{"id":"W6929288060","doi":"10.48448/xk4m-dn57","title":"Recursive Neural Networks with Bottlenecks Diagnose (Non-)Compositionality","year":2022,"lang":"en","type":"other","venue":"Underline Science Inc.","topic":"Prenatal Screening and Diagnostics","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University","funders":"","keywords":"Principle of compositionality; Bottleneck; Information bottleneck method; Metric (unit); Artificial neural network; Ranking (information retrieval); Natural language","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","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0003562943,0.0003133932,0.0004091427,0.0003292334,0.0002977278,0.00006348099,0.0004621925,0.0001543014,0.002741623],"category_scores_gemma":[0.0002389329,0.0002474261,0.00007068535,0.0008099785,0.0009929285,0.00008705338,0.0003116994,0.0007042903,0.00003100624],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002334788,"about_ca_system_score_gemma":0.0005274088,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0006141696,"about_ca_topic_score_gemma":0.0001781,"domain_scores_codex":[0.9974834,0.00004488603,0.0002251361,0.0007074658,0.001005976,0.0005331815],"domain_scores_gemma":[0.9984164,0.0002450667,0.0002138331,0.0006285364,0.0001174565,0.0003787185],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.0004552637,0.001143212,0.03416616,0.0001706986,0.0002446798,0.001715475,0.0001673097,0.01057887,0.00003662065,0.003589058,0.919612,0.02812064],"study_design_scores_gemma":[0.009798472,0.007871273,0.05400508,0.003099996,0.001485397,0.002038785,0.0008445616,0.1923852,0.0001924956,0.001024413,0.7235076,0.003746737],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"other","genre_gemma":"empirical","genre_scores_codex":[0.003719315,0.003639619,0.01860564,0.004780983,0.002063266,0.002314424,0.0005551707,0.001013287,0.9633083],"genre_scores_gemma":[0.6605096,0.0008143026,0.05509452,0.0125094,0.00581478,0.0004220087,0.005543663,0.00128245,0.2580093],"genre_candidate":"other","genre_consensus":null,"teacher_disagreement_score":0.705299,"threshold_uncertainty_score":0.9999978,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0128866749576053,"score_gpt":0.2795446215318533,"score_spread":0.266657946574248,"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."}}