{"id":"W3100985894","doi":"10.1109/micro50266.2020.00071","title":"GOBO: Quantizing Attention-Based NLP Models for Low Latency and Energy Efficient Inference","year":2020,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Topic Modeling","field":"Computer Science","cited_by":163,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto","funders":"","keywords":"Computer science; Quantization (signal processing); Inference; Computation; Latency (audio); Language model; Parallel computing; Computer engineering; Computer hardware; Artificial intelligence; Algorithm","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":[0.000787173,0.0007737059,0.0007022862,0.0006487797,0.0003819125,0.0009373178,0.00180123,0.0009173817,0.004709058],"category_scores_gemma":[0.003968044,0.0004442622,0.0006022362,0.0006638438,0.0006738571,0.002464982,0.001627434,0.001618605,0.0010865],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00106022,"about_ca_system_score_gemma":0.001049051,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01057842,"about_ca_topic_score_gemma":0.0199464,"domain_scores_codex":[0.9996779,0.00007216392,0.00001687112,0.00008294031,0.0001112649,0.00003881525],"domain_scores_gemma":[0.9993151,0.0003938483,0.00005214978,0.0001092018,0.00009777367,0.00003182879],"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":[0.00045985,0.0001595845,0.001246104,0.000350049,0.0001307263,0.0001542345,0.0002763746,0.55861,0.02482651,0.03394753,0.01230179,0.3675372],"study_design_scores_gemma":[0.000009884186,0.00001221917,0.00006574653,0.000006651376,0.000005967386,0.000009789605,0.000009153959,0.987432,0.001909631,0.009809873,0.0007243343,0.000004631388],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.0175721,0.0004563507,0.9743931,0.0003819868,0.00005954436,0.00007112404,0.0003334617,0.004760182,0.001972129],"genre_scores_gemma":[0.5540584,0.0005016627,0.4383315,0.0005703932,0.00009381857,0.0002764828,0.001208903,0.0007525533,0.004206401],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01057842,"threshold_uncertainty_score":0.0210337,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.11445244292019,"score_gpt":0.2010061101867137,"score_spread":0.08655366726652367,"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."}}