{"id":"W4378801057","doi":"10.1145/3583781.3590302","title":"High-Throughput Edge Inference for BERT Models via Neural Architecture Search and Pipeline","year":2023,"lang":"en","type":"article","venue":"","topic":"Advanced Neural Network Applications","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University","funders":"","keywords":"Computer science; Pipeline (software); Throughput; Inference; Enhanced Data Rates for GSM Evolution; Architecture; Artificial neural network; Computer architecture; Parallel computing; Distributed computing; Artificial intelligence; Operating system; Wireless","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":[0.0001194933,0.0001464083,0.0001480757,0.00009056454,0.0001977733,0.00007886707,0.0005919823,0.00004938808,0.000003835838],"category_scores_gemma":[0.00002143073,0.0001225084,0.00003709549,0.0007221575,0.00005561344,0.0003760151,0.0005283365,0.0001605105,0.00002856806],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00001359568,"about_ca_system_score_gemma":0.00002413302,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00003561779,"about_ca_topic_score_gemma":0.00004491309,"domain_scores_codex":[0.998692,0.00002507402,0.0001819747,0.0005245807,0.0001756045,0.0004007359],"domain_scores_gemma":[0.9988055,0.0004302029,0.00002714313,0.0005265184,0.00008988491,0.0001207535],"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":[0.00001188862,0.0000277562,0.00003528778,0.00002526241,0.000007020948,0.00000362409,0.0003635448,0.3968828,0.00211553,0.3059927,0.004167285,0.2903673],"study_design_scores_gemma":[0.0001953351,0.00004089289,0.0001698183,0.000003515938,0.000002027438,0.000007265963,0.000003893894,0.8543317,0.001025103,0.142921,0.001169959,0.0001294204],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01008074,0.00003804012,0.9806775,0.007916061,0.0001041889,0.0004827865,0.0000086885,0.0005389146,0.0001530169],"genre_scores_gemma":[0.752453,0.00004281094,0.2451679,0.0007535393,0.0001357751,0.0001478901,0.00002037825,0.0000161739,0.001262564],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.7423722,"threshold_uncertainty_score":0.4995748,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04471021459728105,"score_gpt":0.3134345893846869,"score_spread":0.2687243747874059,"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."}}