{"id":"W4389518966","doi":"10.18653/v1/2023.findings-emnlp.285","title":"Measuring the Knowledge Acquisition-Utilization Gap in Pretrained Language Models","year":2023,"lang":"en","type":"article","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"ca_institutions":"Polytechnique Montréal; Mila - Quebec Artificial Intelligence Institute; Canadian Institute for Advanced Research; McGill University","funders":"","keywords":"Computer science; USable; Domain knowledge; Robustness (evolution); Parametric statistics; Task (project management); Measure (data warehouse); Knowledge acquisition; Artificial intelligence; Knowledge extraction; Machine learning; Data mining; Engineering; Mathematics","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":[0.0101787,0.001247343,0.001051641,0.001597261,0.0005842894,0.002535576,0.001371262,0.001679188,0.001631437],"category_scores_gemma":[0.06879868,0.0006784963,0.000848899,0.001120596,0.001082913,0.007770457,0.003730233,0.003071562,0.0006950214],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009559178,"about_ca_system_score_gemma":0.001487606,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003483501,"about_ca_topic_score_gemma":0.003883641,"domain_scores_codex":[0.9954112,0.001831995,0.0004657998,0.001212696,0.0007224425,0.0003559887],"domain_scores_gemma":[0.9324485,0.05188706,0.003412347,0.00886152,0.002054239,0.001336256],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.003393441,0.002038603,0.2293886,0.001170728,0.00162692,0.0005319541,0.004237782,0.2553307,0.03866426,0.005280424,0.00348992,0.4548468],"study_design_scores_gemma":[0.0001242003,0.002916195,0.1225603,0.0002048979,0.0005570627,0.0006288245,0.00183275,0.7981226,0.04978834,0.01950756,0.003517686,0.0002396201],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9420607,0.0009091182,0.05140591,0.000353828,0.00002888427,0.00009250313,0.0007420477,0.001285728,0.003121368],"genre_scores_gemma":[0.9813844,0.0001784067,0.01634769,0.00008072292,0.00001401285,0.00008185638,0.001398428,0.00009851388,0.0004159338],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.0101787,"threshold_uncertainty_score":0.05383074,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1207953153189931,"score_gpt":0.3006593014070115,"score_spread":0.1798639860880183,"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."}}