{"id":"W4402671693","doi":"10.18653/v1/2024.acl-long.680","title":"To Distill or Not to Distill? On the Robustness of Robust Knowledge Distillation","year":2024,"lang":"en","type":"article","venue":"","topic":"Bayesian Modeling and Causal Inference","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Alliance de recherche numérique du Canada; Social Sciences and Humanities Research Council of Canada; Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs","keywords":"Robustness (evolution); Computer science; Distillation; Artificial intelligence; Machine learning; Chromatography; Biology; Chemistry","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.008170151,0.002617563,0.001542582,0.001496024,0.001359401,0.003136519,0.003087208,0.002534518,0.005457548],"category_scores_gemma":[0.04190649,0.0008614812,0.001444321,0.001287866,0.002113681,0.00758331,0.005401744,0.005923784,0.003410169],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001136043,"about_ca_system_score_gemma":0.002278639,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01562357,"about_ca_topic_score_gemma":0.01824176,"domain_scores_codex":[0.9938735,0.003181679,0.0002539615,0.001691502,0.0006587738,0.0003405453],"domain_scores_gemma":[0.9811621,0.01345009,0.0004063066,0.003627436,0.001008873,0.0003451141],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.001428859,0.0002999117,0.005103353,0.0005705464,0.0005815201,0.0003394588,0.0007819774,0.5016863,0.008139685,0.03528646,0.02402116,0.4217608],"study_design_scores_gemma":[0.00007077914,0.0001104432,0.0005326103,0.0001075268,0.00006435467,0.0001069962,0.0001488269,0.9426539,0.00706149,0.04422426,0.004857839,0.00006098774],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1558072,0.006123739,0.7862288,0.007138507,0.000726709,0.0001966711,0.003896935,0.02367082,0.01621063],"genre_scores_gemma":[0.7719439,0.001043955,0.2071885,0.002052821,0.0002461155,0.0002033122,0.006486543,0.003020482,0.007814303],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01562357,"threshold_uncertainty_score":0.04320842,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.08253438536572152,"score_gpt":0.3196380303735552,"score_spread":0.2371036450078337,"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."}}