{"id":"W7125609352","doi":"10.1109/icsai68704.2025.11345826","title":"Evaluating Zero-Shot and Few-Shot Learning on the Turkish Massive Multitask Language Understanding Dataset","year":2025,"lang":"","type":"article","venue":"","topic":"Domain Adaptation and Few-Shot Learning","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"ca_institutions":"University of Calgary; Dalhousie University","funders":"","keywords":"Turkish; Task (project management); Multi-task learning; Language acquisition; Task analysis; Feature (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":["metaepi_narrow","sts","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.003409056,0.0005560642,0.0004578969,0.0003944157,0.002320512,0.002119865,0.001055094,0.0002019627,0.0008968753],"category_scores_gemma":[0.00250361,0.0004374544,0.0001295712,0.001197336,0.0003639048,0.0006910517,0.001070281,0.001472442,0.0001812121],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003273525,"about_ca_system_score_gemma":0.0002928494,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001252642,"about_ca_topic_score_gemma":0.00004038272,"domain_scores_codex":[0.9947734,0.001334075,0.0007321889,0.001335142,0.0009452217,0.000879992],"domain_scores_gemma":[0.9944368,0.003905945,0.0004138318,0.0008937813,0.0001188949,0.0002307376],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0003550002,0.0003320246,0.00203917,0.0003284342,0.0007712012,0.0001723579,0.04329922,0.03324984,0.007500196,0.6924245,0.02381041,0.1957176],"study_design_scores_gemma":[0.002093286,0.0004548656,0.001050638,0.000883461,0.0001419235,0.00001439427,0.05387078,0.9309332,0.0006019197,0.001655582,0.007549531,0.0007504441],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.04049025,0.001518427,0.8870906,0.01253577,0.001706338,0.001642805,0.0001526372,0.0003896515,0.05447348],"genre_scores_gemma":[0.9824747,0.0001006721,0.005564646,0.004786056,0.00008741864,0.00002924148,0.0001766911,0.00003326163,0.006747334],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9419844,"threshold_uncertainty_score":0.9998077,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1710522026815548,"score_gpt":0.3854136519028957,"score_spread":0.2143614492213408,"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."}}