{"id":"W4404783219","doi":"10.18653/v1/2024.emnlp-main.329","title":"Benchmarking Vision Language Models for Cultural Understanding","year":2024,"lang":"en","type":"article","venue":"","topic":"Language, Metaphor, and Cognition","field":"Psychology","cited_by":27,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Samsung; Canadian Institute for Advanced Research","keywords":"Benchmarking; Computer science; Artificial intelligence; Business","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0001878943,0.00009876762,0.00009456652,0.00008492284,0.00008325173,0.0001222521,0.00006474658,0.00007888309,0.001919904],"category_scores_gemma":[0.0000065115,0.00007076476,0.0001063189,0.0001241221,0.00001973044,0.0002244568,0.00001576021,0.00008357035,0.0001208879],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0000644314,"about_ca_system_score_gemma":0.000008433182,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00008035456,"about_ca_topic_score_gemma":0.00004121496,"domain_scores_codex":[0.9992679,0.00002953293,0.0001272959,0.0002625542,0.00009774462,0.0002149078],"domain_scores_gemma":[0.9996985,0.0001125107,0.00001608682,0.0001151415,0.00001817761,0.00003963424],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.00005581178,0.00003929483,0.000006322428,0.00007467881,0.0001147056,0.00005242148,0.02319795,0.00001816368,0.002795706,0.8962135,0.02944827,0.04798317],"study_design_scores_gemma":[0.003078307,0.0008693096,0.000178932,0.0005990518,0.0005047726,0.0001944789,0.3060656,0.1669536,0.003041722,0.4958066,0.02126182,0.00144581],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.04477185,0.002572859,0.5799147,0.00027743,0.001870356,0.0003204469,0.00002589878,0.0003695396,0.3698769],"genre_scores_gemma":[0.9882802,0.00001176851,0.001152058,0.0002599414,0.0004397394,0.00004037003,0.00008273818,0.00001998312,0.009713179],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9435084,"threshold_uncertainty_score":0.9989925,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0727058910967509,"score_gpt":0.3697052267384462,"score_spread":0.2969993356416953,"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."}}