{"id":"W4381746849","doi":"10.1109/percomworkshops56833.2023.10150381","title":"Reducing the Cost of GPU Cold Starts in Serverless Deep Learning Inference Serving","year":2023,"lang":"en","type":"article","venue":"","topic":"IoT and Edge/Fog Computing","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Cloud computing; Provisioning; Latency (audio); Pooling; Speedup; Distributed computing; Total cost of ownership; Exploit; Inference; Computer network; Operating system; Artificial intelligence; Computer security","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001876794,0.001415084,0.0009935535,0.0005458206,0.001159207,0.002066654,0.002930367,0.001263781,0.004016583],"category_scores_gemma":[0.00818815,0.0008063175,0.0006390282,0.0007137457,0.001524696,0.00343171,0.002417485,0.002517951,0.001042562],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002222504,"about_ca_system_score_gemma":0.003329922,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01108094,"about_ca_topic_score_gemma":0.01660654,"domain_scores_codex":[0.9984895,0.0002971356,0.00008367822,0.0003186639,0.0004173176,0.0003936535],"domain_scores_gemma":[0.9962047,0.001451709,0.0002826133,0.0009815462,0.0006415258,0.0004378867],"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":[0.004053166,0.001291455,0.0217974,0.0004761869,0.000340377,0.001225817,0.0008489139,0.6653342,0.06582402,0.01587233,0.02591264,0.1970236],"study_design_scores_gemma":[0.00005975861,0.0001524052,0.0006911314,0.000009882478,0.00002375676,0.00005084248,0.00006437262,0.984391,0.01079336,0.002597013,0.00115035,0.00001604242],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.6290935,0.002269737,0.3346947,0.001483693,0.0005144271,0.0002733071,0.0003675067,0.02224665,0.00905643],"genre_scores_gemma":[0.9498766,0.0001490911,0.04715904,0.0003949226,0.00003563854,0.00005389955,0.0003554132,0.0004161654,0.001559118],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.01108094,"threshold_uncertainty_score":0.02203286,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03212428190171373,"score_gpt":0.2712713175877633,"score_spread":0.2391470356860496,"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."}}