{"id":"W6910422042","doi":"10.48448/a52t-ff08","title":"Benchmarking Vision Language Models for Cultural Understanding","year":2024,"lang":"en","type":"other","venue":"Underline Science Inc.","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University; Mila - Quebec Artificial Intelligence Institute; Université de Montréal","funders":"","keywords":"Benchmarking; Set (abstract data type); Cultural diversity; Benchmark (surveying); Cultural background; Comprehension; Foundation (evidence)","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":[0.004867637,0.002117897,0.0006623653,0.002646994,0.0007155904,0.003363171,0.002404338,0.002090834,0.00643626],"category_scores_gemma":[0.01693651,0.0004123444,0.001831357,0.001420555,0.000735365,0.004455965,0.003050255,0.002434931,0.003626026],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002599315,"about_ca_system_score_gemma":0.002284248,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.03106826,"about_ca_topic_score_gemma":0.02916909,"domain_scores_codex":[0.9956185,0.002110865,0.0002418542,0.001082788,0.0006338794,0.0003120773],"domain_scores_gemma":[0.993852,0.0031604,0.0002242229,0.001248704,0.001236652,0.0002779856],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.0007497942,0.001035908,0.02987661,0.0009488675,0.0006180629,0.0002972152,0.001487162,0.1842804,0.009688584,0.01571791,0.05469691,0.7006026],"study_design_scores_gemma":[0.00005464742,0.0002322098,0.005914797,0.0001638454,0.00009260402,0.0001323326,0.001107843,0.9483345,0.01101977,0.01430304,0.01858049,0.00006391661],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.4346333,0.004244012,0.4391717,0.003892181,0.0009076623,0.001179321,0.01716314,0.04491771,0.05389105],"genre_scores_gemma":[0.7930524,0.0005580056,0.1680879,0.0007951175,0.00008465668,0.0005352182,0.02937486,0.001016554,0.006495315],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.03106826,"threshold_uncertainty_score":0.06177485,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.08421030401067177,"score_gpt":0.3806526856515964,"score_spread":0.2964423816409246,"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."}}