{"id":"W6948229243","doi":"10.48448/g5r6-ha27","title":"RAIL-KD: RAndom Intermediate Layer Mapping for Knowledge Distillation","year":2022,"lang":"en","type":"other","venue":"Underline Science Inc.","topic":"","field":"","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Université de Montréal","funders":"","keywords":"Distillation; Layer (electronics); GLUE; Generalizability theory; Fractionating column","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":[0.001185433,0.001696244,0.00112372,0.0008556267,0.0007964563,0.001753097,0.003308993,0.001559277,0.00977922],"category_scores_gemma":[0.004770629,0.0006619255,0.001246175,0.0009533205,0.001238732,0.004058335,0.004341448,0.002981404,0.004864644],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008697341,"about_ca_system_score_gemma":0.002143563,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004127773,"about_ca_topic_score_gemma":0.007392436,"domain_scores_codex":[0.9989567,0.0002765197,0.0000719247,0.0003578929,0.0002028521,0.0001341277],"domain_scores_gemma":[0.9986165,0.000463919,0.00008212285,0.0005558173,0.0001945798,0.00008707328],"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":[0.0005121281,0.0004038017,0.001253012,0.0005211724,0.0001770993,0.0003349979,0.0003140402,0.2117137,0.01512005,0.03802096,0.02329752,0.7083316],"study_design_scores_gemma":[0.00005701922,0.00007403478,0.0001534515,0.00003533611,0.00002649403,0.00009207085,0.00004972755,0.9426236,0.01233033,0.03791006,0.006612013,0.00003585241],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.0148717,0.0005241886,0.9655673,0.000339785,0.0001325301,0.000112216,0.0005588324,0.01343309,0.004460368],"genre_scores_gemma":[0.406192,0.0003751086,0.5750284,0.0006469865,0.00008446429,0.000348394,0.003301913,0.001462737,0.01255995],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.00977922,"threshold_uncertainty_score":0.03271472,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04995969101419159,"score_gpt":0.3265329612535607,"score_spread":0.2765732702393692,"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."}}