{"id":"W4406733441","doi":"10.48550/arxiv.2501.10858","title":"Reliable Text-to-SQL with Adaptive Abstention","year":2025,"lang":"en","type":"preprint","venue":"ArXiv.org","topic":"Distributed and Parallel Computing Systems","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; SQL; Database","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":[0.006895686,0.001193591,0.0009581726,0.0007297088,0.0005448776,0.002316543,0.003770051,0.001703039,0.003247454],"category_scores_gemma":[0.03869656,0.0007094828,0.001000226,0.0006570275,0.001832837,0.005928741,0.004738231,0.002647115,0.001775552],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009043562,"about_ca_system_score_gemma":0.001537154,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002085033,"about_ca_topic_score_gemma":0.002315191,"domain_scores_codex":[0.9927233,0.002945989,0.0004727655,0.001272171,0.002326157,0.0002596678],"domain_scores_gemma":[0.974548,0.01294946,0.00129066,0.008812748,0.001964556,0.0004345355],"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":[0.002513118,0.0007718269,0.01069385,0.000525334,0.0001860521,0.0007436117,0.001715911,0.3975621,0.06236047,0.06375038,0.01932424,0.4398531],"study_design_scores_gemma":[0.0000474883,0.0000959504,0.0002293835,0.000008615201,0.00001339852,0.00007188602,0.00003706937,0.957822,0.0104223,0.02884908,0.002380276,0.00002261045],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.04468338,0.0001761278,0.9159045,0.0005213033,0.00004831994,0.0001658988,0.0003028034,0.03677363,0.001424149],"genre_scores_gemma":[0.6319914,0.0001234044,0.3601567,0.0005211877,0.00007710806,0.0002843647,0.001002996,0.002918851,0.002924141],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.006895686,"threshold_uncertainty_score":0.03646833,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0346877412721206,"score_gpt":0.2570595838781645,"score_spread":0.2223718426060439,"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."}}