{"id":"W3158909043","doi":"10.48550/arxiv.1908.06077","title":"NUQSGD: Provably Communication-efficient Data-parallel SGD via Nonuniform Quantization","year":2019,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Advanced Neural Network Applications","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; Stochastic gradient descent; Quantization (signal processing); Heuristic; Deep neural networks; Gradient descent; Communication complexity; Algorithm; Theoretical computer science; Artificial neural network; Artificial intelligence","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":["metaepi_narrow","open_science"],"consensus_categories":["open_science"],"category_scores_codex":[0.0003164695,0.0003971702,0.0003778735,0.0002426235,0.0003625671,0.0001699716,0.008086503,0.0002769778,0.00001843235],"category_scores_gemma":[0.000029822,0.0004662749,0.0001178724,0.001086149,0.0001564162,0.0007709578,0.008837634,0.0007649466,0.0005314876],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002692536,"about_ca_system_score_gemma":0.0002487523,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001137022,"about_ca_topic_score_gemma":0.00007616921,"domain_scores_codex":[0.9971545,0.0001565618,0.000373882,0.001694187,0.0001795579,0.0004413009],"domain_scores_gemma":[0.9907593,0.0002076032,0.0006117245,0.007959929,0.0003001963,0.0001612545],"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":[0.00001017536,0.0001192571,0.0002279645,0.00003344257,0.00003250287,0.000008213121,0.00006710517,0.7874579,0.00002473254,0.2106631,0.0004433418,0.0009122257],"study_design_scores_gemma":[0.0003340509,0.00002183402,0.0004470931,0.00006103071,0.00004452906,0.000005154023,0.00002332943,0.9719695,0.00003211827,0.02307583,0.003499269,0.0004861952],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01253266,0.0001352128,0.9819862,0.0003984077,0.0003511412,0.001202749,0.00005258185,0.0004812815,0.002859743],"genre_scores_gemma":[0.9527342,0.0004269707,0.04464274,0.0001623139,0.00005303129,0.000007094396,0.0006815429,0.00003528884,0.001256817],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.9402015,"threshold_uncertainty_score":0.9997789,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1012755886281797,"score_gpt":0.2261991095555468,"score_spread":0.1249235209273671,"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."}}