{"id":"W4387424400","doi":"10.1016/j.asoc.2023.110901","title":"BERT models for Brazilian Portuguese: Pretraining, evaluation and tokenization analysis","year":2023,"lang":"en","type":"article","venue":"Applied Soft Computing","topic":"Topic Modeling","field":"Computer Science","cited_by":26,"is_retracted":false,"has_abstract":false,"ca_institutions":"University of Waterloo","funders":"Conselho Nacional de Desenvolvimento Científico e Tecnológico","keywords":"Computer science; Artificial intelligence; Natural language processing; Language model; Transformer; Lexical analysis; Sentence; Encoder; Transfer of learning; Portuguese; Textual entailment; Machine translation; Logical consequence; Linguistics","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":[0.001538786,0.0001262753,0.0001990061,0.0002953634,0.0002313159,0.0001325676,0.00031731,0.00007032215,0.000002005439],"category_scores_gemma":[0.00006976912,0.0001394948,0.00005971087,0.001267204,0.0000173069,0.0002267688,0.0002271348,0.00007508116,0.00000437029],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003750795,"about_ca_system_score_gemma":0.00005605245,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001141901,"about_ca_topic_score_gemma":0.000005585742,"domain_scores_codex":[0.9984384,0.00003110533,0.0003067375,0.0005542756,0.0003744079,0.0002950739],"domain_scores_gemma":[0.9990544,0.0002269093,0.0001310932,0.0003677606,0.0001498722,0.00006996588],"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":[0.000001609789,0.00000570728,0.000312269,0.00001417436,0.00006573436,3.36271e-7,0.002015754,0.5726486,0.0001599497,0.02987246,0.00008329134,0.3948201],"study_design_scores_gemma":[0.0002887514,0.000009421178,0.002302964,0.000007362746,0.00008075607,6.859194e-7,0.00008761259,0.9756243,0.00005068189,0.02136063,0.00004040878,0.0001464719],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.07141633,0.00006352188,0.926266,0.000167493,0.0001136309,0.0004371141,0.00000122198,0.0004178461,0.001116848],"genre_scores_gemma":[0.862605,0.00000336601,0.1370367,0.0001624598,0.00007221313,0.00002975523,0.00004892206,0.00001139342,0.0000301498],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.7911887,"threshold_uncertainty_score":0.5688434,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04187459156628819,"score_gpt":0.3029099270173801,"score_spread":0.2610353354510919,"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."}}