{"id":"W3037585139","doi":"10.18653/v1/2020.repl4nlp-1.10","title":"Exploring the Limits of Simple Learners in Knowledge Distillation for Document Classification with DocBERT","year":2020,"lang":"en","type":"article","venue":"","topic":"Topic Modeling","field":"Computer Science","cited_by":22,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University; University of Waterloo","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Distillation; Computer science; Task (project management); Baseline (sea); Artificial intelligence; FLOPS; Machine learning; Natural language processing; Parallel computing; Engineering","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":[0.002432203,0.001800867,0.0009202693,0.000834341,0.0006292588,0.001793878,0.001885963,0.001703745,0.003320643],"category_scores_gemma":[0.008321455,0.0005437961,0.0004359806,0.0007864666,0.0009823588,0.007139972,0.002233673,0.003599316,0.001694838],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001014751,"about_ca_system_score_gemma":0.00148875,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.007445176,"about_ca_topic_score_gemma":0.01358227,"domain_scores_codex":[0.999329,0.0002225804,0.00003585253,0.0001567132,0.0001407629,0.0001151106],"domain_scores_gemma":[0.9969767,0.002033696,0.0001007086,0.0005112649,0.000250748,0.0001269145],"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":[0.0009588664,0.0005436139,0.003568232,0.0005112276,0.0002039787,0.0002441014,0.000364868,0.4015341,0.01392283,0.01448468,0.01043857,0.5532249],"study_design_scores_gemma":[0.00003128119,0.0001317961,0.0002514607,0.0000302366,0.00002381775,0.00004498739,0.00006563523,0.9811778,0.007352354,0.009039477,0.001831828,0.00001923515],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.4450441,0.008473228,0.4999358,0.003380412,0.0006340054,0.0002292417,0.00131109,0.02103986,0.01995225],"genre_scores_gemma":[0.8836865,0.000958595,0.1065042,0.000631494,0.0001372693,0.0001285233,0.001388873,0.0004793885,0.006085204],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.007445176,"threshold_uncertainty_score":0.01480371,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.2261500682115022,"score_gpt":0.3147457090188217,"score_spread":0.08859564080731949,"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."}}