{"id":"W3161757350","doi":"10.18653/v1/2021.acl-long.86","title":"MATE-KD: Masked Adversarial TExt, a Companion to Knowledge Distillation","year":2021,"lang":"en","type":"preprint","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"ca_institutions":"Carleton University; University of British Columbia","funders":"","keywords":"Computer science; Distillation; Benchmark (surveying); Divergence (linguistics); Adversarial system; Generator (circuit theory); Artificial intelligence; Key (lock); Set (abstract data type); Test set; Machine learning; Natural language processing; Ranking (information retrieval); Power (physics); Programming language; Linguistics","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":[0.003964617,0.001852997,0.001916419,0.001310252,0.001318301,0.003183895,0.005076888,0.004158962,0.02770285],"category_scores_gemma":[0.01912356,0.001271956,0.001307734,0.001052237,0.001591375,0.007621711,0.008963194,0.005344011,0.01629261],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009416541,"about_ca_system_score_gemma":0.001615278,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001863166,"about_ca_topic_score_gemma":0.003609988,"domain_scores_codex":[0.9969718,0.001135398,0.0001498926,0.0006867738,0.0008528157,0.0002032963],"domain_scores_gemma":[0.9934605,0.003440663,0.0001565884,0.002183688,0.0005236773,0.0002348303],"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":[0.001108352,0.0004699448,0.0008106215,0.0007437288,0.0003403391,0.0005047422,0.0003139851,0.1804553,0.005758276,0.1529939,0.2923247,0.3641762],"study_design_scores_gemma":[0.0001221723,0.0000759006,0.0001320078,0.00005766833,0.00002327339,0.0001661896,0.00003532878,0.8036391,0.004825258,0.1574564,0.03341847,0.00004831645],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.004352532,0.00126069,0.9480129,0.002078805,0.001423434,0.0001819453,0.002891528,0.02856188,0.01123629],"genre_scores_gemma":[0.2160459,0.001258429,0.720843,0.002941327,0.00121143,0.0007817801,0.01451166,0.006650307,0.0357563],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.02770285,"threshold_uncertainty_score":0.09267533,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03359519427592566,"score_gpt":0.2843402056826809,"score_spread":0.2507450114067553,"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."}}