{"id":"W4415318169","doi":"10.48550/arxiv.2510.07203","title":"Sunflower: A New Approach To Expanding Coverage of African Languages in Large Language Models","year":2025,"lang":"en","type":"preprint","venue":"ArXiv.org","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"International Development Research Centre","keywords":"Swahili; Comprehension; Languages of Africa; Bantu languages; State (computer science); Language model","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":[0.004437911,0.001056613,0.000737593,0.001524336,0.0008576555,0.002163087,0.002731424,0.00120771,0.01014212],"category_scores_gemma":[0.0144381,0.001048254,0.002708338,0.00113005,0.0008176297,0.007438005,0.003185056,0.002609727,0.001652523],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001448369,"about_ca_system_score_gemma":0.002098641,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01174788,"about_ca_topic_score_gemma":0.028781,"domain_scores_codex":[0.99887,0.0004842527,0.0000912255,0.0002546943,0.0002268999,0.00007289754],"domain_scores_gemma":[0.9950912,0.003514098,0.0002105407,0.0007310746,0.000266658,0.0001863336],"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":[0.0008762772,0.0004291982,0.01022919,0.000840482,0.0004881866,0.0005984418,0.00314249,0.297361,0.006054507,0.2557402,0.0773147,0.3469253],"study_design_scores_gemma":[0.00008988516,0.00005322443,0.0002784294,0.00006845692,0.00004635102,0.00008078005,0.0001253137,0.8410619,0.001271021,0.1103343,0.04656412,0.0000262341],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01346974,0.0003175994,0.9552786,0.001017423,0.00009973276,0.0002801591,0.004227373,0.02092943,0.004380036],"genre_scores_gemma":[0.1454439,0.0004204464,0.8388392,0.0004706661,0.00009147342,0.0008146144,0.006652076,0.003086932,0.004180676],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01174788,"threshold_uncertainty_score":0.03392875,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03171535518445631,"score_gpt":0.3121366835627905,"score_spread":0.2804213283783342,"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."}}