{"id":"W7082275164","doi":"10.48448/3yrn-9195","title":"Global MMLU: Understanding and Addressing Cultural and Linguistic Biases in Multilingual Evaluation","year":2025,"lang":"en","type":"other","venue":"Underline Science Inc.","topic":"Geochemistry and Geologic Mapping","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Kootenay Association for Science & Technology; McGill University","funders":"","keywords":"Cultural diversity; Annotation; Benchmark (surveying); Culturally appropriate; Cultural bias; Language proficiency; Cultural competence; Translation (biology)","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":[],"consensus_categories":[],"category_scores_codex":[0.001139455,0.0002247079,0.0002301706,0.000295789,0.0002431223,0.0004146461,0.0005324819,0.0001576774,0.00003069876],"category_scores_gemma":[0.002859159,0.0002014261,0.00001780474,0.0008852239,0.0007111247,0.0001834246,0.00052398,0.0001690221,0.000001490053],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002749604,"about_ca_system_score_gemma":0.0007075522,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0004157085,"about_ca_topic_score_gemma":0.000603796,"domain_scores_codex":[0.9980148,0.00006142612,0.0002339791,0.0007991769,0.0004885162,0.0004020982],"domain_scores_gemma":[0.9991324,0.0001451249,0.0001571878,0.0002956942,0.0001624514,0.0001071506],"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":[0.00005338157,0.0005910893,0.03433116,0.002171384,0.0001819731,0.0004599518,0.00843821,0.01087886,0.002100759,0.4228539,0.03911983,0.4788195],"study_design_scores_gemma":[0.0009397939,0.00004749106,0.001226657,0.001951688,0.00003932863,0.00005439004,0.001073303,0.9543871,0.000137075,0.03332099,0.006205556,0.0006166698],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.00477854,0.005144197,0.04880225,0.001857977,0.001632828,0.0008870683,0.00004425906,0.0004643717,0.9363885],"genre_scores_gemma":[0.9710196,0.000118099,0.01643743,0.0001468503,0.0001632859,0.000008444347,0.00001962532,0.000009956329,0.01207667],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9662411,"threshold_uncertainty_score":0.8213917,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1505566938512948,"score_gpt":0.3835747598386859,"score_spread":0.2330180659873912,"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."}}