{"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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004018348,0.001284319,0.0009972084,0.001753894,0.0008266675,0.003487578,0.002345769,0.002601808,0.007492875],"category_scores_gemma":[0.01481634,0.0006826122,0.001711974,0.001265649,0.000691974,0.004982705,0.002193539,0.002540439,0.003157031],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002759792,"about_ca_system_score_gemma":0.001674628,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.02989568,"about_ca_topic_score_gemma":0.02737049,"domain_scores_codex":[0.9981955,0.0008474996,0.00009458314,0.0004830855,0.0001964241,0.0001829069],"domain_scores_gemma":[0.9949125,0.002764207,0.0001098951,0.0008754422,0.001045359,0.0002924963],"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":[0.001208616,0.0006968078,0.006855601,0.000351785,0.0007164925,0.0001891468,0.0007237786,0.6216587,0.00397413,0.01909953,0.02945516,0.3150703],"study_design_scores_gemma":[0.00004532144,0.00004521382,0.0005783603,0.0000233668,0.00003333934,0.00002091941,0.0001947717,0.98875,0.001004162,0.007980253,0.001310649,0.0000135813],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.6112398,0.009196873,0.3012456,0.004668471,0.001362776,0.0003937727,0.005375525,0.01712626,0.04939094],"genre_scores_gemma":[0.9346288,0.0004785264,0.05405542,0.0002401377,0.00007795513,0.0001410756,0.00604268,0.0007721622,0.003563107],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.02989568,"threshold_uncertainty_score":0.05944335,"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."}}