{"id":"W4310413652","doi":"10.48550/arxiv.2211.14402","title":"An Analysis of Social Biases Present in BERT Variants Across Multiple Languages","year":2022,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University","funders":"","keywords":"Computer science; Natural language processing; Persian; Artificial intelligence; Set (abstract data type); Linguistics; Adjective; Noun","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":[0.0004160063,0.0002410451,0.0005117867,0.0007345363,0.0001545412,0.0001020325,0.002739225,0.0002201154,0.00007236635],"category_scores_gemma":[0.00009481062,0.0002775498,0.0002591925,0.002317333,0.00009713452,0.0004289467,0.003029123,0.000561047,0.000001247381],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002387369,"about_ca_system_score_gemma":0.0001181113,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003593498,"about_ca_topic_score_gemma":0.001260978,"domain_scores_codex":[0.997971,0.0002969541,0.0002525231,0.0009751564,0.0001720466,0.0003323187],"domain_scores_gemma":[0.9982514,0.000188857,0.0003527313,0.001020758,0.0001137924,0.00007241454],"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":[0.0002540151,0.001968419,0.2310656,0.0003840496,0.001975346,0.00337037,0.02419242,0.6210954,0.001888441,0.1065084,0.0002731641,0.007024412],"study_design_scores_gemma":[0.000439237,0.0000650284,0.04028452,0.00005133721,0.000295703,0.00000137498,0.001208149,0.9477378,0.001250744,0.008028026,0.00005396889,0.0005841626],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8463166,0.0002488655,0.1524063,0.00005314178,0.000103059,0.0002195129,0.0001736939,0.0003444148,0.000134439],"genre_scores_gemma":[0.9883949,0.00002416651,0.01118008,0.0000352015,0.00003013074,0.000002612883,0.0001061803,0.00001322334,0.0002135375],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.3266424,"threshold_uncertainty_score":0.9999677,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0646907881846757,"score_gpt":0.2729201547505337,"score_spread":0.2082293665658579,"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."}}