187848,0.0002398534,0.0008447531,0.0002984389,0.0004348842],"domain_scores_gemma":[0.9991856,0.000003618817,0.0000925428,0.0005599689,0.00004650009,0.0001117637],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"bench_or_experimental","study_design_gemma":"not_applicable","study_design_scores_codex":[0.000007232161,0.00003778336,0.0001506361,0.00008371469,0.00001925989,0.00002536428,0.00005046659,0.000002378122,0.8749256,0.0001051624,0.09959351,0.02499896],"study_design_scores_gemma":[0.0004398109,0.0001249632,0.00009242118,0.0002487994,0.00004745706,0.00004159717,0.000222043,0.0000335413,0.1930282,0.00005782016,0.8051104,0.0005529198],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"other","genre_gemma":"other","genre_scores_codex":[0.07906832,0.004798201,0.0001695558,0.0006958015,0.006460409,0.001089903,0.0001284672,0.0002298869,0.9073595],"genre_scores_gemma":[0.3727697,0.0001635558,0.0007085898,0.0004236428,0.001420709,0.00001725014,0.0001048343,0.0002831778,0.6241086],"genre_candidate":"other","genre_consensus":"other","teacher_disagreement_score":0.7055169,"threshold_uncertainty_score":0.9971155,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01160376732455667,"score_gpt":0.2764850778676766,"score_spread":0.2648813105431199,"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."}}