{"id":"W3176194860","doi":"","title":"SGD in the Large: Average-case Analysis, Asymptotics, and Stepsize Criticality","year":2021,"lang":"en","type":"article","venue":"Conference on Learning Theory","topic":"Statistical Mechanics and Entropy","field":"Physics and Astronomy","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University","funders":"","keywords":"Convergence (economics); Stochastic gradient descent; Applied mathematics; Mathematics; Criticality; Rate of convergence; Limit (mathematics); Statistical physics; Mathematical optimization; Computer science; Mathematical analysis; Physics; Artificial neural network","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":["insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.0007356182,0.0001095171,0.0001919364,0.00003817493,0.0001544124,0.0001307087,0.00008916233,0.00003117898,0.002485776],"category_scores_gemma":[0.0002922751,0.00008311019,0.00006303906,0.0002035265,0.00005058536,0.00003017141,0.00005742674,0.000406191,0.00003090746],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000007499367,"about_ca_system_score_gemma":0.00003700671,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001788713,"about_ca_topic_score_gemma":0.000004377133,"domain_scores_codex":[0.9987053,0.0005345894,0.0001527807,0.00025226,0.000117779,0.0002372424],"domain_scores_gemma":[0.9987094,0.0009072955,0.00003378918,0.0002141164,0.00006965501,0.00006573159],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.00001428022,0.00009517917,0.006431082,0.000005372154,0.00007167155,0.0001642276,0.0004074274,0.00006163056,0.00001459437,0.9845695,0.00002303736,0.008141951],"study_design_scores_gemma":[0.0009846438,0.0001004978,0.02015467,0.00005097824,0.0003580396,0.00002381733,0.01237568,0.1057127,0.0001042446,0.8574991,0.002263579,0.0003720711],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.3181401,0.00006564686,0.6567939,0.001027475,0.0001020107,0.0001170554,0.00006473385,0.00002535072,0.02366372],"genre_scores_gemma":[0.9988331,0.00000717738,0.0002559923,0.0002301521,0.00005208779,0.000008111019,0.00002977123,0.000006878025,0.0005767459],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.680693,"threshold_uncertainty_score":0.9984261,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01944880602390112,"score_gpt":0.2931511756311167,"score_spread":0.2737023696072156,"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."}}