{"id":"W6920575173","doi":"10.60692/r8a9d-3vs67","title":"Exploring the Limits of Simple Learners in Knowledge Distillation for Document Classification with DocBERT","year":2020,"lang":"en","type":"article","venue":"Greater South Information System","topic":"Handwritten Text Recognition Techniques","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University; University of Waterloo","funders":"","keywords":"Distillation; Baseline (sea); Simple (philosophy); Document classification; Knowledge extraction; Knowledge acquisition; Natural language; Question answering","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":[],"consensus_categories":[],"category_scores_codex":[0.0003412987,0.0001061404,0.0001478413,0.0001456098,0.00007276577,0.0001392749,0.0002931433,0.00003107031,0.000001149954],"category_scores_gemma":[0.00003146343,0.00007113582,0.00003923458,0.0004521863,0.00002048564,0.001620055,0.00004218716,0.0000602848,0.00003741657],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00007968928,"about_ca_system_score_gemma":0.00003629913,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000004570389,"about_ca_topic_score_gemma":9.132928e-7,"domain_scores_codex":[0.9989848,0.00006202401,0.0004879454,0.0001403354,0.0001874548,0.000137399],"domain_scores_gemma":[0.9991997,0.00002958397,0.0002822134,0.0002481547,0.0001918158,0.00004855916],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0002898822,0.00001313175,0.06183945,0.002034354,0.00007866526,0.000001169262,0.7228541,0.001122471,0.00008589805,0.05827888,0.0002657164,0.1531362],"study_design_scores_gemma":[0.005544289,0.001235309,0.2415322,0.001181245,0.00007064544,0.00003301552,0.07996932,0.6027328,0.05888396,0.000212402,0.007351738,0.001253039],"study_design_candidate":"qualitative","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.203045,0.000001400831,0.7944009,0.0004137422,0.00005571111,0.0007442474,0.000009506629,0.0002044848,0.001125056],"genre_scores_gemma":[0.9949717,3.114617e-7,0.004510425,0.00006198035,0.00002719585,0.0004040009,0.00001274424,0.000005453974,0.000006202593],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.7919267,"threshold_uncertainty_score":0.2900834,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1506139091073968,"score_gpt":0.2681573197427652,"score_spread":0.1175434106353684,"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."}}