{"id":"W3099658661","doi":"10.18653/v1/2020.emnlp-main.744","title":"An information theoretic view on selecting linguistic probes","year":2020,"lang":"en","type":"article","venue":"","topic":"Natural Language Processing Techniques","field":"Computer Science","cited_by":14,"is_retracted":false,"has_abstract":true,"ca_institutions":"Vector Institute; University of Toronto","funders":"Canadian Institute for Advanced Research","keywords":"Classifier (UML); Artificial intelligence; Modulo; Computer science; Information gain; Natural language processing; Selection (genetic algorithm); Machine learning; Mathematics; Discrete mathematics","routes":{"ca_aff":true,"ca_fund":true,"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":[0.01856108,0.001965633,0.002191571,0.005282156,0.001777894,0.007207692,0.004801463,0.005640959,0.008002097],"category_scores_gemma":[0.09212174,0.001387965,0.001846401,0.002799131,0.01515659,0.02018467,0.005521892,0.005923197,0.001160939],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.003614424,"about_ca_system_score_gemma":0.001785677,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001204921,"about_ca_topic_score_gemma":0.0007219111,"domain_scores_codex":[0.9843212,0.00887585,0.0006220834,0.002334251,0.003255389,0.0005911494],"domain_scores_gemma":[0.9120121,0.07341896,0.00402215,0.006200398,0.003301215,0.001045284],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0002261905,0.00009168437,0.001000179,0.0002540358,0.00009114821,0.00006989529,0.0004567853,0.01954574,0.001725829,0.9371319,0.001946081,0.03746047],"study_design_scores_gemma":[0.00005118488,0.00007828588,0.0002048497,0.00003557032,0.00002383126,0.00004521895,0.00006422326,0.0442394,0.0008842074,0.9530787,0.001262718,0.0000318325],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.009387254,0.0007036564,0.9730442,0.005503797,0.00007179267,0.0001254176,0.0001843132,0.0002457456,0.01073392],"genre_scores_gemma":[0.63033,0.001428358,0.3546485,0.004437899,0.0007863904,0.001681079,0.0005791979,0.0003204754,0.005788024],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01856108,"threshold_uncertainty_score":0.09816152,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01090330771742997,"score_gpt":0.2687331560370854,"score_spread":0.2578298483196554,"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."}}