{"id":"W4411119481","doi":"10.18653/v1/2025.naacl-long.338","title":"Pula: Training Large Language Models for Setswana","year":2025,"lang":"en","type":"article","venue":"","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"Impact","funders":"Clemson University; National Science Foundation","keywords":"Computer science; Training (meteorology); Natural language processing","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00328663,0.003526597,0.001633585,0.001769234,0.00180501,0.002948747,0.004550315,0.002800811,0.01700228],"category_scores_gemma":[0.008283102,0.002353279,0.003625248,0.001454129,0.0007320791,0.005356696,0.004004494,0.007017066,0.01624852],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00155699,"about_ca_system_score_gemma":0.002409299,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01343235,"about_ca_topic_score_gemma":0.03050468,"domain_scores_codex":[0.9983671,0.0006598579,0.0001039157,0.0005719113,0.0001900773,0.0001071815],"domain_scores_gemma":[0.9968507,0.002108023,0.00005612875,0.0005489463,0.0003100444,0.0001261813],"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":[0.001572132,0.0009048478,0.004164213,0.0008163388,0.001985316,0.0005861406,0.000412584,0.1435428,0.005295466,0.00692681,0.2901396,0.5436538],"study_design_scores_gemma":[0.0003522489,0.0001792119,0.0006972771,0.00008738722,0.0002056643,0.0001215714,0.0001786275,0.9528336,0.004408299,0.01897387,0.02190273,0.00005947583],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.06214119,0.00815824,0.6483354,0.003566424,0.003359467,0.0008961617,0.02946663,0.2284121,0.01566442],"genre_scores_gemma":[0.2882746,0.002259157,0.5303735,0.002625822,0.0005703479,0.002736131,0.1401965,0.01083504,0.02212892],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01700228,"threshold_uncertainty_score":0.05687833,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02204869807519197,"score_gpt":0.3135279064304263,"score_spread":0.2914792083552343,"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."}}