{"id":"W3106085430","doi":"","title":"Gradient Estimation with Stochastic Softmax Tricks","year":2020,"lang":"en","type":"article","venue":"Neural Information Processing Systems","topic":"Stochastic Gradient Optimization Techniques","field":"Computer Science","cited_by":12,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto","funders":"","keywords":"Softmax function; Estimator; Gumbel distribution; Computer science; Latent variable; Algorithm; Artificial intelligence; Mathematics; Theoretical computer science; Machine learning; Artificial neural network; Statistics","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":[0.0001428269,0.0001907161,0.0001955848,0.0001910944,0.0002186155,0.0008889035,0.0004869052,0.00006027157,0.000002159281],"category_scores_gemma":[0.000137431,0.0001542118,0.00002907055,0.0009403247,0.00003801444,0.004763487,0.00006806639,0.0001385755,0.00004499954],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00006706872,"about_ca_system_score_gemma":0.00009524771,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000008687055,"about_ca_topic_score_gemma":1.190051e-7,"domain_scores_codex":[0.9984556,0.00003150235,0.0005293632,0.0002158802,0.0005255711,0.0002421084],"domain_scores_gemma":[0.9987842,0.0000388139,0.0004579744,0.0002175485,0.0003383849,0.000163023],"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":[0.00003530895,0.00002563479,0.00003393685,0.000432465,0.00001275063,0.000002775177,0.00875402,0.8223923,0.00002589151,0.02052878,0.001211045,0.1465451],"study_design_scores_gemma":[0.0003414323,0.0002077933,0.00002106082,0.00008465518,0.000007577325,0.00006184882,0.0001413178,0.9984535,0.00009853656,0.0000950175,0.0002940281,0.0001932346],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.0007223745,0.00005918941,0.9957148,0.001019833,0.0001998369,0.0005599303,0.000002669904,0.00122171,0.0004996634],"genre_scores_gemma":[0.9479235,5.071055e-7,0.05112214,0.0007546582,0.00004321095,0.0001099352,0.00002335072,0.00001010635,0.00001253221],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.9472012,"threshold_uncertainty_score":0.8571717,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01854419720311598,"score_gpt":0.2238023871480692,"score_spread":0.2052581899449533,"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."}}