{"id":"W4417068938","doi":"10.1109/iccv51701.2025.02126","title":"Aurelia: Test-Time Reasoning Distillation in Audio-Visual LLMs","year":2025,"lang":"en","type":"preprint","venue":"","topic":"Multimodal Machine Learning Applications","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Kootenay Association for Science & Technology; University of Toronto","funders":"","keywords":"Benchmark (surveying); Process (computing); Case-based reasoning; Visual reasoning; Reasoning system; Deductive reasoning; Code (set theory)","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":[0.000507618,0.0003005209,0.0003554941,0.0003497303,0.0001130607,0.0002716831,0.001338292,0.0003103691,0.0000567661],"category_scores_gemma":[0.0008026726,0.0003066634,0.0001029104,0.0005980091,0.00003243701,0.0001406928,0.002359985,0.00102863,0.0002912509],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002187003,"about_ca_system_score_gemma":0.0003247658,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002625396,"about_ca_topic_score_gemma":0.00005354517,"domain_scores_codex":[0.9977325,0.000120936,0.0004741859,0.0010272,0.0003310553,0.0003141157],"domain_scores_gemma":[0.997843,0.0006117555,0.0002310735,0.001120018,0.0001096415,0.00008445778],"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":[0.00002042852,0.0008046276,0.2229051,0.0004359568,0.0001011893,0.00002876885,0.002180139,0.3723829,0.001126317,0.1944539,0.008385998,0.1971747],"study_design_scores_gemma":[0.000135037,0.0000137106,0.1824304,0.0001618133,0.00000673145,0.00000160168,0.000002344298,0.8137304,0.00007245746,0.002655923,0.0005328603,0.000256746],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.03629496,0.0001039394,0.9108362,0.006144071,0.0004130155,0.0009472929,0.00002267518,0.001374865,0.04386294],"genre_scores_gemma":[0.8099886,0.00001191893,0.184517,0.0002365717,0.00009388127,0.0001631303,0.00006278891,0.00001435187,0.004911697],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.7736937,"threshold_uncertainty_score":0.9999385,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.006768323365754613,"score_gpt":0.2952767481250392,"score_spread":0.2885084247592846,"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."}}