{"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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001009681,0.0001048932,0.0001547593,0.0001459855,0.0001908986,0.0001340121,0.0008824476,0.00004223171,0.000004678893],"category_scores_gemma":[0.0001405716,0.00008161894,0.00003790624,0.001392516,0.00002768023,0.0003098333,0.0008262811,0.0002586714,0.00004261137],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003766779,"about_ca_system_score_gemma":0.00005885615,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0005383358,"about_ca_topic_score_gemma":0.0001065506,"domain_scores_codex":[0.9987156,0.0001208388,0.0002852035,0.0002706697,0.0002348786,0.0003728443],"domain_scores_gemma":[0.9989348,0.0004824854,0.00009948505,0.0003630221,0.00007769198,0.00004252345],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00001381169,0.0001149442,0.2908657,0.0003269696,0.00004597828,0.0001185992,0.04899631,0.2655525,0.01166033,0.02533932,0.003764728,0.3532008],"study_design_scores_gemma":[0.0001685447,0.00002719106,0.05507012,0.0001403551,0.000001724663,0.000002418542,0.0004014948,0.9385945,0.003026172,0.0004203053,0.001993018,0.0001541868],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9129155,0.00004961947,0.08065476,0.0006804047,0.001978712,0.0001872786,4.552999e-8,0.0002946015,0.00323909],"genre_scores_gemma":[0.9972392,0.00001086653,0.002115479,0.00009363312,0.0001581117,0.000006295766,9.185689e-7,0.000007772033,0.0003677715],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.6730419,"threshold_uncertainty_score":0.3328324,"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."}}