{"id":"W4287891037","doi":"10.18653/v1/2022.naacl-main.188","title":"How Gender Debiasing Affects Internal Model Representations, and Why It Matters","year":2022,"lang":"en","type":"article","venue":"Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies","topic":"Topic Modeling","field":"Computer Science","cited_by":15,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Azrieli Foundation","keywords":"Debiasing; Computer science; Data science; Linguistics; Cognitive science; Psychology; Philosophy","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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003353578,0.0001326366,0.0002383626,0.0001222695,0.0003878848,0.00008393549,0.001904699,0.00002706774,0.000001238082],"category_scores_gemma":[0.001683063,0.00009433372,0.0001493076,0.0003048393,0.0002748827,0.00008444134,0.001766844,0.0002403705,4.182122e-8],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001120415,"about_ca_system_score_gemma":0.00005011999,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00006302581,"about_ca_topic_score_gemma":0.00002115334,"domain_scores_codex":[0.9986169,0.00002514274,0.0003076145,0.0002640186,0.0006220341,0.0001643209],"domain_scores_gemma":[0.997311,0.0001989442,0.001578164,0.0002754667,0.0006226621,0.00001378336],"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":[0.00004211625,0.0001301687,0.2011039,0.0003134705,0.0004207025,3.709921e-7,0.01242522,0.08721467,0.00602721,0.6837062,0.002199886,0.006416124],"study_design_scores_gemma":[0.0007909648,0.0001801823,0.02874128,0.000146588,0.0001857772,0.000009326396,0.01268576,0.7454143,0.01051327,0.2005334,0.0003346193,0.0004645211],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9354289,0.0000964768,0.04452937,0.01774427,0.0004706373,0.0008515268,0.0001573231,0.0001648522,0.000556668],"genre_scores_gemma":[0.9843157,0.000002842143,0.01514514,0.0002795707,0.00002195917,0.00003137828,0.000002207257,0.00001025384,0.000190962],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.6581997,"threshold_uncertainty_score":0.3846818,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02677859955779483,"score_gpt":0.2586122176168469,"score_spread":0.2318336180590521,"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."}}