{"id":"W4243640523","doi":"10.31219/osf.io/m6gcn","title":"Attention Mechanism, Transformers, BERT, and GPT: Tutorial and Survey","year":2020,"lang":"en","type":"preprint","venue":"","topic":"Explainable Artificial Intelligence (XAI)","field":"Computer Science","cited_by":80,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Waterloo","funders":"","keywords":"Transformer; Encoder; Computer science; Generative grammar; Artificial intelligence; Speech recognition; Electrical engineering; Engineering; Voltage","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"],"consensus_categories":[],"category_scores_codex":[0.0008038589,0.0002788612,0.0003351863,0.00009800485,0.0001430482,0.0005967719,0.0005762758,0.0002582624,0.00002740947],"category_scores_gemma":[0.00008915138,0.0002703807,0.00006880955,0.000191651,0.00006528574,0.0004074898,0.000947942,0.0003937712,0.00003493501],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00004213333,"about_ca_system_score_gemma":0.0001168103,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002711233,"about_ca_topic_score_gemma":0.001135995,"domain_scores_codex":[0.9978786,0.000180518,0.0003625824,0.0009452221,0.0003226049,0.0003104443],"domain_scores_gemma":[0.9990524,0.0001509102,0.00009798581,0.0003738096,0.0001188005,0.0002060902],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0001016842,0.0001139851,0.001562127,0.0004157489,0.0001731515,0.00007308945,0.004264704,0.00004590703,0.008239611,0.9013829,0.002138573,0.08148853],"study_design_scores_gemma":[0.0007185218,0.0006748633,0.01463365,0.0002154804,0.00009966759,0.00004981662,0.0004957974,0.2900787,0.0458368,0.6400071,0.00482519,0.002364428],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.03923064,0.0001329131,0.9534251,0.002453596,0.00231272,0.0005168082,0.00002104734,0.0002409161,0.001666253],"genre_scores_gemma":[0.982805,0.0003909961,0.01567179,0.000311685,0.0002969945,0.0000311716,0.00003745784,0.00002259491,0.0004323625],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9435743,"threshold_uncertainty_score":0.9999748,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.05466022564196835,"score_gpt":0.2753320060763013,"score_spread":0.2206717804343329,"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."}}