{"id":"W4392698767","doi":"10.1016/j.neucom.2024.127468","title":"Sparsity in transformers: A systematic literature review","year":2024,"lang":"en","type":"article","venue":"Neurocomputing","topic":"Ferroelectric and Negative Capacitance Devices","field":"Engineering","cited_by":40,"is_retracted":false,"has_abstract":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Computer science; Transformer; Scalability; Machine learning; Artificial neural network; Computational complexity theory; Artificial intelligence; Computer engineering; Algorithm; 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":[],"consensus_categories":[],"category_scores_codex":[0.0001851674,0.0001261514,0.0002498275,0.0001119838,0.00002040617,0.00005947239,0.0000842667,0.00003550955,0.00000460241],"category_scores_gemma":[0.00002585275,0.0001080419,0.00006523191,0.0008799443,0.000006632259,0.000146896,0.000005419961,0.0003087341,0.00003050417],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003949797,"about_ca_system_score_gemma":0.000006447611,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":3.774292e-7,"about_ca_topic_score_gemma":0.000001698143,"domain_scores_codex":[0.9992678,0.00004096498,0.0002393445,0.0001562289,0.00009869609,0.0001970055],"domain_scores_gemma":[0.9997499,0.0001258497,0.00001075653,0.00007014776,0.00001372765,0.00002959531],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"systematic_review","study_design_gemma":"systematic_review","study_design_scores_codex":[0.000001635183,0.00001049792,0.00006152064,0.9808384,0.00003954294,0.0004744306,0.001743966,0.001050846,0.001262724,0.0005184639,0.002199483,0.01179849],"study_design_scores_gemma":[0.0001263075,0.00003082721,0.0002497596,0.5780618,0.00005136572,0.0001739439,0.00004412709,0.4183315,0.0005875774,0.00009427475,0.001928447,0.0003200523],"study_design_candidate":"systematic_review","study_design_consensus":"systematic_review","genre_codex":"review","genre_gemma":"empirical","genre_scores_codex":[0.05717943,0.9246278,0.007099878,0.0002266171,0.0006806017,0.0009790458,0.000003969333,0.0009621141,0.008240527],"genre_scores_gemma":[0.9799541,0.01950471,0.0001164799,0.0002877465,0.00006580623,0.00001754456,0.000002237454,0.00002236084,0.00002901583],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9227747,"threshold_uncertainty_score":0.440582,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.007780102721587496,"score_gpt":0.2181571800578452,"score_spread":0.2103770773362577,"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."}}