{"id":"W4404781872","doi":"10.18653/v1/2024.findings-emnlp.422","title":"Intermediate Layer Distillation with the Reused Teacher Classifier: A Study on the Importance of the Classifier of Attention-based Models","year":2024,"lang":"en","type":"article","venue":"","topic":"Neural Networks and Applications","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University","funders":"","keywords":"Classifier (UML); Distillation; Computer science; Artificial intelligence; Machine learning; Chemistry; Chromatography","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":[0.001404848,0.001206914,0.0008156431,0.0005732447,0.0004378504,0.0009671762,0.001848805,0.0008201677,0.002608466],"category_scores_gemma":[0.008536497,0.0004372981,0.0005680764,0.0007056574,0.000709668,0.003333312,0.001994425,0.002849764,0.0009983102],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008070713,"about_ca_system_score_gemma":0.001427002,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.007356272,"about_ca_topic_score_gemma":0.009752259,"domain_scores_codex":[0.9992616,0.0002432515,0.00004092873,0.0001950363,0.0001525141,0.0001066955],"domain_scores_gemma":[0.9963472,0.002235322,0.0001550461,0.0006396571,0.0004553071,0.000167537],"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":[0.000698536,0.0002932621,0.00327418,0.0002961742,0.0001066317,0.0001812066,0.0003059983,0.1408644,0.03036648,0.008493177,0.004206771,0.8109133],"study_design_scores_gemma":[0.00002352051,0.0001580198,0.0005316265,0.00001817681,0.00003215493,0.00007477948,0.00003803075,0.976851,0.01868106,0.001871334,0.001705966,0.00001439657],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1438077,0.001854807,0.8434969,0.0005352097,0.0001457186,0.00009818094,0.0001217442,0.005171249,0.004768482],"genre_scores_gemma":[0.7690546,0.0007694146,0.2208604,0.0003003138,0.0001166349,0.0001087945,0.0005588428,0.0005685443,0.007662558],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.007356272,"threshold_uncertainty_score":0.01462692,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04510357437240407,"score_gpt":0.27000634110827,"score_spread":0.2249027667358659,"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."}}