{"id":"W4406134357","doi":"10.1109/iccv51701.2025.00156","title":"AVTrustBench: Assessing and Enhancing Reliability and Robustness in Audio-Visual LLMs","year":2025,"lang":"en","type":"preprint","venue":"","topic":"Digital Rights Management and Security","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Université de Montréal; Kootenay Association for Science & Technology; University of Toronto","funders":"","keywords":"Robustness (evolution); Audio visual; Reliability (semiconductor); Computer science; Reliability engineering; Multimedia; Engineering; Chemistry","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":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0008288934,0.0002757274,0.0003990678,0.0002552374,0.00009367855,0.001622171,0.000571178,0.000186797,0.0000100755],"category_scores_gemma":[0.00009511531,0.0002337739,0.00004786358,0.0002435221,0.00009268012,0.0008795356,0.004596436,0.0004332291,0.000001431287],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00007341554,"about_ca_system_score_gemma":0.000116426,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002134263,"about_ca_topic_score_gemma":0.0002872957,"domain_scores_codex":[0.9979246,0.00009297849,0.0003982682,0.001038389,0.0002474323,0.0002983631],"domain_scores_gemma":[0.9990055,0.0001948675,0.0001089844,0.0005468795,0.00006008011,0.00008368826],"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":[0.00003047085,0.001104754,0.04074637,0.007312449,0.0001628615,0.0002494067,0.004548142,0.01062798,0.00005216806,0.4453892,0.001627338,0.4881488],"study_design_scores_gemma":[0.000790378,0.00006493156,0.07760697,0.00134074,0.00004022256,0.000005001452,0.0002477474,0.8023593,0.000525862,0.113148,0.002461164,0.001409621],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5505019,0.0001952672,0.3972836,0.0008319058,0.0005804264,0.0003901433,0.000001711375,0.0001899318,0.05002506],"genre_scores_gemma":[0.9782766,0.00007483298,0.01940405,0.0001041079,0.00003040455,0.00001654924,0.000005349373,0.000005064159,0.002083096],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.7917314,"threshold_uncertainty_score":0.9994143,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01555325027566546,"score_gpt":0.2816976066732604,"score_spread":0.266144356397595,"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."}}