{"id":"W3123224975","doi":"","title":"When does preconditioning help or hurt generalization","year":2021,"lang":"en","type":"article","venue":"Institutional Repositories DataBase (IRDB)","topic":"Stochastic Gradient Optimization Techniques","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto","funders":"","keywords":"Generalization; Reproducing kernel Hilbert space; Computer science; Matching (statistics); Kernel (algebra); Population; Variance (accounting); Algorithm; Applied mathematics; Hilbert space; Mathematical optimization; Mathematics; Statistics; Discrete mathematics","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":[0.0001922318,0.0002077783,0.0001851661,0.0001426346,0.000686076,0.0005368853,0.0005379861,0.00008739935,0.0001680281],"category_scores_gemma":[0.0007589659,0.0001709611,0.00006538103,0.000583662,0.00014956,0.002361956,0.0004481094,0.0001380325,0.000026006],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000214235,"about_ca_system_score_gemma":0.0008228167,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00007523216,"about_ca_topic_score_gemma":0.00005447122,"domain_scores_codex":[0.9979143,0.0000865188,0.0004504022,0.0006549064,0.0006213661,0.0002725048],"domain_scores_gemma":[0.9980841,0.0001346622,0.000171921,0.0008238652,0.0006487568,0.0001366687],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.00001951962,0.0001380618,0.001075716,0.00005053108,0.00005377904,0.0002708975,0.0003508804,0.003794364,0.004212527,0.9798188,0.009311201,0.0009036722],"study_design_scores_gemma":[0.002101712,0.0002846933,0.003591753,0.001295718,0.0001377727,0.004058101,0.0002510137,0.1043264,0.6281241,0.05475965,0.1987787,0.002290414],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.002194063,0.0001540769,0.9901135,0.0007325812,0.003346167,0.0001979443,0.0001490659,0.0005962402,0.002516366],"genre_scores_gemma":[0.1084869,0.00008695352,0.8838874,0.0007730479,0.001167122,0.000208226,0.002111804,0.0000269415,0.003251664],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.9250592,"threshold_uncertainty_score":0.6971592,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02050156646347252,"score_gpt":0.2614843825587116,"score_spread":0.2409828160952391,"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."}}